Draft version June 21, 2024

Typeset using LATEX twocolumn style in AASTeX631

Adiabatic Mass Loss In Binary Stars. IV. Low and Intermediate Mass Helium Binary Stars
Lifu Zhang ,1, 2 Hongwei Ge ,1, 3, 4 Xuefei Chen ,1, 2, 3, 4 and Zhanwen Han 1, 2, 3, 4

1 Yunnan Observatories, Chinese Academy of Sciences, Kunming 650216, China

2 University of Chinese Academy of Sciences, Beijing 100049, People’s Republic of China
3 Key Laboratory for Structure and Evolution of Celestial Objects, Chinese Academy of Sciences, P.O. Box 110, Kunming 650216,

People’s Republic of China

arXiv:2406.13146v1 [astro-ph.SR] 19 Jun 2024

4 International Centre of Supernovae, Yunnan Key Laboratory, Kunming 650216, People’s Republic of China

ABSTRACT
The unstable mass transfer situation in binary systems will asymptotically cause the adiabatic ex-
pansion of the donor star and finally lead to the common envelope phase. This process could happen
in helium binary systems once the helium donor star fills its Roche-lobe. We have calculated the
adiabatic mass loss model of naked helium stars with a mass range of 0.35 M⊙ to 10 M⊙ , and every
mass sequence evolved from the He-ZAMS to the cooling track of white dwarf or carbon ignition. In
consideration of the influence of stellar wind, massive helium stars are not considered in this paper.
Comparing stellar radius with the evolution of the Roche-lobe under the assumption of conservative
mass transfer, we give the critical mass ratio qcrit = MHe /Maccretor as the binary stability criteria of
low and intermediate-mass helium binary stars. On He-MS, the result shows 1.0 < qcrit < 2.6, which
is more unstable than the classical result of polytropic model qcrit = 3. After early He-HG, the qcrit
quickly increases even larger than 10 (more stable compared with widely used result qcrit = 4), which
is dominated by the expansion of radiative envelope. Our result could be useful for these quick mass
transfer binary systems such as AM CVns, UCXBs, and helium novae, and it could guide the binary
population synthesis for the formation of special objects such as SNe Ia and GW sources.

Keywords: Binary evolution(154) — Helium-rich stars(715) — Stellar evolution(1599) — Common

envelope evolution(2154)

1. INTRODUCTION Depending on whether the primary star could be lim-

The observation has shown that a large rate of stars ited near the range of the Roche lobe during RLOF, the
exist in binary systems (e.g., Sana et al. 2012; Moe & binary mass transfer could be separated into the sta-
Di Stefano 2017; Chen et al. 2024). If those systems are ble phase and common envelope (CE) phase (Paczynski
close enough, with the evolution of the primary stars, 1976). For the first phase, the donor star remains in
the primaries will expand and fill its Roche-lobe. Finally the Roche lobe and it stays in thermal equilibrium dur-
it will starts the Roche-lobe overflow (RLOF). The mag- ing the mass transfer stage until it loses enough mate-
nitude of binary mass transfer has been proved in the rial and shrinks smaller than its Roche lobe. Secondary
last few decades. Mass transfer in binary systems cre- could have enough time (longer than thermal or nuclear
ates variable events that can hardly be explained in the timescale) to accrete mass to become Algol-like binaries
single-star systems such as double neutron stars (NSs) (Crawford 1955) or create the accretion-disk cataclysmic
and black holes (BHs), type Ia supernova (SN Ia), and variables and low-mass X-ray binaries (LMXBs) (Giac-
gravitational wave (GW) sources (Han et al. 2020; Tau- coni 2005). However, if the donor star loses the stability
ris & van den Heuvel 2023). and considerably exceeds the Roche-lobe radius, a vio-
lent mass loss of the donor star will come out (Sober-
man et al. 1997). It breaks the thermal equilibrium and
Corresponding author: Hongwei Ge triggers the adiabatic expansion. In this case, the stellar
[email protected] radius expands considerably greater than binary separa-
tion. Such unstable mass transfer leads to the CE phase.
2 Zhang et al.

The stability in binary mass transfer is a conclusive issue M = 1.0 M⊙ and M = 5.0 M⊙ Hertzsprung gap (HG)
for the outcome of binary evolution. If the mass transfer stars.
is stable, the accretor will have enough time to increase One inescapable case is the helium binary system. It
mass. If unstable, the donor star will expand further contains a helium dominated core or shell burning star
than binary separation and form the CE. During the and a secondary. The helium stars have a faster life than
CE phase, the reduction of binary orbital energy over- normal stars with the same mass and could also become
comes the binding energy and ejects most of the donor donors during RLOF. The helium star is formed by the
envelope in a short time. CE evolution is able to create ejection of the hydrogen envelope of the progenitor due
a short-period binary system or cause binary merge dur- to binary mass transfer or fast stellar wind. Different
ing CE eject. The difference in binary evolution between from a helium white dwarf (He WD), the center of a he-
stable and unstable mass transfer leads to very different lium star is capable of starting helium ignition. The for-
fates, such as progenitors of SNe Ia, the merge of dou- mation channels of helium stars have been studied by us-
ble white dwarfs, and the formation of short-period hot ing the binary population synthesis method (Han et al.
subdwarf stars. Because the timescale is very short for 2002) and found as the dominant contribution of UV-
the CE phase, it is difficult to find these systems during upturn of elliptical galaxies (Han et al. 2007). Low-mass
unstable mass transfer. Though, there are some poten- helium stars, especially for stellar mass M < 1.0 M⊙ ,
tial remnants of CE ejection or CE merge (Tylenda & are very likely to be observed as B/O type hot sub-
Soker 2006; Tylenda et al. 2011; Ivanova et al. 2013). dwarf stars (sdB and sdO). For massive helium stars
In the meantime, the accretion of a secondary star can M > 10 M⊙ , it is believed to refer to some objects like
also lead to uncontrollable expansion and cause CE evo- Wolf-Rayet stars (WR).
lution. In this article, we focus on the stability criteria of sdB stars lay on the ultra-horizontal branch of the
the donor and ignore the influence of such an accreting Hertzsprung–Russell diagram (HRD), denser and dim-
mechanism. mer than O/B type MS stars (Heber 2009, 2016). Their
In the 1990s, the polytropic model was developed to high temperature and luminosity suggest a helium-
solve the stability criteria problem (Hjellming & Web- burning center instead of hydrogen. Different from pure
bink 1987; Soberman et al. 1997). It simplifies the stellar helium composite stars, a large amount of sdB stars have
structure and is widely used in binary population syn- features of hydrogen absorption lines in the spectrum,
thesis (Hurley et al. 2002; Claeys et al. 2014). Other which is believed to be the symbol of a thin hydrogen
studies are trying to discover the stability criteria of envelope (< 10−2 M⊙ ) near the stellar surface. Besides,
some specific binary systems. Compared with studies sdBs are commonly found in binary systems (Maxted
of the population of observed systems, more results re- et al. 2001; Copperwheat et al. 2011). Most possible
garding the stability of mass transfer have been obtained companions in short-period systems (P<15 d) are white
(e.g., Tauris & Savonije 1999; Podsiadlowski et al. 2002; dwarfs (WD), and some systems could even be as short
Han et al. 2003; Vos et al. 2019, 2020; Leiner & Geller as 1 hr (Schaffenroth et al. 2022), which makes those
2021). However, most of them have not given us the re- sdBs possible ultra-compact X-ray binaries (UCXB) and
sults in the complete parameter space. In our previous low-frequency gravity wave (GW) source. In addition,
Papers (Ge et al. 2010, 2015, 2020a), we introduced the one-third of long-period sdB binaries are composed of
model sequences of adiabatic mass loss from the main MS stars (Vos et al. 2018). With the evolution of he-
sequence (MS) star to the Red Giant Branch (RGB) lium stars, some of the close sdB binary systems could
star and Asymptotic Giant Branch (AGB) star. The experience RLOF.
result shows a great improvement for RGB and AGB The formation of sdBs has a tight relationship with
stars. Our studies are adjusted for the stability predic- binary mass transfer. During this stage, the radius of
tion of classical studies (e.g., Paczyński 1965; Paczyński the primary star (e.g. HG, RGB, AGB) could expand
et al. 1969; Hjellming & Webbink 1987) and have given over its Roche lobe and the stellar mass flew through the
a more stable parameter space for Giant-Branch (GB) inner Lagrangian point (L1 ). In the last two decades,
stars and massive stars. Recently, the 1D simulation of studies in stellar evolution simulation and binary popu-
the mass-transfer evolution by MSEA code (Temmink lation synthesis have drawn a good picture to show that
et al. 2023) has simulated the mass-transfer evolution in mass transfer in close binary systems is a compelling
the mass range from 1 M⊙ to 8 M⊙ and successfully ex- case for removing most of the hydrogen envelope (the
pands the criteria to giant branch stars. Their result and systems are also called stripped helium stars) and leave
our work have formed a good mutual confirmation on a thin hydrogen shell behind (Han et al. 2020). Due to
the low luminosity and strong gravity on the sdB sur-
 He Binary Stars 3

