Astronomy & Astrophysics manuscript no.

0164pap February 2, 2008

(DOI: will be inserted by hand later)

Statistical properties of exoplanets

III. Planet properties and stellar multiplicity
A. Eggenberger, S. Udry, and M. Mayor

Observatoire de Genève, 51 ch. des Maillettes, CH-1290 Sauverny, Switzerland

arXiv:astro-ph/0402664v1 27 Feb 2004

Received / Accepted

Abstract. Among the hundred or so extrasolar planets discovered to date, 19 are orbiting a component of a
double or multiple star system. In this paper, we discuss the properties of these planets and compare them to the
characteristics of planets orbiting isolated stars. Although the sample of planets found in multiple star systems is
not large, some differences between the orbital parameters and the masses of these planets and the ones of planets
orbiting single stars are emerging in the mass–period and in the eccentricity–period diagrams. As pointed out by
Zucker & Mazeh (2002), the most massive short-period planets are all found in multiple star systems. We show
here that the planets orbiting in multiple star systems also tend to have a very low eccentricity when their period
is shorter than about 40 days. These observations seem to indicate that some kind of migration has been at work
in the history of these systems. The properties of the five short-period planets orbiting in multiple star systems
seem, however, difficult to explain with the current models of planet formation and evolution, at least if we want
to invoke a single mechanism to account for all the characteristics of these planets.

Key words. planetary systems – binaries: general

1. Introduction discussed the period distribution and the mass–period

diagram for extrasolar planets orbiting single stars. As
Studies of stellar multiplicity among solar-type stars of the pointed out by Zucker & Mazeh (2002), planets orbiting
solar neighbourhood have shown that about 40% of the G a component of a multiple star system seem to have dif-
and K dwarfs can be considered to be real single stars ferent characteristics than planets orbiting single stars.
(Duquennoy & Mayor 1991; Eggenberger et al. 2003). As Zucker & Mazeh (2002) showed that there is a significant
the majority of solar-type stars belong to double or multi- correlation in the mass–period diagram for planets orbit-
ple star systems1 , it is of interest to consider the existence ing single stars, while there may be an anticorrelation in
of planets in such an environment. Searches for extraso- this same diagram for planets found in multiple star sys-
lar planets using the radial velocity technique have shown tems. The difference is mainly due to a paucity of massive
that giant planets exist in certain types of multiple star planets with short periods, and to the fact that the most
systems (see Table 1 for further details). The number of massive short-period planets are all found in binaries.
such planets is, however, still low, in part because close bi-
The characteristics of extrasolar giant planets have
naries are difficult targets for radial velocity surveys and
forced considerable modifications of the standard model of
were consequently often rejected from the samples. Due to
planet formation. It is now usually believed that planets
the limitations of the available observational techniques,
form within a protoplanetary disc of gas and dust orbiting
most detected objects are giant (Jupiter-like) planets; the
a central star, but the precise modes by which this forma-
existence of smaller mass planets in multiple star systems
tion takes place are still debated, especially for giant plan-
is thus still an open question.
ets (e.g. Pollack et al. 1996; Boss 1997; Bodenheimer et al.
The orbital characteristics and the mass distribution
2000; Boss 2000; Wuchterl et al. 2000; Boss 2003). Two
of extrasolar planets can give us an insight into their for-
major models have been proposed to explain giant planet
mation mechanisms and their subsequent evolution. In
formation (see Sect. 4.3), each with its advantages and lim-
the first paper of this series, Udry et al. (2003b; Paper I)
itations, but there is currently no consistent model that
Send offprint requests to: Anne Eggenberger, e-mail: accounts for all the observed characteristics of extrasolar
Anne.Eggenberger@obs.unige.ch planets. Observational constraints are thus needed, not
1
In this paper, double and multiple star systems will be only to specify our understanding of planet formation and
called multiple star systems. evolution, but also to possibly discriminate between the
2 A. Eggenberger et al.: Statistical properties of exoplanets III

Table 1. Planets orbiting a component of a multiple star system with confirmed orbital or common proper motion
(CPM stands for common proper motion and SB for spectroscopic binary).

