Can Neutron Star Mergers Alone Explain the R-process

Enrichment of the Milky Way?

Dany VANBEVEREN and Nicki M ENNEKENS

Astronomy & Astrophysics Research Group, Vrije Universiteit Brussel

∗ Corresponding author: [email protected]
arXiv:2410.17650v1 [astro-ph.SR] 23 Oct 2024

Abstract
Mennekens and Vanbeveren (2014) studied the effect of double compact star mergers on the
Galactic chemical enrichment of r-process elements. LIGO merger detections since 2015 and
new r-process element yields as function of neutron star + neutron star (NS+NS) and neutron
star + black hole (NS+BH) mass requires an update of the 2014 computations. The results of
the update are the scope of the present paper.

1. Introduction
To answer the title-question we proceed as follows. The chemistry-data of long-lived stars
(G-dwarfs) gives us the observed Galactic evolution of Eu as function of Fe. With a detailed
Galactic chemical evolutionary code (that includes as much as possible all effects of single
star evolution as well as those of close binaries) we calculate the temporal variation of the
double white dwarfs (progenitors of Type Ia supernovae responsible for a significant fraction
of the Fe-enrichment), the double NS binary mergers and mixed system mergers (NS+NS and
NS+BH mergers) possibly responsible for a significant r-process chemical enrichment. Linking
the predicted NS+NS and NS+BH population, the merger rates caused by gravitational wave
radiation and kilo-nova simulations associated with these mergers, predicting the Galactic Eu
evolution is straightforward. Comparison with the observed run answers the title-question.
Since single star and binary population synthesis plays a crucial role a few facts are worth
repeating.

2. The massive star interacting binary fraction

In the present paper the term ‘interacting close binary’ stands for a binary where during
the evolution of the components Roche lobe overflow/mass transfer will happen. For circular
systems this means binaries with an orbital period less than (roughly) 10 years but it may be
much larger when the orbit is eccentric. We further use the abbreviation MCB (massive close
binary) and we distinguish O-type stars (ZAMS mass larger than ∼20 M⊙ ) and B0-B3- (lu-
minosity class V, IV and III)-type stars (ZAMS mass between 10 M⊙ and 20 M⊙ ). The mass

1
ratio q = mass secondary/mass primary; the secondary/primary is the binary component with
the smaller/larger mass on the ZAMS.
Garmany et al. (1980) studied the MCB fraction among all known O type stars brighter
than mv = 7 and north of -50◦ (a total of 67 O type single stars or primaries of binaries).
They concluded that 33% of the O type stars are the primary of a massive close binary (+/-
13% accounting for small number statistics) with mass ratio q larger than 0.2 and period P <
100 days. When the period distribution is flat in the Log up to 10 years (Popova et al., 1982;
Mason et al., 1998) the real interacting O type binary fraction may be considerably larger.
The MCB fraction in the B0-B3 (luminosity class V, IV and III) spectral range has been
investigated by Vanbeveren et al. (1998b,c). It was concluded that ∼32% is primary of an inter-
acting close binary. Notice the similarity with the MCB fraction of O-type stars. As for O-type
stars also the B0-B3 MCB fraction can be considerably larger.
The study of the MCB fraction becomes much more complicated if one realizes that a
significant number of the single stars in a massive star population may have had a binary history
(e.g. single stars that became single after the SN explosion of the companion, or became single
after the merger of both binary components) and the following procedure may therefore be
mandatory.
Starting from the MCB fraction at birth (on the ZAMS) one calculates the content of the
massive star population using a detailed massive star/binary population code. The question
‘what must be the massive close binary fraction at birth in order to explain the observed MCB
fraction in a population where star formation happened continuously in time’ was investigated
by Vanbeveren et al. (1998a,b,c) and it was concluded that to explain the O and B0-B3 results
discussed above, the MCB fraction at birth must be at least 70%. Note however that in 1998
the physics of mergers was poorly understood. The MCB merger rate may therefore be larger
than the 1998-value and as will be explained later on a model where the MCB fraction at birth
equals 100% cannot be excluded. Interestingly, about 14 years later the MCB fraction has been
studied over again by Sana et al. (2012) and de Mink et al. (2013) who essentially confirmed
the percentages discussed above. An interesting candidate of a single star with a binary history
is ζ Pup. In the early eighties this star was considered as a gift from the gods as far as single
star evolution is considered. However, the star is a runaway, e.g. it became runaway due to the
supernova explosion of a companion or it became runaway due to the dynamical interaction in
a dense cluster but also here at least one binary was involved in the formation. So ζ Pup may
be a gift from the gods as far as binary evolution is concerned (see also Vanbeveren, 2012).
The WR+OB binary fraction ∼30-40% (Vanbeveren and Conti, 1980) and many subse-
quent studies have confirmed this number. The O-type binaries are the WR+OB progenitors
and we can try to answer the question ‘what must be the WR+OB progenitor binary fraction in
order to explain the 30-40% WR+OB binary fraction’. This was studied by Vanbeveren et al.
(1998a,b,c) as well and it was concluded that a MCB fraction at birth >70% also explains the
observed WR+OB binary fraction.
Last not least, 15% of the red supergiants seems to have a binary component (Neugent et al.,