face, the hydrogen envelope of low-mass helium stars In our simulation, we consider a normal Population
could hardly be ejected by the stellar wind during most I metallicity situation (Z = 0.02). We use the over-
of the time of star evolution. shooting parameter δ=0.12 (Schroder et al. 1997; Pols
WR binary systems are much more massive (M > et al. 1998) and the mixing length parameter α=2.0
10 M⊙ ) and brighter than sdB. Strong and broad emis- (Pols et al. 1998). They are good fits to the sun. In
sion lines in their spectrum suggest that the stellar wind the case of focusing on low and intermediate-mass he-
significantly influences its stellar evolution (Hamann lium stars (M < 10 M⊙ ), which more likely bring out
et al. 2006). Because of this, a WR star might eject weak wind around 10−6 ∼ 10−8 M⊙ /yr in most time of
its hydrogen envelope through stellar wind instead of evolution, we override stellar wind during the evolution
RLOF. Despite all this, almost all WR stars we have stage of helium burning. Since observation shows little
observed are in binary systems. Oddly, helium stars are evidence of helium star structure after the core burning
barely found in those intermediate masses (2.0 M⊙ < period, no wind assumption brings another advantage:
M < 10 M⊙ ). It is possibly because of rare quantity it makes more models evolve thicker shells at the later
(compared to sdBs) and less brightness (compared to stage to help simulate larger stars to cover the unknown
WRs). The detection of HD 45166, believed to be a parameter space of helium binaries. On the other hand,
”quasi Wolf-Rayet” (qWR) binary system (Steiner & helium stars have lower mass limits, just like normal low-
Oliveira 2005), first gives us a glance at these stars. Re- mass stars. Once the total mass is lower than 0.35 M⊙ ,
cently, medium-resolution spectra observation in Magel- the center of the star would not form the environment for
lanic Clouds gives us 25 candidates of intermediate he- helium ignition and finally lead to the He WD phase. To
lium binary star with an MS companion (Götberg et al. sum up, we settle the helium star mass sequences from
2023). Their observation shows that intermediate-mass 0.35 M⊙ to 10 M⊙ .
helium stars are similar to WR stars on the spectrum, The observation of sdB/O has ensured that a thin
but the evolution characteristic is close to low mass he- hydrogen shell widely exists on helium stars, though it
lium stars by the weak wind on He-MS. is less than 10−3 M⊙ in most cases. Such a thin hy-
This article will use the adiabatic mass loss model to drogen shell cannot start a shell nuclear reaction and
simulate the unstable mass transfer process of low and become a major influence of energy during the helium
intermediate-mass helium stars. The massive helium star evolution. In this article, we simply override the
star will be studied in the future. Then, we will analyze hydrogen shell and consider the naked helium star ap-
the stability criteria of helium binary systems by com- proximation to simplify the calculation. Producing and
paring the radius between the donor star and its Roche evolving a certain helium star experiences three parts
lobe. Section 2 will introduce the method of creating in our simulation. Firstly, we build a typical zero-age
helium star sequences as donor stars. Based on differ- main sequence (ZAMS) star and let it evolve through
ent radius features, we divide helium stars into different MS. After this part, a helium core is formed at the star’s
stages and select several models for adiabatic mass loss. center. Secondly, at the HG stage, we stop abundance
Section 3 gives the specific process of adiabatic mass change due to helium and heavy metal ignition and start
loss of helium mass sequences. Firstly, we give a sample ejecting the hydrogen envelope by using multiple times
to introduce radius change during adiabatic mass loss of Reimers wind (Baschek et al. 1975). It finally strips
on Radius-Mass diagram. Then we will give the criti- the envelope entirely and exposes the helium core inside.
cal mass ratio qcrit for judging unstable mass transfer. Then, by reducing and increasing the stellar mass, we
Finally, Section 5 will be the summary of this article. can get the mass sequences of such naked helium stars.
At last, once a certain helium core mass is gotten, stop
wind mass loss and re-open the abundance change, such
2. BUILD HELIUM STAR SEQUENCES a helium star starts to evolve as a helium-zero-age main
sequence (He-ZAMS) star.
We must simulate variable helium stars in differ-
We take 1.6 M⊙ helium star as a sample to introduce
ent masses and evolutionary stages to study adiabatic
different stages of helium star evolution. Figure 1 shows
mass loss in the helium binary system. To develop a
the evolution track of a 1.6 M⊙ helium star in the HRD.
sequence of naked low and intermediate-mass helium
The method for creating the helium star in the previous
stars, here we use STARS code which was developed
introduction is shown in sub-figure ’a’. We focus on the
by Eggleton (1971, 1972, 1973) and Paxton (2004).
helium star evolution stage particularly in sub-figure ’b’.
It is a one-dimensional (spherically symmetric) non-
The evolution of the stellar radius is shown in Figure
Lagrangian code. We have already introduced this code
2. The blue and red lines are the radius when the helium
in the second section of Paper I (Ge et al. 2010).
4 Zhang et al.

Figure 1. Evolutionary track of a naked 1.6 M⊙ helium star in the HRD. In this paper, ’log’ represents base-10 logarithm. In
sub-figure ’a’, we show the complete processes of building the helium star. The black dash line shows the 10 M⊙ progenitor
star from ZAMS to HG. Its hydrogen envelope is ejected during the green line. We stop the mass-losing stage until stellar mass
reaches 1.6 M⊙ . The blue line shows the helium star evolution stage. The code is killed at carbon ignition. Here we set the
carbon ignition limit at the of carbon luminosity when LC > 100 L⊙ . Sub-figure ’b’ is the helium star evolution stage. It is
divided into three phases by radius: Radius-increasing phase on He-MS, Radius-decreasing phase on He-MS, and He-HG/GB
phase.

star is expanding, and the gray dash line is for shrinkage. bon/Oxygen core (CO core) behind. Helium star com-
The shadow area is the convective envelope zone. prises a CO core and a helium envelope after He-TAMS.
After helium ignition at He-ZAMS, 3 α reaction starts After He-MS, the CO core star shrinks, and a helium-
at the center, forming a convective core. Similar to burning shell is developed at the bottom of the stellar
the hydrogen burning process on the MS phase of envelope. At this phase, the helium star expands sig-
intermediate-mass normal stars, helium burning finally nificantly like an ordinary star at HG. For those helium
burns off helium inside the convective zone and makes stars at a certain mass range, like 1.6 M⊙ helium star,
star maintain its luminosity. Such helium main sequence their expansion at this phase is so strong that developing
(He-MS) shall maintain for about a nuclear timescale a convective envelope similar to normal AGB stars. By
until center helium is exhausted at the terminal age of the different structures of the helium envelope, we de-
helium main sequence (He-TAMS). part this phase into helium Hertzsprung Gap (He-HG)
On early He-MS, sustaining 3 α reaction, the average and helium Giant Branch (He-GB). The helium-burning
relative atomic mass at the center convective core keeps region keeps supplying mass for the CO core during the
increasing. It finally led to radius expansion. At the He-HG and the He-GB phases.
end of He-MS, the center helium fraction decreases lower On the one hand, more massive stars build up enough
than 0.2 so that 3 α shall not be able to keep burning mass for carbon ignition. However, due to the roughness
at the center. With a decrease of helium and increase of elements and star nets, our work does not simulate
of carbon in the center, the 4 α reaction gradually takes the evolution after carbon ignition. Here, we technically
advantage of later He-Ms and turns appreciable carbon set the limit of the carbon ignition when the luminosity
into Oxygen. Conversely, star goes through the shrink- of carbon ignition LC > 100 L⊙ . If the luminosity of
age stage at the later He-MS phase. At He-TAMS, carbon ignition is lower than this limit, the helium star
the convective core gradually disappears, leaving a car- will evolve to the shrinkage stage and end as a WD. If
 He Binary Stars 5