Star ab ap Mp sin i ep Notes References

(AU) (AU) (MJ )
HD 40979 ∼6400 0.811 3.32 0.23 CPMa 12,11
Gl 777 A ∼3000 4.8 1.33 0.48 CPM 1,23
HD 80606 ∼1200 0.469 3.90 0.927 CPM 22
55 Cnc ∼1065 0.115 0.84 0.02 CPM 8,21,19,2
0.24 0.21 0.34
5.9 4.05 0.16
16 Cyg B ∼850 1.6 1.5 0.634 CPM 24,5,15
Ups And ∼750 0.059 0.71 0.034 CPM 17,24,2,3
0.83 2.11 0.18
2.50 4.61 0.44
HD 178911 B ∼640 0.32 6.292 0.1243 CPM 28,30
HD 219542 B ∼288 0.46 0.30 0.32 CPM 7
Tau Boo ∼240 0.05 4.08 0.018 orbit 13,24,2
HD 195019 ∼150 0.14 3.51 0.03 CPM 24,1,10
HD 114762 ∼130 0.35 11.03 0.34 CPM 24,16,18
HD 19994 ∼100 1.54 1.78 0.33 orbit 13,25,20
HD 41004 A ∼23 1.33 2.5 0.39 SB 29,31,27
γ Cep ∼22 2.03 1.59 0.2 SB 4,6,14
Gl 86 ∼20 0.11 4.0 0.046 CPM, SBb 9,26

Notes: (a) According to Halbwachs (1986), this pair has only a probability of 60% to be physical. The physical nature of
this binary has however been confirmed later on the basis of CORAVEL radial velocity measurements (Halbwachs, private
communication); (b) The multiplicity status of this system has still to be clarified.
References: (1) Allen et al. 2000; (2) Butler et al. 1997; (3) Butler et al. 1999; (4) Campbell et al. 1988; (5) Cochran et al. 1997;
(6) Cochran et al. 2002; (7) Desidera et al. 2003; (8) Duquennoy & Mayor 1991; (9) Els et al. 2001; (10) Fischer et al. 1999; (11)
Fischer et al. 2003; (12) Halbwachs 1986; (13) Hale 1994; (14) Hatzes et al. 2003; (15) Hauser & Marcy 1999; (16) Latham et al.
1989; (17) Lowrance et al. 2002; (18) Marcy et al. 1999; (19) Marcy et al. 2002; (20) Mayor et al. 2003; (21) McGrath et al.
2002; (22) Naef et al. 2001; (23) Naef et al. 2003; (24) Patience et al. 2002; (25) Queloz et al. 2000a; (26) Queloz et al. 2000b;
(27) Santos et al. 2002; (28) Tokovinin et al. 2000; (29) Udry et al. 2003a; (30) Zucker et al. 2002; (31) Zucker et al. 2003.

proposed models. In this context, the detection and the 3. Statistics of planets in multiple star systems
characterization of planets orbiting in multiple star sys-
tems, even if more difficult to carry out than the study of Although the sample of planets found in multiple star sys-
planets orbiting isolated stars, may bring new constraints tems is not large, a preliminary comparison between the
and additional information. characteristics of these planets and the ones of planets or-
biting isolated stars can be made. Here, we will discuss the
This paper is organized as follows. The sample of plan-
mass–period and the eccentricity–period diagrams for ex-
ets found in multiple star systems is presented in Sect. 2.
trasolar planets, focusing on possible differences between
Some trends seen in the statistics are then emphasized in
the two populations. Our sample of planetary candidates
Sect. 3. Models of formation and evolution of giant planets
orbiting a component of a multiple star system consists of
in binaries are briefly reviewed in Sect. 4 and their pre-
all the systems listed in Table 1. The total sample of ex-
dictions are compared to the observations in Sect. 5. Our
trasolar planetary candidates is made of 115 objects2 with
conclusions are drawn in Sect. 6.
a minimum mass M2 sin i ≤ 18 MJ . The orbital elements
used for the analysis are the ones deduced from our radial
velocity data or the most recent version of the ones given
2. Known planets in multiple star systems
in the literature.
Among the extrasolar planets discovered to date, some of
them are orbiting a component of a multiple star system.
3.1. The mass–period diagram
Planets have been found around stars known to be part
of a wide common proper motion pair, known to be in a Figure 1 shows the distribution of all the extrasolar plan-
visual binary or in a spectroscopic binary. Alternatively, etary candidates in the M2 sin i–log P plane. Two in-
searches for faint companions to stars hosting planets have teresting features emerge from this plot: there are no
revealed a few new systems. These observations, summa- short-period extrasolar planets with a mass M2 sin i &
rized in Table 1, show that giant planets can form and
2
survive in certain types of multiple star systems. see e.g. http://obswww.unige.ch/Exoplanets/
 A. Eggenberger et al.: Statistical properties of exoplanets III 3