2
2020) and also here this number is compatible with a >70% MCB binary fraction at birth.
The population of early B-type supergiants deserves special attention. Figure 1 illustrates
the HR-diagram of about 2500 massive stars in stellar aggregates in the Solar Neighborhood
(data taken from Humphreys and McElroy, 1984). When compared to massive single star evo-
lution we conclude that the supergiant stars in the red box are post core hydrogen burning,
hydrogen shell burning stars. These stars are expected to cross the box towards the red on the
thermal Kelvin-Helmholts timescale which is very short (a factor 100-1000 shorter than the
corresponding core hydrogen burning timescale). Therefore, compared to the number of stars
in the core hydrogen burning box, the number of stars in the red box of Figure 1 should be
100-1000 smaller than in the core hydrogen burning box and by inspecting Figure 1 it is clear
that this is far from being the case. Humphreys and McElroy also presented the results for
the Magellanic Clouds and the effect discussed above is also visible there. Furthermore, only
few stars in the red box seem to be close binaries (probably less than 20%) which is odd at first
glance considering the fact that the MCB fraction at birth of the progenitors of these supergiants
is >70%.
A famous example of the stars in the red box is the blue supergiant B-type progenitor
of SN1987A, a star in the Large Magellanic Cloud with a mass ∼20 M⊙ . Binary models
for SN1987A have been presented by Podsiadlowski et al. (1990); Podsiadlowski (1992) and
by De Loore and Vanbeveren (1992) following original suggestions of Hellings (1983, 1984).
These studies learned us that a post-core hydrogen burning star with a mass ∼20 M⊙ with an
under-massive helium core and an over-massive hydrogen shell will remain in the blue part of
the HR-diagram during most of its core helium burning phase, populating in this way the red
box of Figure 1. So, question: how to make a (massive) post-core hydrogen burning star with
an under-massive helium core and an over-massive hydrogen shell? Vanbeveren et al. (2013)
and Justham et al. (2014) proposed the merger of a massive case B close binary before the onset
of core helium burning of the primary (case Br). In Figure 2 we show the evolutionary track
of 3 binary models (we refer to our 2013 paper for a discussion of the evolutionary results and
details about the computational method) where soon after the onset of RLOF both stars enter
a contact phase from whereon the secondary merges with the primary. Merging is treated as a
fast accretion process. At the end of the merging process the merger rapidly restores thermal
equilibrium and becomes a core helium burning star but with a helium core that is under-massive
with respect to its total mass. The merger remains a blue star till the end of core helium burning,
e.g. these mergers populate the red box of Figure 1. Note that the merger is N overabundant, He
overabundant, CO underabundant. Interestingly, Menon et al. (2024) obtained chemistry data
for 59 blue supergiants in the Large Magellanic Cloud and compared them with the theoretically
predicted chemistry of mergers. The authors conclude that for at least half of these supergiants
the observations correspond to the merger predictions, a result that strongly support the merger
model for these stars. Note that the merger model naturally explains why the blue supergiant
binary fraction is so low.
The situation now is as follows: a significant percentage of case Br binaries merge and
become single blue supergiants. A significant percentage of the MCBs evolve via a quasi-