of the burning shell reaches the surface of the helium

star, helium ignition stops and forms a carbon/oxygen
white dwarf (CO WD) in the end. However, we cannot
give more details of the following evolution due to the
small number of stellar shells and elements. The possi-
ble carbon flash and supernova phases are still directly
unknown in our simulation.
Here we compare our results with the final outcome of
AGB stars by Doherty et al. (2017) and give a predic-
tion of their ending. The final status of our helium star
sequences is shown in Figure 3.
For helium stars with MHe < 1.4 M⊙ , they will lead
to the cooling track and end as CO WD. The carbon ig-
nition power is always lower than 100 L⊙ . Such a weak
power of carbon flash can not make the burning flame
dredge into the center. It shall form a Ne shell inside
the CO core, identified as a carbon/oxygen-neon white
dwarf (CO-Ne WD). For helium stars in the mass range
of 1.4 M⊙ < MHe < 2.3 M⊙ , they have enough core
Figure 2. Radius evolution of 1.6 M⊙ helium star from He- mass for carbon ignition, and the flame of carbon flash
ZAMS to C ignition. We change the ticks of He-MS and is strong enough to dredge into the center of the degen-
He-HG/GB to enhance the tendency of radius change. Dot erate core. After the carbon is exhausted, it finally turns
lines represent the maximum radius of He-MS. The shadow
into a degenerate oxygen/neon white dwarf (ONe WD).
area is the convective envelope zone. The models we choose
for adiabatic mass loss are black dots. For some massive ONe WDs, if the degenerate core is
heavier than around 1.375 M⊙ , the electron-capture pro-
not, the fierce carbon ignition flame will cause a tem- cess will happen and lead to ECSN. For more massive
perature spike inside of the core. It finally reaches the helium stars (MHe > 2.3 M⊙ ), helium stars eventually
boundary of our program calculation. form a non-degenerate CO core and start carbon and
Carbon ignition is a complex process. For degenerate oxygen ignition in the center. It finally cause the iron-
CO core cases, carbon ignition might be unstable and collapse event in the center and forms the CCSN. Due
cause carbon flash events or become supernovae (SNe). to the stars’ stripped hydrogen envelope during the first
In this paper, we roughly separate carbon flash and non- mass transfer, the supernovae should not obtain hydro-
degenerate core (similar to more massive stars) by center gen lines and become SNe Ib (Tauris & van den Heuvel
degeneracy Ψc = 3 (Doherty et al. 2017). The carbon 2023). We also notice that the very detailed simulation
flash is similar to helium flash for some intermediate- by Woosley (2019) also gives the final outcomes for he-
mass RGB stars going through the horizontal branch. lium stars. The main difference is the consideration of
Simulation shows that carbon flash is off-center as well. the stellar wind. In our simulation, with out the wind,
It is still unclear whether the carbon flash will reverse the minimum mass of carbon ignition will be lower due
the degenerate core like helium flash or lead to the to the helium shell burning. Excluding that, the results
electron-capture supernova (ECSN) phase. For helium are similar.
star with a non-degenerate core, carbon ignition creates To study the mass transfer stage, we pay more at-
an onion structure inside the core, finally leading to the tention to the structures of the different helium stars.
iron core-collapse supernova (CCSN). In the case of the The main stages for mass transfer are limited to He-
very lifetime after carbon ignition, the helium envelope MS/He-HG/He-GB. Notably, some helium stars do not
is unlikely to be wholly ejected. Finally it leaves helium experience He-HG/He-GB. Figure 4 shows some repre-
emission lines in the supernova spectrum (Tauris & van sentative samples of evolution stages in our helium stars
den Heuvel 2023). Both situations would quickly evolve sequences. Different from the red giant branch (RGB)
to a supernova phase, which makes it worthless to con- stars, only a narrow mass range (0.9 M⊙ < MHe <
sider binary mass transfer after carbon ignition. There- 2.0 M⊙ ) of helium star can evolve to He-GB phase.
fore, our simulation stops at carbon ignition. On the For MHe > 2.0 M⊙ helium stars, they failed to de-
other hand, a low-mass helium star could not start car- velop a convective envelope before carbon ignition. For
bon ignition due to its thin envelope. When the flame MHe < 0.9 M⊙ helium stars, their helium envelopes are
6 Zhang et al.

Figure 3. The final status of helium stars. We roughly separate them by the carbon luminosity LC and center degeneracy Ψc .
For MHe < 1.4 M⊙ stars, they will become CO WDs. For 1.4 M⊙ < MHe < 2.3 M⊙ stars, off-center carbon flash reserves the
degenerate core, and they end as ONe WDs. For MHe > 2.3 M⊙ helium stars, carbon ignition starts in the center and they lead
to CCSNe. Due to the helium envelope, the hydrogen lines do not exist in the spectrum (SN Ib). The special situations, CO-Ne
WD and ECSN, can not be certain in our simulation. We roughly give their top limits in the figure.

too thin to develop a convective zone before the cooling simulate stellar evolution during mass loss from surface
track of WD. Another unique part lies on MHe < 0.6 M⊙ to center. Such adiabatic mass loss could cause varying
helium stars: they barely expand during the He-HG degrees of stellar expansion, and the stellar radius may
phase out of the super-thin envelope. This radius be- finally be larger than the separation of the binary sys-
havior means their maximum radius on He-HG could be tem. This process is likely to become the CE evolution
lower than it on He-MS. phase.
To study mass transfer in close binary systems, we We precisely simulate the adiabatic mass loss pro-
give more attention to the radius change of the donor cesses of different helium stars and give the stability
star. According to the previous part, helium stars can criteria for the CE phase. Section 3.1 gives a sample of
divide into three stages based on radius change: the first 1.6 M⊙ helium stars. We describe the method of getting
expanding stage on early He-MS; the shrinkage stage stability criteria in this Subsection and finally sum up
on later He-MS and the second expanding stage on He- all mass sequences in Section 3.2. In this section, we
HG/GB. The symbols are shown in Figure 1-b. To help show our results as a critical mass ratio (qcrit ) in the
adiabatic mass loss simulation, we pick multiple models Mass-Radius diagram.
on three stages of radius change and sign them up as
bot markers in Figure 2.
3.1. Adiabatic Response And Stability Criteria
3. ADIABATIC MASS LOSS This section will show a sample of 1.6 M⊙ helium star
We use the method in Paper I (Ge et al. 2010) to sim- sequence. Like the description in Section 2, 1.6 M⊙ he-
ulate the adiabatic mass loss situation for these selected lium star could evolve through all stages from He-MS
models. In this part, we keep locking the entropy pro- to He-GB. That makes it a representative case of study
file within the mass coordinate of every helium star and adiabatic mass loss at each stage. Our work is based
 He Binary Stars 7

Figure 4. Mass sequences of helium stars in the HRD. We divided them into three parts out of their envelope feature. Blue
lines: carbon ignition before developing convective envelope; Green lines: could evolve convective envelope; Red lines: envelope
exhausted before developing convective envelope. The models that start carbon ignition are shown as black stars.

on the adiabatic mass loss model, already introduced in at the bottom of the envelope region enlarges the mass
Paper I (Ge et al. 2010). of the CO core. In this 1.6 M⊙ case, the CO core shall
To clear the most primary input quantities, we give keep electron degenerate. When the degenerate CO core
the stellar entropy profiles in Figure 5-a. It shows some reaches a certain status, carbon flash will start at a spe-
representative evolution moment of 1.6 M⊙ helium star. cific region (off-center). The final fate of different stars
These moments are shown on the HRD in this figure. have been discussed in Section 2. Before carbon ignition,
On the He-MS evolution stage, the convective core con- the overly expanding convective envelope could signifi-
tracts with the decrease of the helium fraction, which cantly influence the efficiency of internal energy conduc-
causes a decrease in core entropy. Outside the core, en- tion. It finally leads to the formation of a negative (su-
tropy increases with the mass profile to show a much peradiabatic) specific entropy profile at this evolution
more stable structure than the convective zone. After stage.
He-MS, the convective core disappears due to the helium We put these stellar models into adiabatic mass loss
exhaustion at the center. Meanwhile, the core contract simulation, and the characteristic adiabatic response of
leads to gravitational potential energy release. At the a star to mass loss is shown in Figure 5-b. In these cases,
He-HG stage, the helium envelope starts expanding and radius characteristics can be concluded into two modes
entropy increases rapidly by absorbing the gravitational by their envelope structure. Before the base of He-GB,
potential energy from the center. Helium stars must the stellar envelope is radiative and keeps a relatively
keep a radiative envelope until the base of He-GB. Here, dense area near the surface. Entropy rapidly increases
the surface area is getting too thin and cool to keep from the core to the surface. With the process of adi-
the structure. The convective envelope develops from abatic mass loss, although it is experiencing adiabatic
surface to core and forms an equal-entropy zone inside expansion, the cold and low-entropy shell is going to
this area. On the He-HG/GB stages, helium burning be quickly exposed, finally leading to the contraction
8 Zhang et al.