Fig. 1. Planetary minimum mass versus orbital period for Fig. 2. Eccentricity versus orbital period for all the ex-
the known extrasolar planetary candidates. Planets orbit- trasolar planetary candidates. Same symbol coding as in
ing a single star are represented by open circles, while Fig. 1. The dashed line approximately delimits the zone
planets orbiting a component of a multiple star system where there are no planets belonging to multiple star sys-
are represented by filled circles. The dashed line approx- tems.
imately delimits the zone where only extrasolar planets
belonging to multiple star systems are found.
systems is, nevertheless, smaller than the mean mass of
planets orbiting single stars. This difference comes from
5 MJ , and the most massive short-period planets are al- the fact that no very massive planet has been found on a
most all found in multiple star systems (Udry et al. 2002; long-period orbit around a component of a multiple star
Zucker & Mazeh 2002; Paper I). Indeed, planetary candi- system. Again, this cannot be attributed to observational
dates with a mass M2 sin i & 2 MJ and a period P . 40 selection effects, because several planets with a smaller
days are all orbiting a component of a multiple star sys- mass and on long-period orbits have been found in multi-
tem, the only exception being HD 162020 b. As explained ple star systems.
in Udry et al. (2002), HD 162020 b is probably a brown The lack of planets with M2 sin i ≥ 5 MJ and P ≥ 100
dwarf with a true mass much larger than its minimum days in multiple star systems can, however, be due to the
mass, and should therefore be removed from our diagram. small number of planets found in multiple star systems. To
If it is true that no planet with a mass M2 sin i & 2 MJ check this point, we computed the probability that, draw-
orbiting a single star has been found with a period P . 40 ing a random subsample of 10 planets (i.e. the number of
days, the two populations of planets are somewhat mixed long-period planets found in multiple star systems when
together for periods between 40 and 150 days. The orbital selection effects are taken into account) with P ≥ 100
period below which there is no more massive planet or- days out of the total population of planets with P ≥ 100
biting a single star is thus not well defined, and such a days, we would have no planet with M2 sin i ≥ 5 MJ (see
unique and well defined limit may, in fact, not exist. Appendix A.1 for further details). Using the hypergeomet-
The paucity of massive short-period planets cannot ric distribution (Appendix A), this probability is 3.25 %
be attributed to observational selection effects since these (selection effects have been taken into account by discard-
planets are the easiest to detect. Moreover, even if the ing from the counts the planets with a radial velocity
sample of planets orbiting a component of a multiple star semiamplitude K < 15 ms−1 ). Thus, if the planets with
system is small and incomplete, the presence of a few can- P ≥ 100 days orbiting in multiple star systems have the
didates in a zone of the diagram where there are no other same properties as the planets with P ≥ 100 days orbit-
planets is significant. We will come back to these differ- ing single stars, the probability not to have a single planet
ences in Sect. 5. with P ≥ 100 days and M2 sin i ≥ 5 MJ among our sam-
For periods longer than ∼100 days, the distribution ple of planets found in multiple star systems is 3.25 %.
of the planetary companions in the M2 sin i–log P plane This result shows that this trend could be real. It is, how-
is not very different for the two samples. In this period ever, not possible to exclude with a high confidence level
range, the mean mass of planets found in multiple star that the lack of planets with M2 sin i ≥ 5 MJ and P ≥ 100
4 A. Eggenberger et al.: Statistical properties of exoplanets III