3
 Figure 1: The HR-diagram of 2500 massive stars in stellar aggregates in the Solar
Neighborhood taken from Humphreys and McElroy (1984). The single star evolu-
tionary tracks are Geneva tracks (Schaller et al., 1992). The red box contains the
early B type supergiants discussed in the text.

conservative RLOF and become WR+OB binaries. What makes the difference? And how
important is the blue supergiant merger model when we repeat the exercise where we link the
MCB fraction and the MCB fraction at birth as discussed above? We are prepared to stick
out our neck: accounting for the observed large number of blue supergiants and the highly
plausible merger model for these supergiants (which makes them single stars), repeating the
exercise discussed above let us conclude that the MCB fraction at birth is close to 100%. The
simulations discussed in section 4 have been calculated with a MCB fraction at birth = 70%.
With 100% the results change of course but the overall conclusions remain the same.

3. The effect of LBV mass loss on the evolution of massive single stars and
binaries
Luminous Blue Variables (LBVs) have a luminosity Log L/L⊙ > 5.4 and they occupy a
region in the HR-diagram that corresponds to the hot part of the hydrogen shell burning phase
of stars with an initial mass higher than 20 M⊙ . The observed LBV-mass loss rates range be-
tween 10−5 and 10−3 M⊙ /yr and on top of that LBVs are known to experience major mass
eruption phases by a process that is as yet unclear. The observations reveal that the region in the
HR-diagram which is located to the right of the region occupied by the LBVs with a luminosity
Log L/L⊙ > 5.5 (corresponding to stars with an initial mass higher than 30-40 M⊙ ) is almost

4
 Figure 2: Evolutionary tracks of 3 case B binary models (30+20M⊙ , 15+9M⊙,
9+6.7M⊙) where merging of the two stars starts at point A (merging is treated as a
fast accretion process, where the secondary star is accreted onto the primary).

void. There are a few yellow hypergiants but no red supergiants. This is also observed in the
Magellanic Clouds. Based on this fact and on the observed mass loss rates it was suggested by
Humphreys and Davidson (1979, 1984) that the LBV-mass loss has to be so high that it prohibits
a hydrogen shell burning star with initial mass >30-40 M⊙ from evolving into a red supergiant,
and that this scenario is metallicity independent. From evolutionary point of view this is a very
robust criterion allowing to calculate the LBV-mass loss rate that is needed. This was inves-
tigated by the Geneva stellar evolution team (Schaller et al. (1992), see also Vanbeveren et al.
(1998b,c)) and it was concluded that when a massive star becomes an LBV and loses mass at
a few times 10−3 M⊙ /yr up to 10−2 M⊙/yr stars with an initial mass >30-40 M⊙ lose their
hydrogen rich layers and they become Wolf-Rayet stars without becoming RSGs first.
As discussed above it is conceivable that LBV stellar wind mass loss suppresses the red-
ward evolution during hydrogen shell burning of stars with an initial mass higher than 30-40
M⊙ and that stars in this mass range lose most of their hydrogen rich layers by this LBV mass
loss process. As a consequence it cannot be excluded that the LBV mass loss rate suppresses
the RLOF/common envelope phase in case Br/case Bc/case C binaries when the mass loser has
a mass higher than 30-40 M⊙ (the LBV scenario of massive binaries as it was introduced by
Vanbeveren, 1991). In these binaries the orbital period variation satisfies

 2
P M1,0 + M2,0
= (1)
P0 M1 + M2

where the subscript ‘0’ stands for values at the beginning of the LBV mass loss phase.
Eq. (1) predicts a significant binary period increase during the LBV mass loss phase and