Figure 5. In sub-figure ’a’, we show the initial entropy profiles of 1.6 M⊙ helium stars in different evolution stages. Sub-figure
’b’ shows radius changes during adiabatic mass loss.

of the stellar radius during mass loss. For the He-MS with the radius inside the envelope, which could cause
stage, when the surface reaches the inside of the ini- the thermal timescale mass transfer rate.
tial stellar envelope, the adiabatic mass loss would not We use Kelvin–Helmholtz timescale (τKH ) as a good
lead to a rapid radius change due to an almost constant approximation of thermal timescale and calculate the
entropy profile. Such details of radiative response are thermal timescale mass transfer rate as ṀKH =
introduced in Paper II (Ge et al. 2015). On the other MHe /τKH . In most cases, the binary mass transfer will
hand, the convective envelope gives us another picture be stable if Ṁ < ṀKH . If the Roche-lobe radius (RL )
of radius evolution. A convective envelope has a less is equal to the radius RKH , the mass transfer rate will
dense structure. During the adiabatic mass loss, the equal to ṀKH . As a critical situation, the unstable mass
high-entropy area is exposed due to the superadiabatic transfer quickly appears once Roche-lobe reaches RKH ,
zone. The helium stars expand rapidly and keep the then adiabatic mass loss shall happen in the binary sys-
maximum radius until the convective envelope is wholly tem.
ejected. Such details of superadiabatic expansion re- Stellar radius and RKH responses are described in Fig-
sponse are very similar to the behavior of the RGB star. ure 6. The black solid line represents the radius re-
It was introduced in Paper I (Ge et al. 2010). sponse, and the grey dashed line is RKH . Clearly, the
With the help of adiabatic mass loss response, we donor star can not start the unstable RLOF at the be-
now consider the specific situation in helium binary ginning of binary mass transfer in this sample. To show
mass transfer. A non-negligible He-HG stage issue is the boundary reaches the limit of ṀKH , we introduce a
the delayed unstable mass transfer. The mass transfer mass–radius exponent ζad ,
rate is strongly related to the difference between stel-  
lar and Roche-lobe radius, so reaching the dynamical ∂ ln RKH
ζad = (1)
mass transfer rate at the beginning is impossible. The ∂ ln M KH
mass transfer rate will likely gradually reach a thermal
timescale mass transfer rate. Then, internal energy will It represents the donor star’s adiabatic response to mass
not be able to be rebalanced inside the stellar envelope, loss and we have already introduced it in Paper I (Ge
which will finally accelerate to an adiabatic mass trans- et al. 2010). Here, we choose a He-TAMS and a He-
fer. Here, we use the method that we introduced in GB model to introduce how to get qcrit in the helium
Paper I (Ge et al. 2010), replacing the stellar radius binary system. We can easily conclude that RKH is
 He Binary Stars 9

Figure 6. Radius evolution tracks of two 1.6 M⊙ helium stars during adiabatic mass loss within the mass coordinate. In
sub-figure ’a’, we show the total radiative structure of a He-TAMS star. In sub-figure ’b’, it is a convective envelope on the giant
branch. The black solid line is the radius change. The red dash line is the initial radius profile. Because of adiabatic expansion,
we can see that the surface radius is always larger than the initial profile. Using the method as the trigger of unstable mass
transfer in Paper I (Ge et al. 2010), we show the maximum radius of Roche-lobe as a grey dash line. Assuming binary mass
transfer is conservative and the mass ratio is known, the RL can be rebuilt as a blue dash line during mass transfer.

much smaller than the stellar radius on He-GB, but He- in Figure 7. In this figure, qcrit is a function of ra-
TAMS does not show such scene instead. Actually, that dius. With the evolution of the helium star, the radius
is special for convective envelopes due to the low-density is increasing on the early He-MS and He-HG/GB, as
area. In this work, we assume a conservative situation we introduced in Section 2. In the early He-MS and
for binary mass transfer, which means all mass through He-HG stage, due to radius increasing, qcrit have to get
RLOF from helium donor transfer to accretor, and no larger to reach the RKH profile. Things are quite dif-
material is lost from the binary system. Paper I (Ge ferent in the later He-MS and He-GB. For the shrinking
et al. 2010) has shown how the law of conservation of stage on later He-MS, the radius of the inner envelope
mass and angular momentum would influence the RL . changes faster than the surface (see the difference in
In conservative mass transfer, the RL shall determined Figure 5-b), and qcrit is still getting larger alone with
by the initial mass ratio (q). We define it as equal to stellar evolution. After the He-HG stage, the convec-
donor mass divided by accretor mass (here, the donor is tive envelope quickly evolves inward from the surface.
especially represented by a helium star). When the mass of the surface convective envelope in-
creases to around 10−3 M⊙ /yr, the stellar mass transfer
q = MHe /Macc (2)
suddenly becomes unstable due to the development of
Blue lines show the RL by different q. If it is constantly the low-density envelope. With a helium star ascending
larger than RKH , the mass transfer rate will not be high the He-GB, the convective zone dredges inward and the
enough to start the adiabatic mass loss. Keep increasing stellar radius increases rapidly. After the beginning of
q, unstable mass transfer stage comes out when RL is He-GB, the quickly increasing radius and the short ther-
lower than RKH . Here, we define the minimum initial mal timescale dominate the stability criteria, and qcrit
mass ratio q as the stability criteria qcrit , just letting RL increases during most of He-GB.
tangents with RKH . For He-GB stars, our simulations face a great chal-
Using the method, we calculated the qcrit during lenge in calculating adiabatic expansion ( the failed data
1.6 M⊙ helium star evolution track using the selected is shown in Figure 7). In this narrow parameter space,
models in Section 2. The stability criteria are shown
10 Zhang et al.

Figure 7. qcrit for 1.6 M⊙ helium stars from He-ZAMS to C ignition. The results keep increasing on He-MS and He-HG. Due to
the development of the convective envelope, qcrit firstly decreases at the beginning of He-GB and then increases at later He-GB
because of the radius increasing along with evolution. The ’Failed data’ area is for the models that we can not appropriately
simulate due to the steep entropy profile at the superadiabatic zone near the stellar surface. The red dash dot line is the
maximum mass ratio of Darwin instability (DWI), which is discussed in Section 4.

the radius is increasing too fast to calculate the appro- In this section, we introduced the method of calcu-
priate response for adiabatic mass loss. This difficulty lating the stability criteria of adiabatic mass loss and
is caused by the increases of the entropy in the supera- the results of a 1.6 M⊙ helium star. Generally, stellar
diabatic zone. The mass loss brings a higher entropy models with higher qcrit become more stable during adi-
surface and leads to faster expansion. Due to the posi- abatic mass loss. It might not be reasonable for helium
tive feedback, our simulation fails to solve the structure stars to have a much more stable mass transfer chan-
response in the superadiabatic zone around the surface nel on He-HG/GB (qcrit > 10). We expected it to be
area, which is similar to the situation on AGB stars in dominated by the short thermal timescale mass transfer
Paper III (Ge et al. 2020a). When it comes to a later during outer Lagrangian point (L3 ) at these stages. It
stage before carbon ignition, we notice that the supera- will be discussed in Section 4.
diabatic zone evolves to a less steep entropy profile (see
3.2. Critical Mass Ratio In M-R Parameter Spaces
Figure 6). We calculated the stability criteria here. The
structure of helium stars on He-GB is similar to low Using the same method in Subsection 3.1, we have
and intermediate-mass normal RGB stars (both a con- calculated all the models in different mass sequences.
vective envelope with a degenerate and non-degenerate To show the key process of radius increasing, we use
core). Comparing the qcrit tendency on RGB in Paper II M − R parameter space (also called Webbink diagram)
(Ge et al. 2015), we have not seen special changes on the to show the stability criteria of helium star mass se-
giant branch. So here we assume the qcrit is increasing quences. Given an intuitive radius of the donor star, we
at a later stage of He-GB. could easily find the evolution stage at the beginning
of the RLOF. There are several mechanisms to make
the donor fill its Roche lobe. The most possible one is
 He Binary Stars 11

Figure 8. Initial results of stability criteria for helium mass sequences in Mass-Radius (M − R) parameter space. We use
1/qcrit to highlight the lower results that qcrit < 10. Here, we ignore the shrinkage at the later MS stage, and just consider
the expanding phase of evolution. We show them in three parts: Early He-MS stage (Case BA) in sub-figure ’a’; He-HG stage
(Case BBe) in sub-figure ’b’; and He-GB stage (Case BBl) in sub-figure ’c’. The shadow area represents failed data introduced
in Section 3.1.
the radius expanding during evolution. Some processes diagram. Noticing that helium star evolves faster than
(like magnetic braking (Rappaport et al. 1983), gravity the accretor, the initial mass of the helium star may be
wave, and tidal instability) could reduce the separation lower than the companion. In other words, the initial
of the binary system to create the possibility for the mass ratio of the mass transfer stage could have q < 1.
donor to fill its Roche lobe even during the shrinkage We define the Case BA phase as when helium stars
stage. However, when the RLOF is undergoing, the an- fill their Roche-lobe at the He-MS stage. Case BA shall
gular momentum transfer with the mass flow will dom- end at the maximum radius on He-MS, and the helium
inate the binary separation and period change. Other star could unlikely fill the Roche lobe in the later He-MS
mechanism can be overridden at this stage. It is obvious stage due to the decreasing radius. At the He-HG/GB
to find that stellar expansion is the primary stage for bi- stage, the helium star could fill the Roche lobe when
nary mass transfer. Suppose we ignore the possibilities the radius is larger than the maximum radius on He-MS,
of mass transfer on the shrinkage stage, dominated by which we define as Case BB. Depending on the structure
quick binary orbit reducing, and just consider the donor of the envelope, Case BB can be separated by Case BBe
radius expanding as the only reason for filling its Roche (radiative) in He-HG stage and Case BBl (convective)
lobe. We can find all mass transfer phases on the M − R in He-GB stage.
12 Zhang et al.