days may solely be due to small-number statistics and that 3.3. Remarks
planets with P ≥ 100 days orbiting in multiple star sys-
tems and around isolated stars may, in fact, belong to the The limit at ∼40 days is valid for both the eccentricity–
same population. A larger sample of long-period planets period and the mass–period diagrams, but in the latter
orbiting in multiple star systems will be required to settle the limiting period is less well defined. If we plot the evo-
this question. lution of the mean mass or the highest mass (averaged on
the three highest values) of planets orbiting single stars as
a function of the period (see Fig. 6 of Paper I), there is a
3.2. The eccentricity–period diagram jump at a period of ∼40 days, which reflects the distribu-
tion of these planets in the mass–period diagram. Thus,
The distribution of the extrasolar planetary candidates about the same limiting period is obtained by consider-
in the e–log P plane is illustrated in Fig. 2. In this dia- ing the distribution of planets orbiting single stars in the
gram, we note that all the planets with a period P . 40 mass–period diagram on the one hand, and by consider-
days orbiting in multiple star systems have an eccentricity ing the distribution of planets orbiting in multiple star
smaller than 0.05, whereas longer period planets found in systems in the eccentricity–period diagram on the other
multiple star systems can have larger eccentricities. Some hand. This is intriguing.
of the very short-period planets are so close to their par- Even if the orbital parameters of the binaries hosting
ent star that tidal dissipation in the planet could have planets are not exactly known, the projected separations
circularized their orbit, even if they were originally eccen- of these systems (see Table 1) indicate that the five plan-
tric (Rasio et al. 1996). For longer periods, the orbits are ets with a period shorter that 40 days reside in very differ-
not necessarily circularized anymore and any eccentricity ent types of systems. There is thus no obvious correlation
is possible. For the three planets with a period between 10 between the properties of these planets and the known
and 40 days orbiting in multiple star systems, the circu- orbital characteristics of the binaries or the star masses.
larization time (due to tidal dissipation in the planet) is The history of planets found in multiple planet systems
τc & 1012 years. This is clearly too long to explain the low is probably different from the one of ”single” planets. For
eccentricities of these planets by invoking tidal dissipation the analysis presented here, all planets have been consid-
alone. ered, but we have checked that our conclusions remain
The hypergeometric distribution (Appendix A) was unchanged if the planets belonging to multiple planet sys-
again used to test if the difference observed in this dia- tems are removed from the samples. Among the five short-
gram may solely be due to the small size of the sample period planets found in multiple star systems, two also
of short-period planets found in multiple star systems. In belong to multiple planet systems. This is something that
this case, we computed the probability to have 5 planets we must keep in mind when discussing the properties of
with log P ≤ 1.6 and e < 0.05 in a subsample of 5 plan- these planets.
ets with log P ≤ 1.6 drawn out of the total population
of planets with log P ≤ 1.6 (see Appendix A.2 for more
details). When selection effects are taken into account (by 4. Models of formation and evolution of giant
discarding from the counts the planets with a radial ve- planets in binaries
locity semiamplitude K < 15 ms−1 ), this probability is
3.77 %. This means that the probability to have 5 planets Let us now turn to planet formation models and consider
with log P ≤ 1.6 and e < 0.05 in the sample of short- their predictions regarding the existence and the survival
period planets orbiting in multiple star systems is 3.77 % of giant planets in binary star systems. We will then be
if these planets have the same properties as short-period in position to compare and confront our observations with
planets orbiting isolated stars. On the basis of the current their results.
samples, there is thus a trend towards a difference in the There are different points to take into account when
properties of short-period planets orbiting in multiple star considering the formation of giant planets in binaries.
systems and around isolated stars. The alternative state- Indeed, the stellar companion affects all the stages of
ment, namely that short-period planets orbiting in multi- planet formation as well as the subsequent evolution of
ple star systems and around isolated stars belong to the the planet once it has formed. The major points that have
same population can, nevertheless, not be excluded. The been studied in the literature are briefly described in this
observation that the three extrasolar planets with periods section.
between 10 and 40 days orbiting a component of a multiple
star system have very low eccentricities is interesting and
could be a clue to their formation and/or subsequent evo- 4.1. Binary-disc interactions
lution history. We will come back to this point in Sect. 5. Binary stars can in principle interact with three types of
For periods longer than ∼40 days, there is no signifi- discs: two circumstellar and one circumbinary. Transfer of
cant difference between planets found in multiple star sys- angular momentum between the binary and the disc leads
tems and around isolated stars. The mean eccentricity is to a truncation of the inner/outer edge for a circumbi-
similar for the two samples. nary/circumstellar disc, respectively.
 A. Eggenberger et al.: Statistical properties of exoplanets III 5