5
we demonstrated in Mennekens and Vanbeveren (2014) that this LBV scenario affects critically
the predicted population of double-BH-binary mergers.
The results of the present paper (Table 1) with the label “LBV 40” (resp. “LBV 20”) are
calculated assuming an average LBV wind (the average of the eruption + intereruption mass
loss) of a few 10−3 M⊙ /yr up to 10−2 M⊙ /yr that prohibits the redward evolution of stars with
initial mass > 40 M⊙ (resp. 20 M⊙). Notice that the results presented here would be entirely
similar when other average values would be adopted provided that these values are high enough
so that they prevent the occurrence of the RLOF. The label “LBV off” correspond to results
computed by switching off the LBV eruption mass loss so that the redward evolution for stars
with initial mass > 20 M⊙ is not suppressed and, when such a star has a close companion,
RLOF/common envelope evolution will happen.
As a note added in proof, Cheng et al. (2024) proposed a parameter model for eruptive
mass loss of LBVs and implemented this in the MESA stellar evolutionary code. Their results
essentially confirm the Geneva and Brussels simulations discussed above.

4. Compact object mergers and r-process elements

In Mennekens and Vanbeveren (2014, 2016) we used a combination of the Brussels pop-
ulation synthesis code and Galactic chemical evolution model (for an extended review see
De Donder and Vanbeveren, 2004) to study the effect of binary star evolution on the produc-
tion of r-process elements. These elements can be produced and released during the merger
of two NSs or one NS and one BH, the latter being termed a ‘mixed’ merger. The research
question is whether these compact object mergers alone can produce enough r-process elements
to match observations. The amount of r-process elements released due to a particular merger
depends on the masses of both compact objects, and were previously based on the calculations
by Korobkin et al. (2012). These NSs and BHs are obviously the end products of (binary) star
evolution, and their masses and distance (determining their merger timescale due to the emis-
sion of gravitational wave radiation) are determined by their previous evolution. In particular,
we investigated the influence of still uncertain processes during the life of a binary star on the
amount of r-process elements produced and the time at which they are released. The most im-
portant of these processes are the mass transfer efficiency during stable Roche-lobe overflow
(RLOF) and the energy conversion efficiency during unstable common envelope (CE) evolu-
tion. As is common in binary evolution studies, we use the parameters β and α to denote these
properties. For RLOF, β is the fraction of matter lost by the donor star that is actually accepted
by the gainer star. β = 1 thus denotes conservative RLOF, while β < 1 implies non-conservative
mass transfer. For CE evolution, α is defined as the efficiency of the transfer of orbital energy
into escape energy of the CE. As can be seen from the results in Mennekens and Vanbeveren
(2014, 2016) these parameters (among others) have an enormous influence on the amount of
r-process elements produced and the time at which they are released. Hence, these papers in-
clude a number of models which combine different assumptions and parameters. Models which
produce results that are incompatible with observations can then be excluded. A number of
observations can be used. First of all, the number of predicted Galactic double NS mergers