Figure 9. Fitting qcrit for helium mass sequences in M − R parameter space. This result is using RBF interpolating to fit
the initial result of qcrit . Shadow area represents failed data introduced in Subsection 3.1. To highlight the large results on
He-HG/GB and following the observation tradition, we use 1/qcrit instead to show this parameter space.

Figure 8 shows the qcrit on the M −R parameter space. In this part, we will analyze our results and compare
In the early He-MS stage, the results of qcrit are within them with other research and observations.
the range of 1.0 to 2.6. On the He-HG stage, it ranges Our result show 1.0 < qcrit < 2.6 for He-MS donor
from 2.0 to several hundred. On early He-GB, adiabatic stars. During the evolution of a given mass helium star,
mass transfer quickly becomes unstable due to convec- qcrit is getting larger from He-ZAMS to He-TAMS. How-
tive envelope, but qcrit does not decrease below 10.0. ever, with the increase of stellar mass, qcrit of the same
The distribution of initial results can be used to fit the evolutionary stage helium stars are getting smaller. This
stability criteria on M − R parameter space. Because of trend of helium stars is similar to normal MS stars. The
the continuity of stellar structure with mass increasing criteria from He-HG to He-GB also have the same ten-
and evolution, the criteria qcrit also shows continuity on dency as HG and RGB stars, but the critical mass ra-
the M − R diagram. Due to the initial data being un- tios qcrit are much larger (compared with Paper II by
evenly distributed, we use Radial Basis Function (RBF) Ge et al. 2015). This result is due to the much larger
interpolation to solve the fitting result. Figure 9 shows entropy profile of helium stars. Considering that unsta-
fitting qcrit for the helium sequences we studied. ble mass transfer should only occur when the mass ratio
The fitting results are very suitable for binary pop- is larger than qcrit , our results may suggest that more
ulation synthesis. Due to the extreme mass ratio be- binary systems tend to avoid dynamical timescale mass
ing rare to be found and separated both in theory and transfer for helium donor stars compared with ordinary
observation, in Figure 9, we use 1/qcrit as the symbol HG and RGB donor stars.
instead of qcrit and ignore the specific difference when However, for some developed He-HG and He-GB stars,
1/qcrit < 0.05. This way of expression is based on the we notice that the qcrit are extremely large and we barely
observer’s tradition. see such mass ratio in observation. The extreme mass
ratio systems may be dominated by the Darwin insta-
4. DISCUSSION bility (Darwin 1879; Eggleton & Kiseleva-Eggleton 2001;
In the previous sections, we illustrated the method Sargsyan et al. 2019), it will finally lead to the merging
and concluded the results of the stability criteria in he- stage in a short timescale by the tidal effect. By assum-
lium binary systems using the adiabatic mass loss model. ing a tidally locked binary system and overriding the
 He Binary Stars 13

rotational angular momentum of secondary star, we can et al. 2015).

calculate the maximum mass ratio for helium binary sys-
tem (Rasio 1995). We give the dimensionless gyration 3 1
log (P/d) = log (RL /R⊙ ) − log (M/M⊙ )
radius of helium star by using the moment of inertia. A 2 2 (3)
sample is shown in Figure 7. Luckily, compared with + log g(q) − 0.45423,
HG and GB stars, the helium stars are compact enough
to bring a very small moment of inertia. As a result, where P represents the orbital period in days, g(q) is a
Darwin instability (DWI) in the helium binary allows a very weak function of mass ratio q and we use qcrit as
very large critical mass ratio on He-HG/GB. After He- the initial mass ratio:
GB stage,The qcrit in our research is larger than max- ! 32
 12 2 1
imum mass ratio of DWI. Still, the actual rotation of 0.6 + q − 3 ln (1 + q 3 )

2q
helium and secondary star certainly decrease the max- g(q) = . (4)
1+q 0.6 + ln 2
imum mass ratio. We can only give a rough prediction
that DWI may influence our conclusion in the systems
Using this relation, we change the criteria to the mass
with high stellar rotation speed or after He-GB.
ratio-orbital period (q − P ) parameter space and com-
Furthermore, the stability criteria of helium stars on
pare the critical results with other theories and obser-
the He-GB stage are still uncertain. By the influence
vations. We recalculate the relation of qcrit and binary
of a very short KH thermal timescale, the mass transfer
orbital period in Figure 10 and Figure 11. The color bar
could turn to the unstable phase even when Ṁ ≤ ṀKH
in these two figures represents the stellar mass. We light
and the stellar radius will quickly reach the outer La-
the mass on the expansion stage of He-MS and He-HG.
grangian point. On the other hand, the thick shell of
In the polytropic model, the results of qcrit are con-
delayed unstable mass transfer during He-HG leads to
stant values. For He-MS stars, qcrit = 3, for He-HG
thermal timescale mass transfer and causes unexpected
stars qcrit = 4 and for He-GB stars qcrit = 0.78 when
expansion. These helium stars may undergo a thermal
the accretor is non-degenerate star (Hurley et al. 2002;
timescale mass transfer process (Ge et al. 2020b) before
Claeys et al. 2014). Compared with the constant cri-
reaching the dynamical timescale mass transfer. In the
teria of the polytropic model, the qcrit in our research
future, we will study such processes in helium binary
is evolving with the stellar status. In Figure 10, we
systems.
give the specific difference. Our result indicates a more
This study calculates the critical mass ratio by assum-
unstable parameter space compared with polytropic re-
ing that mass and angular momentum are conserved
sults of q = 3 for He-MS donors. Our results suggest
during mass transfer. Fortunately, the non-conserved
that more helium binary systems can evolve to unsta-
mass transfer will only change the orbital evolution. In
ble mass transfer phases like contact binaries, and fewer
other words, the donor’s response during adiabatic mass
systems could evolve through stable mass transfer to
loss is independent of the binary system’s orbital evolu-
become AM CVn or UCXB systems. For He-HG donor
tion. We can easily present the critical mass ratio for
stars, our results on parameter space cover both sides
non-conserved cases, such as the results applied in the
of q = 4, which means the specific difference from poly-
double WDs study by Li et al. (2023). Picco et al. (2024)
tropic model relies on stellar mass and structure. Gen-
also provide a method to derive the critical mass ratio
erally speaking, it is more unstable than the polytropic
of non-conserved mass transfer based on Ge et al.’s se-
model for early He-HG stars when MHe > 2, but it is
ries work, and they systematically study the forming of
more stable on the other part which covers more pa-
merging double compact objects.
rameter space. Thus, we predict the birth rate of SNe
Although we have given qcrit in M − R parameter
Ia should be more significant in the Case BB channel.
space, the mass and radius can not be directly observed.
More long period helium binaries will go through the
The calculation of stellar mass and radius must intro-
stable mass transfer stage and produce long period dou-
duce the stellar structure model, which causes unneces-
ble WD systems. However, further study of the realistic
sary effects when compared with the stability criteria.
thermal timescale mass transfer still needs to confirm
Here, if we consider the stellar radius as the Roche lobe
this prediction.
radius R = RL , the relation of orbital period, mass, ra-
We also compare the network of helium binary pro-
dius, and mass ratio is easily found (see Paper II by Ge
genitors of SN Ia (Wang et al. 2009, 2014). In their
simulation, the mass transfer rate Ṁ > 10−5 M⊙ /yr is
considered as the criterion of the unstable mass transfer
situation (see dot line in Figure 10). They lock the initial
14 Zhang et al.

Figure 10. The relation of 1/qcrit and stellar binary orbital period. The color bar represents the stellar mass. We light the
color on the expansion stage of He-MS and He-HG. The dotted line is a critical boundary for the network of progenitors of SN
Ia when Maccretor = 1.2 by Wang et al. 2009. Horizontal lines are the criteria of the polytropic models on He-MS and He-HG by
Claeys et al. 2014. The data K24 of ϕ Per-type binaries is from Klement et al. 2024. We also put two indirect helium binaries,
NGC 247 ULX-1 and V445 Pup, in this figure.