Artymowicz & Lubow (1994) investigated the approx- Pollack et al. 1996; Boss 1997; Mayer et al. 2002). The
imate sizes of discs as a function of binary mass ratio and core accretion mechanism begins with the collisional ac-
eccentricity for systems with circumstellar and circumbi- cumulation of planetesimals and planetary embryos in a
nary gaseous discs. Their results show that for a binary protoplanetary disc, as for terrestrial planet formation. In
with a mass parameter µ = M2 /(M1 + M2 ) = 0.3, the the outer parts of the disc, where the amount of solid ma-
inner edge of a circumbinary disc is typically located at terial is increased by the presence of ices, embryos may
rt = 2.0 ab (where ab is the binary semimajor axis) for reach about 10 M⊕ in ∼106 years and begin to grow an
nearly circular binaries and at rt = 3.0 ab for eb = 0.5. atmosphere of disc gas to form giant planets like Jupiter
The outer edge of a circumprimary disc lies at rt = 0.4 ab and Saturn in ∼107 years. In the disc instability model, a
for nearly circular binaries and at rt = 0.18 ab for eb = 0.5. gravitationally unstable disc fragments directly into self-
For a circumsecondary disc, the outer truncation radius is gravitating clumps of gas and dust that can contract and
located near rt = 0.27 ab for nearly circular binaries and become giant gaseous protoplanets. Coagulation and sed-
near rt = 0.15 ab for eb = 0.4. imentation of dust grains to the protoplanet center could
form a solid core. This process occurs over a dynamical
time scale: clump formation and dust grain sedimentation
4.2. Long-term stability of orbits proceed nearly simultaneously in ∼103 years. Though they
Assuming that planets can form in binary stars, do long- are different, these two mechanisms share a common char-
term stability regions exist for planetary orbits in these acteristic: giant planets should only form in the relatively
systems? Holman & Wiegert (1999) studied the long-time cool outer regions of protoplanetary discs.
survival of planets in different regions of phase space near Boss (1998) considered the influence of a binary com-
a binary star system. Circumprimary as well as circumbi- panion on giant planet formation via disc instability.
nary planets were studied for different values of the binary 3D hydrodynamical models of discs with 0.04 M⊙ were
eccentricity and mass ratio. For a binary with a mass pa- evolved in time, subject to the gravity of a binary star
rameter µ = 0.3, the largest stable orbit around the pri- companion placed on a circular orbit at 40 AU. In the ab-
mary star has a critical semimajor axis rc = 0.37 ab for sence of the binary companion, the disc is stable, but in
eb = 0.0 or rc = 0.14 ab for eb = 0.5. For the same binary, the presence of the binary companion the disc forms a
the smallest circumbinary orbit has a critical semimajor multi-Jupiter-mass protoplanet in 0.002 Myr.
axis rc = 2.3 ab for eb = 0.0 or rc = 3.9 ab for eb = 0.5. The evolution of two stars, each orbited by a circum-
The study of Holman & Wiegert (1999) was re- stellar disc, was simulated by Nelson (2000) using a two
stricted to planets in initially circular motion. dimensional smoothed particle hydrodynamic code. Each
Pilat-Lohinger & Dvorak (2002) extended this type component of the binary had a mass of 0.5 M⊙ and a bi-
of analysis and determined the variation of the stable nary eccentricity of 0.3 was considered. The system was
zone due to an increase of the initial planet eccentricity. evolved over 2700 yr (8 binary orbits). During and after
An increase of the planet eccentricity reduces the stable periastron each disc developed strong two-armed spiral
zone, this reduction being of course less pronounced structures which decayed to a smooth condition over the
that the one due to the same increase of the binary next half binary period; this cycle repeating with little
eccentricity. variations. The spiral structures decay was due to internal
As shown by Holman & Wiegert (1999) a companion heating in the discs, which increased their stability against
star orbiting beyond about 5 times the planetary distance spiral arm growth. Giant planet formation via gravita-
is not a serious threat to the long-term stability of plane- tional collapse is therefore unlikely in this system. In fact,
tary orbits. Nevertheless, this result only applies to orbits the temperatures in the discs are so high, that some grain
with a low mutual inclination. Innanen et al. (1997) inves- species, including water ices, are vaporized everywhere.
tigated the stability of planetary orbits in binary systems Giant planet formation by the core accretion mechanism
with emphasis on the inclination of the orbital planes. As is thus unlikely as well in this system.
an example, they studied the stability of the solar system
under the presence of an hypothetical distant companion
placed at 400 AU with different inclinations and masses.
4.4. Evolution of an embedded planet in a binary
Due to the Kozai mechanism (Kozai 1962), the system is A different approach was considered by Kley (2001) who
unstable at high inclination when the companion mass is studied the evolution of a giant planet still embedded in
larger than 0.05 M⊙ . A low-mass companion does, how- a protoplanetary disc around the primary component of
ever, not destabilize the system, even when the inclination a binary system. A 1 MJ planet was placed on a circular
is high. orbit at 5.2 AU from a 1 M⊙ star. The secondary star had
a mass of 0.5 M⊙ and an eccentricity of 0.5. The binary
4.3. Giant planet formation in binaries semimajor axis was varied from 50 to 100 AU. The sim-
ulations show that the companion alters the evolutionary
Two mechanisms have been proposed to explain giant properties of the planet: the mass accretion rate is in-
planet formation: core accretion and disc instability (e.g. creased and the inward migration time is reduced.
6 A. Eggenberger et al.: Statistical properties of exoplanets III