6
must be high enough to match observations. For this, we use the lower limit of 3 mergers per
Myr found by Kim et al. (2010). Any models producing less mergers are eliminated. Second,
models which produce too much r-process elements to be compatible with observations can be
eliminated. Those which produce too few r-process elements (all of them, as will be seen later)
can obviously not be eliminated, as there may be other sources of r-process element enrichment.
Third, one can also look at the aLIGO rates for NS+NS and NS+BH mergers predicted by the
models. This will be discussed later, as no observational aLIGO rates were available in 2014.
The conclusion in 2014 was that some models could be eliminated on the grounds above, but
none of the models succeeded in producing sufficient r-process elements through NS+NS and
NS+BH mergers alone. A possible solution is of course to include r-process element production
though other means, such as single star type II supernovae at low metallicity, which then gave
a number of reasonable models. It is notable that for most or all of these models, the mixed
mergers (NS+BH) provided a notable contribution to the r-process element enrichment of the
same order of magnitude as the contribution of double NS mergers. All these conclusions are
graphically illustrated in Fig 7 of Mennekens and Vanbeveren (2014).
Fig 2(b) of Kobayashi et al. (2023) presented a comparative study of various attempts at
the theoretical reproduction of the observed [Eu/Fe] (representative for r-process elements) vs.
[Fe/H] rates. The conclusion is that none of these models produce enough r-process elements
to match observations. This comparison is based on the yields of a single NS+NS merger
calculation, and assumes that the same yields are produced in NS+BH mergers (whereas, at
least in 2014 with the Korobkin yields, we found those to be notably larger). It should be
noted that the Brussels model used in this comparative study (named model 2 in our 2014
paper) is not our model with the highest allowable r-process element production. That is
to say, there were models in our 2014 study with higher, but (assuming the yields used by
Kobayashi et al., 2023) not too high, r-process element production. Also, the star formation
rate assumed by Kobayashi et al. (2023) is quite different from our own, which means that their
Fig 2(b) is more “smeared out” in the horizontal (temporal) direction than the panels in Fig 7 of
Mennekens and Vanbeveren (2014).
Recently, the Brussels code was updated with the results from Just et al. (2015) to deter-
mine the amount of r-process elements released during a particular merger. These yields are
based on a larger and more diverse number of merger calculations. A very large difference with
the yields used before, is that with the Just yields, mixed mergers (NS+BH) hardly contribute to
the r-process element enrichment, especially at higher metallicity. Hence, the [r/Fe] vs. [Fe/H]
now almost only depends on the amount of double NS mergers. With the updated code, we
have calculated a number of models consisting of various combinations of parameters. Table 1
summarizes these models. Some of these (the first five lines in Table 1) are the same models as
in 2014, while others consist of new variations. The left hand side of the table denotes the input
parameters of these various models.
With the same considerations as in 2014, models that produce insufficient Galactic NS+NS
mergers and models that (at any time during the evolution) produce too much r-process elements
can be rejected. Notable models that can thus be eliminated are those with β = 0.5, as they pro-

7
duce too few Galactic mergers. Stable mass transfer thus has to be (near) conservative in order
to match observations. Among the models with β = 1, those with α = 1 (i.e. high common
envelope efficiency) produce too much r-process elements at certain times and can thus also be
eliminated. While models with lower CE efficiency (α = 0.5) can not strictly be excluded (as
there may be other r-process element sources), the best match with observations is found for a
moderate CE efficiency of α = 2/3. The [r/Fe] vs. [Fe/H] evolution predicted by this model is
shown in Fig 3 of the present paper. The solid line represents the enrichment when including
only NS+NS mergers. The dashed line includes the enrichment by both NS+NS and NS+BH
mergers. This figure can thus be directly compared to the black lines in the panels of Fig 7 in
Mennekens and Vanbeveren (2014) which also provide this distinction. The dots are observed
values as discussed in the same paper. In this case, there is an excellent match between predic-
tion and observation at higher metallicity (i.e. the right half of the figure). For low metallicity
(corresponding to the first ∼100 Myr of Galactic evolution) however, there is still insufficient
r-process element production, and another source is needed to match observations. As in 2014,
a suggestion (although certainly not the only one) could be the contribution from single star
type II supernovae at low metallicity. It is obvious that due to the Just yields, the difference
between the solid and dashed line in the figure is much smaller than between the corresponding
lines in any of the panels in Fig 7 of Mennekens and Vanbeveren (2014). This demonstrates that
with the Just yields, NS+BH mergers barely contribute to the r-process element enrichment, and
only at very low metallicity. We now revisit the comparison by Kobayashi et al. (2023). With
our updated code using the Just yields, our [r/Fe] vs. [Fe/H] for β = 1 and α = 0.5 (model 2
from 2014, second line in Table 1) now indeed looks very much like the one shown in Fig 2(b)
of Kobayashi et al. (2023). However others, such as the one shown in Fig 3 of the present paper,
would lie much higher.
Thanks to the availability of aLIGO data, a third comparison can now be made which was
not possible in 2014. Combined data from the first three aLIGO runs (Abbott et al., 2023) seems
to suggest that there is a near parity between the number of observed NS+NS and NS+BH merg-
ers. To be exact, the Collaboration’s analysis suggests that aLIGO has detected two NS+NS
mergers and four NS+BH mergers (as well as 69 BH+BH mergers). Given the small numbers,
one must obviously allow for a large statistical error. This ratio is obviously also a quantity that
is predicted by our various theoretical models. The ‘best’ model mentioned above, with β = 1
and α = 2/3, which (apart from early times) satisfactorily reproduces the [r/Fe] vs. [Fe/H]
rates, predicts 24 times more NS+BH merger detections than NS+NS merger detections. Re-
gardless of small number statistics concerning the observations, this number appears to be too
high. There are a number of possible solutions for this mismatch. First, one could assume that
the limit for direct BH formation may be higher than our standard value of 40 M⊙. A number
of models has been calculated where this limit has been brought to 70 M⊙ (Schneider et al.,
2021). This way, the number of NS+BH mergers is greatly reduced, without much influencing
the (already satisfactory) [r/Fe] vs. [Fe/H] evolution as the mixed mergers hardly contribute to
the r-process element enrichment. A second possible solution (on its own or in combination
with the first) is to assume that the supernova kick imparted on a BH may be larger than in
our standard model. We normally use an expression (described in Mennekens and Vanbeveren,