accretor’s mass and change the periods and the initial 0.17 M⊙ to 0.45 M⊙ , the structure difference between
helium star’s mass (in this figure, we show the results He WD to helium star is significant. By using our cri-
when Macc = 1.2), which makes it difficult to compare. teria, we expect the mass transfer to be stable if the
We noticed that the network is incomplete, which causes donor star is a helium star at a similar mass range from
the stability criterion to be incoherent for some period. 0.35 M⊙ to 0.45 M⊙ with the 1.4 M⊙ NS companion.
Still, we can see our result on He-MS is more unstable, As we seek the comparison object from observed he-
which can be explained by the bias of conservative mass lium star binaries, we find it challenging to test our
transfer. The He-HG stage gives an opposite prediction. result directly. The critical problem is the lack of he-
Firstly, their limitation may be too strict to cover the lium binary systems at the early stage of mass trans-
thermal timescale mass transfer into a stable situation. fer, though the possibility of this stage is high. If a
Secondly, delayed unstable mass transfer plays a signif- helium star fills its Roche lobe on He-MS, the orbital
icant role during the He-HG stage. period will be only a few minutes to hours. The pos-
For low-mass helium stars, there is an overlap between sible binary objects will be AM CVns (accreting WD
the possible mass ranges of helium stars and He WDs companion) and ultra-compact X-ray binaries (UCXBs;
from 0.35 M⊙ to 0.45 M⊙ . Sometimes, the observation accreting NS companion), which are hydrogen-poor in
cannot separate them clearly, especially in short orbital the spectrum with degenerate companions. Due to the
period systems. We notice that the stability studies of weak observational features, these systems make solv-
NS with He WD systems have shown no space for un- ing the orbital parameters and type of component dif-
stable mass transfer on Mass-Period parameter space ficult. Only a few AM CVns are believed to have gone
(Chen et al. 2022). Though the mass range covers through the helium channel and just passed the stable
 He Binary Stars 15

Figure 11. Same as the previous figure, but use qcrit instead. This figure focuses on the He-MS and early He-HG stage at the
short-period side. The data points are the potential helium star channel AM CVn stars in Solheim 2010.

mass transfer (Solheim 2010). We plot them in Fig- For mass-transferring helium stars on the He-HG/GB
ure 11. However, for these systems, the mass of the stage, the initial orbital periods are from days to hun-
donor is almost entirely lost. They have entered the dreds of days. With the extremely short timescale of
late stage of mass transfer, which makes them less suit- He-HG/GB stars and the limited mass range around
able to compare with the stable criteria. Besides, due to 2.0 M⊙ > MHe > 0.9 M⊙ , We expect their possibility
the qcrit on He-MS is lower than the polytropic model, to be observed is slim. They will likely be found as
our results prevent more massive He-MS stars to be- low-mass X-ray binaries (LMXBs) with an NS compan-
come the donor of AM CVn systems. The upper mass ion. If the companion is a B-type MS, the accretor may
limit is MHe = 0.95 M⊙ when MWD = 0.4 M⊙ and spin up and become Be star due to angular momentum
MHe = 1.77 M⊙ when MWD = 0.8 M⊙ . transfer. It may form the ϕ Per-type binary, which con-
Our results are unsuitable to compare with some sists of sdO and Be star. Here, we use the recent data
sdB/O binary systems, which are going through the last (Klement et al. 2024). The periods of these systems are
stage of stable mass transfer to strip the hydrogen en- longer than 60 days. Due to the massive mass of Be
velope and expose its helium core. These systems are stars, the mass ratios of these systems are not near to
considered as stripped helium stars. Typical systems our stability criteria. Still, they are not good candidates
including VFTS 291 (Villaseñor et al. 2023) and LB-1 for comparison due to the mass ratio being around 0.05
(Liu et al. 2019; Shenar et al. 2020) do not start RLOF to 0.2, which is far from our criteria for He-HG. How-
because of the evolution of helium star. In fact, they are ever, the criteria suggest a stable mass transfer stage
likely preparing to reach He-ZAMS after the hydrogen fitting the observation. Similar to observed AM CVns,
shell is ejected. Hence, we do not analyze these systems they are not at the beginning of mass transfer. In ad-
in this article. dition to the direct observation, some systems show in-
direct evidence of helium binary mass transfer. If the
16 Zhang et al.

secondary is NS or BH, the stable mass transfer at a in 2.0 M⊙ > MHe > 0.9 M⊙ could experience the fierce
high accreting rate will create a high X-ray luminosity expansion and create a convective envelope on the He-
and become an ultra-luminous X-ray source (ULX). A GB. For MHe > 2.0 M⊙ donors, the convective envelope
potential ULX which has a helium donor is NGC 247 fails to be built by the premature carbon ignition.
ULX-1. However, the distance is too far to detect the Then, we analyze the adiabatic mass loss response of
binary parameters. Recently, Zhou et al. (2023) gives a different helium donor stars. After we get the radius
prediction of helium star mass by assuming a 1.4 M⊙ NS change of the adiabatic mass loss process, we rebuild
accretor (MHe = 0.6 ∼ 2.0 M⊙ , Porb = 2.4 ∼ 21 d), and the Roche-lobe radius and compare them in the different
the rough result is shown on Figure 10. Besides, the only mass ratio systems. We conclude the results by giving
known helium nova V445 Pup is believed to have evolved the stability criteria qcrit on M − R diagram. Here, the
from a WD accreting mass from a helium donor due to binary mass and angular momentum transfer are con-
the dominant helium lines in the nova’s spectrum. A served. Generally speaking, the qcrit increases with the
helium nova outburst caused by the helium flash of the evolution of the helium star. For main-sequence helium
accreting shell on the WD surface is the most favorable donor stars, the results show 1.0 < qcrit < 2.6. After
explanation for this event, which reveals a stable mass early He-HG, the qcrit quickly increases larger than 10
transfer stage before this event. Here, we use the result for helium donors. With increased stellar mass, the qcrit
(MWD = 0.8 M⊙ , MHe = 0.67 M⊙ by using d = 8.2 kpc) of the He-MS is getting smaller. This result is similar
calculated by Banerjee et al. (2023). It is also shown to the tendency of a typical star from MS to RGB.
in Figure 10. We can easily find that neither is a good In the last section, we compare the results with dif-
limit for stability criteria. All the systems are far from ferent stability criteria. Our result indicates a more
the qcrit . unstable parameter space for He-MS donors than the
For some sdB/O binaries, they have yet to reach the polytropic model, which indicates that more He-MS
RLOF stage. Using our criteria, we could predict the shall go through CE or contact binary systems, and the
mass transfer stage. We take the eclipsing sdO binary birth rate of SNe Ia through Case BA should be lower.
system HD49798 as a sample (MWD = 1.28 ± 0.05 M⊙ , Our critical mass ratios for the early He-HG donors are
MHe = 1.50 ± 0.05 M⊙ and P = 1.55 d by Brooks et al. smaller than previous results, which is more unstable.
2017). The mass ratio and period are shown in Fig- However, the qcrit rapidly increases after early He-HG
ure 10. Due to the q < qcrit in the q − P parameter and becomes highly stable. Such extreme mass ratio
space, the Case BB mass transfer should be stable. This systems may only be influenced by Darwin instability
method can also be used to guide helium binary popu- on late He-GB. The specific evolution at this stage may
lation synthesis. rely on further research on the non-conserved thermal
As we can see, there have been few detections of he- timescale mass transfer.
lium binaries during RLOF up to now. The observation Finally, we compare our results with some observed
of helium binaries is still strongly required. We hope helium binary systems going through RLOF. For AM
there will be more detections, especially for high-time CVns and ϕ Per-type binaries, our criteria do fit these
resolution surveys, to find more systems to compare with objects, but the mass ratio of these systems varies far
different criteria. These systems will be very important from qcrit . Other indirect helium binaries like helium
for us to understand the mass-transfer evolution of he- nova and sdB ULXs are also not good candidates for
lium stars. validating our theoretical results. Our results can be
applied to the binary population synthesis code to study
5. SUMMARY helium binary systems.
This study attempts to give stability criteria of dy-
namical mass transfer in low and intermediate-mass he- 6. ACKNOWLEDGMENTS
lium binary systems. In this article, we use the adiabatic This project is supported by the National Natu-
mass loss model (Paper I Ge et al. 2010, II Ge et al. ral Science Foundation of China (NSFC, grant Nos.
2015, III Ge et al. 2020a) to study the stellar structure 12288102, 12090040/3, 12173081), National Key R&D
response of helium donor during binary mass transfer. Program of China (grant Nos. 2021YFA1600403,
We first simulate the evolution of a single helium star 2021YFA1600401), the Key Research Program of Fron-
as the donor. To get the most developed structure of He- tier Sciences, CAS (No. ZDBS-LY-7005), CAS-
HG/GB stars, we ignore the stellar wind of helium stars. Light of West China Program, Yunnan Funda-
In our simulation, MHe < 0.6 M⊙ helium stars can only mental Research Projects (Nos. 202101AV070001,
go through the He-MS mass transfer stage. Helium stars 202201BC070003), Yunnan Revitalization Talent Sup-
 He Binary Stars 17

port Program - Science & Technology Champion 202302AN360001). L.Z thanks the selfless help of H.G,
Project (No. 202305AB350003), and International X.C and Z.H. L.Z also appreciate Jingxiao.Luo and
Centre of Supernovae, Yunnan Key Laboratory (No. Luhan.Li for the guidance of possible observation ob-
jects of helium binaries.