4.5. Summary and Zucker & Mazeh (2002) have shown that this effect is
statistically significant.
The main effect a companion has on a protoplanetary disc
is a truncation of its radius and an induction of waves Now, if we consider the evolution of a protoplanet or-
which, upon dissipation, transfer angular momentum be- biting the primary star of a binary system, we have seen
tween the binary and the disc. Comparing the results pre- that the presence of the companion alters the evolutionary
sented in Sect. 4.1 and in Sect. 4.2, we see that a planet can properties of the planet, in particular the migration and
almost always persists for a long time, wherever it forms mass growth rates are enhanced (Kley 2001). These dif-
in a truncated protoplanetary disc. The effect a secondary ferences may explain why the most massive short-period
star has on the efficiency of planet formation is, however, planets are found in multiple star systems: either they are
less clear. According to Nelson (2000), the companion has more massive because of the higher mass accretion rate, or
a negative influence, slowing or inhibiting altogether giant they are massive planets like the ones found with periods
planet formation. Boss (1998) claims the opposite, namely longer than ∼100 days around single stars, but orbiting
that giant planet formation via gravitational collapse is closer-in in multiple star systems because of the higher
favoured in binaries. More comprehensive studies will be migration rate. Both of these effects are in fact probably
needed, not only to clear up the case of binaries with a sep- mixed together and present at the same time.
aration of 40 or 50 AU, but also to explore what happens
for binaries with different projected separations, mass ra- Still regarding migration via the gravitational interac-
tios and eccentricities. Anyway, we expect the secondary tion of a Jupiter-mass protoplanet with a gaseous disc,
star to have an influence on planet formation, at least for models indicate that a realistic upper limit for the masses
close binaries, and Kley (2001) has shown that this re- of closely orbiting giant planets is ∼5 MJ , if they originate
mains true for the subsequent evolution of giant planets in protoplanetary discs similar to the minimum-mass solar
in binaries. Planets found in multiple stars systems may nebula (Nelson et al. 2000). Examples of large (> 5 MJ )
thus have different characteristics than planets orbiting planets at small orbital distances can, however, be ob-
isolated stars. tained due to migration in discs with different masses or
viscosities (Trilling et al. 1998). If such an upper mass
limit exists for short-period planets, and if the scenario
5. Discussion proposed by Kley (2001) is correct, we would expect to
find an upper mass limit for short-period planets orbiting
As mentioned in Sect. 4.3, in situ formation is very un- in multiple star systems that is larger than the one valid
likely for short-period Jupiter-mass planets. Formation at for planets orbiting single stars. Therefore, there should
larger distances followed by inward migration seems to exist a zone in the mass–period diagram where only plan-
be a better explanation to the existence of these planets ets orbiting in multiple star systems are found, the pres-
(e.g. Lin et al. 1996; Bodenheimer et al. 2000). The high ence of a stellar companion being the reason that enables
masses and low eccentricities of short-period planets or- a planet to reach a small separation with a mass larger
biting in multiple star systems (Sect. 2 and Figs. 1 and than the limit corresponding to planets orbiting isolated
2) seem also to indicate that some kind of migration has stars.
been at work in the history of these systems. A few differ-
ent migration mechanisms have been proposed to explain In the simulations by Kley (2001), the planet eccentric-
the existence and the characteristics of short-period giant ity is also modified: it first grows due to the perturbations
planets. We will now briefly discuss two of them and see induced by the secondary star, but then declines because
if they might explain some of the features emphasized in of the damping action of the disc. The final result is a
Sect. 2. rapid decay of the planet semimajor axis and a damping
of the initial eccentricity.
5.1. Planet–viscous disc interaction
Taken at face value, these arguments may provide an
One proposed migration mechanism involves the grav- explanation for the observation that the most massive
itational interaction of a protoplanet with the gaseous short-period planets are all found in multiple star systems
disc out of which it formed (Goldreich & Tremaine 1979, and have very small eccentricities. It should, however, be
1980; Ward 1986, 1997). Subject to such an interaction, noticed that several of the multiple star systems known
high mass planets will migrate more slowly than low to host planets are probably very different from the ones
mass planets in a given disc because they create larger studied by Kley (2001). The five planets with a period
gaps. Moreover, some of these planets will experience shorter than 40 days orbit in binaries with very differ-
mass loss as they come close to the central star. In over- ent separations (from ∼20 to ∼1000 AU) and it seems not
all, we thus expect to find more massive planets at in- likely that the perturbations produced by a wide compan-
termediate and large semimajor axes, the population of ion would influence the evolution of a protoplanet orbit-
close-in objects being dominated by smaller mass planets ing at or below a few AU. This, however, deserves further
(Trilling et al. 2002). This is indeed observed (Paper I) study.
 A. Eggenberger et al.: Statistical properties of exoplanets III 7