8
2014) to calculate the fallback and resulting kick velocity during compact object formation,
which results in almost 100% fallback (and thus a very small kick) in the case of a BH. Now
we have calculated a number of models where we assume zero fallback (and thus a large kick)
in the case of BH formation. Lastly, one may also adapt the luminous blue variable (LBV) sce-
nario. In our standard models, it is assumed that stars with an initial mass above 40 M⊙ evolve
according to this scenario. We now also present some calculations where this limit is lowered
to 20 M⊙ , as well as some others where this LBV scenario is avoided altogether even by mas-
sive stars (denoted “off” in Table 1, a possibility also discussed in Mennekens and Vanbeveren,
2014). As can be seen in the right half of the table, most of these models result in a [r/Fe]
vs. [Fe/H] evolution that is acceptable (and thus all very similar to the one shown in Fig 3).
That is to say, not too high at any time, and not too low at high metallicity. The [r/Fe] lying
too low at low metallicity is present in each model, and thus as mentioned before requires an
alternative r-process element source. Table 1 also shows the ratio of expected NS+BH/NS+NS
aLIGO detections, which as a reminder is expected to be about two, but may statistically lie
anywhere between 0.33 and 15. Also the predicted amount of Galactic NS+NS mergers (with
its observational lower limit of three per Myr) is shown. It should also be noted (as shown in
Table 1) that most of our models predict no aLIGO BH+BH merger detections from a binary
evolutionary origin. The only exception are the models with LBV off, but most of those then
immediately have a BH+BH merger detection rate that is much higher (compared to NS+NS
and NS+BH merger detections) than observations suggest. Therefore we certainly keep an open
mind concerning other pathways to BH+BH mergers that have been suggested in the literature
over the past decade.
The conclusion remains that none of our models is able to reproduce the observed amount
of r-process element enrichment at early stages in Galactic evolution (at low metallicity) through
compact object mergers alone, and thus certainly not when including only double NS mergers.
Other sources of r-process elements, such as single stars, must be included to obtain an accept-
able result. The [r/Fe] vs. [Fe/H] at later times (higher metallicity) can be reproduced more
satisfactorily with the Just yields than with the Korobkin yields used by us in 2014. How-
ever, they remain extremely sensitive to binary star evolutionary parameters such as β , α and
assumptions about BH formation and LBV evolution.