APPENDIX

A. PHYSICAL PARAMETERS AND MASS TRANSFER STABILITY CRITERIA OF HELIUM STARS

Here we show the data of mass sequence 1.6 M⊙ as samples. The stellar radius versus qcrit in this table is shown in
Figure 7. The completed data of mass sequences are released in the machine readable table. The columns of Table 1
are as follows:
1. k — sequence model number, indicate the order from He-ZAMS to C ignition;
2. age — evolution age measured from He-ZAMS model (k=1 model);
3. MCO — carbon/oxygen core mass in solar unit;
4. log R — initial helium stellar radius before mass loss;
5. log tKH — Kelvin–Helmholtz timescale of the initial model;
6. log L — initial helium stellar luminosity before mass loss;
7. log Teff — initial helium stellar effective temperature before mass loss;
8. log g — the acceleration of gravity on the stellar surface;
9. k 2 — the dimensionless gyration radius. k 2 = I/M R2 , I is moment of inertia, M is stellar mass, R is stellar radius;
10. Ψc — the degeneracy at the center of helium star;
11. XHe — helium fraction at the center;
12. ζad — mass–radius exponent at Ṁ = ṀKH . The ’...’ marker represents the failed data;
13. qcrit — critical mass ratio. The ’...’ marker represents the failed data;
14. type — evolutionary stage of helium star;

REFERENCES
Banerjee, D. P. K., Evans, A., Woodward, C. E., et al. Doherty, C. L., Gil-Pons, P., Siess, L., & Lattanzio, J. C.
2023, ApJL, 952, L26, doi: 10.3847/2041-8213/acdf56 2017, PASA, 34, e056, doi: 10.1017/pasa.2017.52
Baschek, B., Kegel, W. H., & Traving, G. 1975, Problems Eggleton, P. P. 1971, MNRAS, 151, 351,
in stellar atmospheres and envelopes doi: 10.1093/mnras/151.3.351
Brooks, J., Kupfer, T., & Bildsten, L. 2017, ApJ, 847, 78, —. 1972, MNRAS, 156, 361, doi: 10.1093/mnras/156.3.361
doi: 10.3847/1538-4357/aa87b3 —. 1973, MNRAS, 163, 279, doi: 10.1093/mnras/163.3.279
Chen, H.-L., Tauris, T. M., Chen, X., & Han, Z. 2022, ApJ, Eggleton, P. P., & Kiseleva-Eggleton, L. 2001, ApJ, 562,
930, 134, doi: 10.3847/1538-4357/ac6608 1012, doi: 10.1086/323843
Chen, X., Liu, Z., & Han, Z. 2024, Progress in Particle and Ge, H., Hjellming, M. S., Webbink, R. F., Chen, X., & Han,
Nuclear Physics, 134, 104083, Z. 2010, ApJ, 717, 724,
doi: 10.1016/j.ppnp.2023.104083 doi: 10.1088/0004-637X/717/2/724
Claeys, J. S. W., Pols, O. R., Izzard, R. G., Vink, J., & Ge, H., Webbink, R. F., Chen, X., & Han, Z. 2015, ApJ,
Verbunt, F. W. M. 2014, A&A, 563, A83, 812, 40, doi: 10.1088/0004-637X/812/1/40
doi: 10.1051/0004-6361/201322714 —. 2020a, ApJ, 899, 132, doi: 10.3847/1538-4357/aba7b7
Copperwheat, C. M., Morales-Rueda, L., Marsh, T. R., Ge, H., Webbink, R. F., & Han, Z. 2020b, ApJS, 249, 9,
Maxted, P. F. L., & Heber, U. 2011, MNRAS, 415, 1381, doi: 10.3847/1538-4365/ab98f6
doi: 10.1111/j.1365-2966.2011.18786.x Giacconi, R. 2005, ARA&A, 43, 1,
Crawford, J. A. 1955, ApJ, 121, 71, doi: 10.1086/145965 doi: 10.1146/annurev.astro.43.090303.091253
Darwin, G. H. 1879, Proceedings of the Royal Society of Götberg, Y., Drout, M. R., Ji, A. P., et al. 2023, ApJ, 959,
London Series I, 29, 168 125, doi: 10.3847/1538-4357/ace5a3
18 Zhang et al.

Table 1. Physical parameters and mass transfer stability of a 1.6 M⊙ helium star

k age MCO log R log tKH log L log Teff log g k2 Ψc XHe ζad qcrit type
yr M⊙ R⊙ yr L⊙ K cm/s2
01 0.00e+00 0.62 -1.2282 5.3630 3.0737 1.5680 5.7092 0.076423 -2.178 0.96232 2.567 1.981
02 7.93e+05 0.65 -1.2134 5.3355 3.0948 1.5684 5.6964 0.074497 -2.183 0.85903 2.639 2.015
03 1.30e+06 0.68 -1.1914 5.3055 3.1152 1.5684 5.6772 0.072549 -2.199 0.76800 2.717 2.051
04 1.62e+06 0.70 -1.1777 5.2864 3.1284 1.5685 5.6653 0.071325 -2.209 0.71338 2.767 2.075
05 2.08e+06 0.73 -1.1586 5.2589 3.1476 1.5687 5.6488 0.069580 -2.223 0.63831 2.842 2.110
06 2.51e+06 0.75 -1.1412 5.2328 3.1661 1.5688 5.6336 0.067936 -2.235 0.57052 2.915 2.144
07 2.91e+06 0.78 -1.1252 5.2078 3.1842 1.5690 5.6197 0.066380 -2.246 0.50849 2.987 2.177
08 3.36e+06 0.80 -1.1085 5.1803 3.2045 1.5693 5.6052 0.064672 -2.257 0.44245 3.070 2.216
He-MS
09 3.89e+06 0.83 -1.0901 5.1473 3.2295 1.5698 5.5893 0.062643 -2.267 0.36666 3.173 2.264
10 4.42e+06 0.85 -1.0739 5.1140 3.2558 1.5705 5.5751 0.060584 -2.273 0.29236 3.283 2.316
11 5.53e+06 8.85 -1.0556 5.0474 3.3144 1.5727 5.5593 0.052898 -2.255 0.14475 3.530 2.431
12 6.57e+06 0.92 -1.0975 5.0059 3.3740 1.5776 5.5956 0.052897 -2.153 0.03628 3.750 2.534
13 6.83e+06 0.93 -1.1472 5.0070 3.3945 1.5809 5.6388 0.052143 -2.064 0.01369 3.798 2.556
14 6.91e+06 0.93 -1.1827 5.0114 3.4055 1.5831 5.6697 0.051834 -2.001 0.00672 3.816 2.565
15 6.96e+06 0.93 -1.2300 5.0146 3.4229 1.5861 5.7107 0.051228 -1.907 0.00221 3.857 2.584
16 6.98e+06 0.93 -1.2845 5.0007 3.4605 1.5904 5.7581 0.049233 -1.740 0.00014 4.001 2.652
17 6.98e+06 0.93 -1.3273 4.9769 3.5029 1.5945 5.7953 0.046641 -1.473 0 4.194 2.742
20 7.01e+06 0.93 -1.3027 4.8610 3.6081 1.5987 5.7739 0.037867 -0.803 0 4.878 3.062
23 7.04e+06 0.94 -0.9596 4.5189 3.8012 1.5934 5.4759 0.024070 -0.278 0 6.860 3.995
26 7.08e+06 0.96 -0.6154 4.2417 3.9289 1.5846 5.1769 0.016051 0.348 0 9.032 5.020
29 7.11e+06 0.98 -0.2796 4.0070 4.0177 1.5742 4.8852 0.011134 0.923 0 11.467 6.173
32 7.13e+06 0.99 0.0713 3.7898 4.0826 1.5617 4.5804 0.007811 1.459 0 14.336 7.535
35 7.14e+06 1.00 0.4144 3.5969 4.1264 1.5483 4.2824 0.005597 1.943 0 17.495 9.040
38 7.15e+06 1.01 0.7598 3.4161 4.1573 1.5339 3.9824 0.003997 2.385 0 21.111 10.768
41 7.15e+06 1.01 1.1045 3.2447 4.1790 1.5188 3.6831 0.002837 2.775 0 25.240 12.745
44 7.15e+06 1.02 1.4282 3.0887 4.1944 1.5041 3.4019 0.002043 3.091 0 29.720 14.897
He-HG
47 7.16e+06 1.02 1.7907 2.9182 4.2075 1.4872 3.0870 0.001401 3.415 0 35.571 17.713
50 7.16e+06 1.02 2.1209 2.7662 4.2160 1.4714 2.8002 0.000985 3.675 0 41.810 20.723
53 7.16e+06 1.02 2.4709 2.6077 4.2225 1.4541 2.4962 0.000673 3.914 0 49.569 24.475
56 7.16e+06 1.03 2.8246 2.4494 4.2272 1.4363 2.1890 0.000459 4.117 0 58.886 28.990
59 7.16e+06 1.03 3.1355 2.3115 4.2301 1.4203 1.9189 0.000337 4.269 0 68.506 33.662
62 7.16e+06 1.03 3.4013 2.1945 4.2317 1.4063 1.6880 0.000280 4.388 0 77.840 38.202
65 7.16e+06 1.03 3.8480 1.9993 4.2329 1.3824 1.3000 0.000276 4.550 0 96.621 47.356
68 7.17e+06 1.03 4.1820 1.8543 4.2328 1.3640 1.0100 0.000420 4.650 0 113.639 55.667
71 7.17e+06 1.03 4.5686 1.6878 4.2314 1.3422 0.6741 0.001966 4.849 0 134.536 65.890
72 7.17e+06 1.03 4.6919 1.6364 4.2292 1.3350 0.5670 0.005537 5.272 0 129.514 63.432
73 7.17e+06 1.03 4.8085 1.5768 4.2383 1.3290 0.4658 0.017204 6.235 0 48.879 24.141
74 7.17e+06 1.04 4.9281 1.4616 4.3015 1.3263 0.3619 0.034318 8.458 0 21.630 11.016
75 7.18e+06 1.04 5.0420 1.3476 4.3660 1.3240 0.2629 0.043957 11.863 0 21.696 11.048
76 7.18e+06 1.05 5.1558 1.2345 4.4297 1.3216 0.1641 0.050977 16.669 0 27.539 13.849
He-GB
77 7.18e+06 1.05 5.2599 1.1315 4.4875 1.3195 0.0737 0.057009 22.408 0 ... ...
78 7.19e+06 1.05 5.3876 1.0061 4.5575 1.3167 -0.0372 0.064529 31.089 0 ... ...
79 7.19e+06 1.06 5.5021 0.8939 4.6199 1.3142 -0.1366 0.071669 40.711 0 ... ...
80 7.19e+06 1.07 5.6200 0.7766 4.6860 1.3118 -0.2391 0.079331 54.250 0 135.291 66.260
81 7.20e+06 1.11 5.7341 0.6566 4.7564 1.3099 -0.3382 0.085401 74.684 0 241.300 118.316
Note—Table 1 is published in its entirety in the machine-readable format.
A portion is shown here for guidance regarding its form and content.
 He Binary Stars 19