References
Abbott, R., Abbott, T., Acernese, F. and the LIGO, Virgo and KAGRA Collaboration (2023)
Physical Review X, 13, 011048. https://doi.org/10.1103/PhysRevX.13.011048.

Cheng, S., Goldberg, J., Cantiello, M., Bauer, E., Renzo, M. and Conroy, C. (2024) The Astro-
physical Journal, accepted (arXiv:2405.12274). https://doi.org/10.48550/arXiv.2405.12274.

De Donder, E. and Vanbeveren, D. (2004) New Astronomy Reviews, 48, 861.

https://doi.org/10.1016/j.newar.2004.07.001.

De Loore, C. and Vanbeveren, D. (1992) Astronomy & Astrophysics, 260, 273.

9
Table 1: Overview of models computed in this study. Left part contains
input parameters, right part contains output predictions. ‘std’ denotes
the standard fallback mechanism (see text).
BH BH LBV [r/Fe] aLIGO aLIGO Gal NS+
NS+BH BH+BH
lim fall- lim vs. NS+NS NS+NS NS merg
β α (M⊙ ) back (M⊙ ) [Fe/H] ratio ratio (/Myr)
1 1 40 std 40 high 17.7 0 42.1
1 0.5 40 std 40 low 29.1 0 8.65
0.5 1 40 std 40 low 350 0 2.05
1 1 40 std off high 9.48 315 42.1
1 0.5 40 std off low 67.5 1880 8.65
1 2/3 40 std 40 ok 24.3 0 19.0
1 2/3 70 std 40 ok 11.2 0 19.0
1 2/3 40 0 40 ok 11.1 0 19.0
1 2/3 70 0 40 ok 3.73 0 19.0
1 2/3 40 std off ok 23.3 820 19.0
1 2/3 70 std off ok 10.5 27.3 19.0
1 2/3 40 0 off ok 9.72 158 19.0
1 2/3 70 0 off ok 3.64 3.71 19.0
1 2/3 40 std 20 ok 8.12 0 19.4
1 2/3 70 std 20 ok 3.41 0 19.5
1 2/3 40 0 20 ok 3.42 0 19.4
1 2/3 70 0 20 ok 0.96 0 19.5

10
 Figure 3: r-process element evolution for a model with β = 1, α = 2/3 and other
assumptions standard (= sixth line in Table 1). Solid line is enrichment due to
NS+NS only, dashed line due to NS+NS and NS+BH combined. Dots are observa-
tions.

de Mink, S., Langer, N., Izzard, R., Sana, H. and de Koter, A. (2013) The Astrophysical Journal,
764, 166. https://doi.org/10.1088/0004-637X/764/2/166.

Garmany, C., Conti, P. and Massey, P. (1980) The Astrophysical Journal, 242, 1063.
https://doi.org/10.1086/158537.

Hellings, P. (1983) Astrophysics & Space Science, 96, 37. https://doi.org/10.1007/BF00661941.

Hellings, P. (1984) Astrophysics & Space Science, 104, 83.

https://doi.org/10.1007/BF00653994.

Humphreys, R. and Davidson, K. (1979) The Astrophysical Journal, 232, 409.

https://doi.org/10.1086/157301.

Humphreys, R. and Davidson, K. (1984) Science, 223, 243.

https://doi.org/10.1126/science.223.4633.243.

Humphreys, R. and McElroy, D. (1984) The Astrophysical Journal, 284, 565.

https://doi.org/10.1086/162439.

Just, O., Bauswein, A., Ardevol Pulpillo, R., Goriely, S. and Janka, H. (2015) Monthly Notices
of the Royal Astronomical Society, 448, 541. https://doi.org/10.1093/mnras/stv009.

Justham, S., Podsiadlowski, P. and Vink, J. (2014) The Astrophysical Journal, 796, 121.
https://doi.org/10.1088/0004-637X/796/2/121.

11
Kim, C., Kalogera, V. and Lorimer, D. (2010) New Astronomy Reviews, 54, 148.
https://doi.org/10.1016/j.newar.2010.09.010.