Hamann, W. R., Gräfener, G., & Liermann, A. 2006, A&A, Podsiadlowski, P., Rappaport, S., & Pfahl, E. D. 2002,
457, 1015, doi: 10.1051/0004-6361:20065052 ApJ, 565, 1107, doi: 10.1086/324686
Han, Z., Podsiadlowski, P., & Lynas-Gray, A. E. 2007, Pols, O. R., Schröder, K.-P., Hurley, J. R., Tout, C. A., &
MNRAS, 380, 1098, Eggleton, P. P. 1998, MNRAS, 298, 525,
doi: 10.1111/j.1365-2966.2007.12151.x doi: 10.1046/j.1365-8711.1998.01658.x
Han, Z., Podsiadlowski, P., Maxted, P. F. L., & Marsh, Rappaport, S., Verbunt, F., & Joss, P. C. 1983, ApJ, 275,
T. R. 2003, MNRAS, 341, 669, 713, doi: 10.1086/161569
doi: 10.1046/j.1365-8711.2003.06451.x Rasio, F. A. 1995, ApJL, 444, L41, doi: 10.1086/187855
Han, Z., Podsiadlowski, P., Maxted, P. F. L., Marsh, T. R., Sana, H., de Mink, S. E., de Koter, A., et al. 2012, Science,
& Ivanova, N. 2002, MNRAS, 336, 449, 337, 444, doi: 10.1126/science.1223344
doi: 10.1046/j.1365-8711.2002.05752.x Sargsyan, V. V., Lenske, H., Adamian, G. G., &
Han, Z.-W., Ge, H.-W., Chen, X.-F., & Chen, H.-L. 2020, Antonenko, N. V. 2019, arXiv e-prints, arXiv:1907.01877,
Research in Astronomy and Astrophysics, 20, 161, doi: 10.48550/arXiv.1907.01877
doi: 10.1088/1674-4527/20/10/161 Schaffenroth, V., Pelisoli, I., Barlow, B. N., Geier, S., &
Heber, U. 2009, ARA&A, 47, 211, Kupfer, T. 2022, A&A, 666, A182,
doi: 10.1146/annurev-astro-082708-101836 doi: 10.1051/0004-6361/202244214
—. 2016, PASP, 128, 082001, Schroder, K.-P., Pols, O. R., & Eggleton, P. P. 1997,
doi: 10.1088/1538-3873/128/966/082001 MNRAS, 285, 696, doi: 10.1093/mnras/285.4.696
Hjellming, M. S., & Webbink, R. F. 1987, ApJ, 318, 794, Shenar, T., Bodensteiner, J., Abdul-Masih, M., et al. 2020,
doi: 10.1086/165412 A&A, 639, L6, doi: 10.1051/0004-6361/202038275
Hurley, J. R., Tout, C. A., & Pols, O. R. 2002, MNRAS, Soberman, G. E., Phinney, E. S., & van den Heuvel,
329, 897, doi: 10.1046/j.1365-8711.2002.05038.x E. P. J. 1997, A&A, 327, 620,
Ivanova, N., Justham, S., Avendano Nandez, J. L., & doi: 10.48550/arXiv.astro-ph/9703016
Lombardi, J. C. 2013, Science, 339, 433, Solheim, J. E. 2010, PASP, 122, 1133, doi: 10.1086/656680
doi: 10.1126/science.1225540 Steiner, J. E., & Oliveira, A. S. 2005, A&A, 444, 895,
Klement, R., Rivinius, T., Gies, D. R., et al. 2024, ApJ, doi: 10.1051/0004-6361:20052782
962, 70, doi: 10.3847/1538-4357/ad13ec Tauris, T. M., & Savonije, G. J. 1999, A&A, 350, 928,
Leiner, E. M., & Geller, A. 2021, ApJ, 908, 229, doi: 10.48550/arXiv.astro-ph/9909147
doi: 10.3847/1538-4357/abd7e9 Tauris, T. M., & van den Heuvel, E. P. J. 2023, Physics of
Li, Z., Chen, X., Ge, H., Chen, H.-L., & Han, Z. 2023, Binary Star Evolution. From Stars to X-ray Binaries and
A&A, 669, A82, doi: 10.1051/0004-6361/202243893 Gravitational Wave Sources,
Liu, J., Zhang, H., Howard, A. W., et al. 2019, Nature, 575, doi: 10.48550/arXiv.2305.09388
618, doi: 10.1038/s41586-019-1766-2 Temmink, K. D., Pols, O. R., Justham, S., Istrate, A. G., &
Maxted, P. F. L., Heber, U., Marsh, T. R., & North, R. C. Toonen, S. 2023, A&A, 669, A45,
2001, MNRAS, 326, 1391, doi: 10.1051/0004-6361/202244137
doi: 10.1111/j.1365-2966.2001.04714.x Tylenda, R., & Soker, N. 2006, A&A, 451, 223,
Moe, M., & Di Stefano, R. 2017, ApJS, 230, 15, doi: 10.1051/0004-6361:20054201
doi: 10.3847/1538-4365/aa6fb6 Tylenda, R., Hajduk, M., Kamiński, T., et al. 2011, A&A,
Paczyński, B. 1965, AcA, 15, 89 528, A114, doi: 10.1051/0004-6361/201016221
Paczynski, B. 1976, in Structure and Evolution of Close Villaseñor, J. I., Lennon, D. J., Picco, A., et al. 2023,
Binary Systems, ed. P. Eggleton, S. Mitton, & MNRAS, 525, 5121, doi: 10.1093/mnras/stad2533
J. Whelan, Vol. 73, 75 Vos, J., Bobrick, A., & Vučković, M. 2020, A&A, 641,
Paczyński, B., Ziólkowski, J., & Zytkow, A. 1969, in A163, doi: 10.1051/0004-6361/201937195
Astrophysics and Space Science Library, Vol. 13, Mass Vos, J., Németh, P., Vučković, M., Østensen, R., & Parsons,
Loss from Stars, ed. M. Hack, 237, S. 2018, MNRAS, 473, 693, doi: 10.1093/mnras/stx2198
doi: 10.1007/978-94-010-3405-0 30 Vos, J., Vučković, M., Chen, X., et al. 2019, MNRAS, 482,
Paxton, B. 2004, PASP, 116, 699, doi: 10.1086/422345 4592, doi: 10.1093/mnras/sty3017
Picco, A., Marchant, P., Sana, H., & Nelemans, G. 2024, Wang, B., Meng, X., Chen, X., & Han, Z. 2009, MNRAS,
A&A, 681, A31, doi: 10.1051/0004-6361/202347090 395, 847, doi: 10.1111/j.1365-2966.2009.14545.x
20 Zhang et al.

Wang, B., Meng, X., Liu, D. D., Liu, Z. W., & Han, Z. Zhou, C., Feng, H., & Bian, F. 2023, ApJ, 947, 52,
2014, ApJL, 794, L28, doi: 10.1088/2041-8205/794/2/L28 doi: 10.3847/1538-4357/acc5eb
Woosley, S. E. 2019, ApJ, 878, 49,
doi: 10.3847/1538-4357/ab1b41