Kobayashi, C., Mandel, I., Belczynski, K., Goriely, S., Janka, T., Just, O., Ruiter, A., Vanbev-
eren, D., Kruckow, M., Briel, M., Eldridge, J. and Stanway, E. (2023) The Astrophysical
Journal Letters, 943, L12. https://doi.org/10.3847/2041-8213/acad82.

Korobkin, O., Rosswog, S., Arcones, A. and Winteler, C. (2012) Monthly Notices of the Royal
Astronomical Society, 426, 1940. https://doi.org/10.1111/j.1365-2966.2012.21859.x.

Mason, B., Gies, D., Hartkopf, W., Bagnuolo, W., ten Brummelaar, T. and McAlister, H. (1998)
The Astronomical Journal, 115, 821. https://doi.org/10.1086/300234.

Mennekens, N. and Vanbeveren, D. (2014) Astronomy & Astrophysics, 564, A134.

https://doi.org/10.1051/0004-6361/201322198.

Mennekens, N. and Vanbeveren, D. (2016) Astronomy & Astrophysics, 589, A64.

https://doi.org/10.1051/0004-6361/201628193.

Menon, A., Ercolino, A., Urbaneja, M., Lennon, D., Herrero, A., Hirai, R., Langer, N., Schoote-
meijer, A., Chatzopoulos, E., Frank, J. and Shiber, S. (2024) The Astrophysical Journal Let-
ters, 963, 42. https://doi.org/10.3847/2041-8213/ad2074.

Neugent, K., Levesque, E., Massey, P., Morrell, N. and Drout, M. (2020) The Astrophysical
Journal, 900, 118. https://doi.org/10.3847/1538-4357/ababaa.

Podsiadlowski, P. (1992) Publications of the Astronomical Society of the Pacific, 104, 717.
https://doi.org/10.1086/133043.

Podsiadlowski, P., Joss, P. and Rappaport, S. (1990) Astronomy & Astrophysics, 227, 9.

Popova, E., Tutukov, A. and Yungelson, L. (1982) Astrophysics & Space Science, 88, 55.
https://doi.org/10.1007/BF00648989.

Sana, H., de Mink, S., de Koter, A., Langer, N., Evans, C., Gieles, M., Gos-
set, E., Izzard, R., Le Bouquin, J.-B. and Schneider, F. (2012) Science, 337, 444.
https://doi.org/10.1126/science.1223344.

Schaller, G., Schaerer, D., Meynet, G. and Maeder, A. (1992) Astronomy & Astrophysics Sup-
plement, 96, 269.

Schneider, F., Podsiadlowski, P. and Muller, B. (2021) Astronomy & Astrophysics, 645, A5.
https://doi.org/10.1051/0004-6361/202039219.

Vanbeveren, D. (1991) Astronomy & Astrophysics, 252, 159.

Vanbeveren, D. (2012) Astronomical Society of the Pacific Conference Series, 465, 342.
https://doi.org/10.48550/arXiv.1109.6497.

12
Vanbeveren, D. and Conti, P. (1980) Astronomy & Astrophysics, 88, 230.

Vanbeveren, D., De Donder, E., Van Bever, J., Van Rensbergen, W. and De Loore, C. (1998a)
New Astronomy, 3, 443. https://doi.org/10.1016/S1384-1076(98)00020-7.

Vanbeveren, D., Mennekens, N., Van Rensbergen, W. and De Loore, C. (2013) Astronomy &
Astrophysics, 552, A105. https://doi.org/10.1051/0004-6361/201321072.

Vanbeveren, D., Van Rensbergen, W. and De Loore, C. (1998b) Astronomy & Astrophysics
Review, 9, 63. https://doi.org/10.1007/s001590050015.

Vanbeveren, D., Van Rensbergen, W. and De Loore, C. (1998c) The Brightest Binaries, Kluwer
Academic Publishers, Dordrecht.

13