5.2. Kozai migration – the most massive (M2 sin i & 2 MJ ) short-period plan-
ets all orbit in multiple star systems;
Another mechanism that may be at work in wide binaries – the planets found in multiple star systems tend to have
is the so-called Kozai migration (Wu 2003; Wu & Murray
a very low eccentricity when their period is shorter
2003). In such a case, the Kozai mechanism (Kozai
than 40 days.
1962; see also Holman et al. 1997; Innanen et al. 1997;
Mazeh et al. 1997) produces large cyclic oscillations of the These observations seem to indicate that migration has
planet eccentricity. During the periods of high eccentric- played an important role in the history of the short-period
ity, the periastron is small and, consequently, tidal dissi- planets orbiting in multiple star systems and that migra-
pation becomes important and gradually removes energy tion may be induced differently in binaries than around
from the planetary orbit, eventually leading to circulariza- single stars.
tion. The Kozai mechanism coupled with tidal dissipation
From the theoretical point of view, it has been shown
is thus a viable method by which a planet can migrate
(Kley 2001) that the presence of a companion star affects
towards the central star.
the properties of a Jupiter-mass planet still embedded in a
Tidal dissipation depends sensitively on the nearest
disc around the primary component of a binary by increas-
approach distance and is important only if the planet can
ing the migration and mass accretion rates. Alternatively,
reach a high eccentricity. As the eccentricity oscillations
the Kozai mechanism may be (or have been) at work in bi-
only depend on the inclination of the planet orbital plane
naries hosting planets and will modify some of the orbital
relative to the binary orbital plane, the initial inclination
parameters of the planet. This mechanism can be efficient
between these two planes is a key parameter. Moreover,
in wide binaries and, coupled with tidal dissipation, it may
for the Kozai mechanism to be effective, the compan-
also lead to inward migration. Even if these two mecha-
ion must provide the dominant contribution to the ap-
nisms may be invoked to explain the characteristics of a
sidal precession of the planet (see Holman et al. 1997 and
few planets orbiting in binaries, none of them seem to be
Wu & Murray 2003 for more details). The efficiency of
able to account for all the properties of the five planets
the Kozai migration is, however, fairly independent of the
orbiting in multiple stars systems with a period P . 40
planet mass, and this mechanism will work for planets of
days. Nonetheless, it is also possible that diverse mecha-
relatively large masses (Wu 2003).
nisms may have been at work in these systems, but leading
Although the Kozai mechanism may be efficient in bi-
to a similar final state and similar planet properties. New
naries with large semimajor axes, several requirements
studies dedicated to this issue will be needed to settle this
must be simultaneously satisfied for it to operate, and
question and to find a satisfactory explanation to the ex-
such a mechanism will not apply to a large fraction of
istence and the characteristics of the short-period planets
planetary systems. Furthermore, even if the Kozai migra-
found in multiple star systems.
tion has been efficient during a period in the evolution
From the observational point of view, a larger sample
of a planet, it does not imply that its orbit is now cir-
of planets orbiting in multiple star systems will be required
cular. Up to now, the Kozai mechanism and the Kozai
to confirm or refute the preliminary trends emphasized in
migration have been considered to explain the high eccen-
this paper. In this context, the search for planets in mul-
tricity of given planetary candidates such as 16 Cyg B b
tiple star systems, even if more difficult to carry out than
and HD 80606 b (Holman et al. 1997; Mazeh et al. 1997;
the search for planets around single stars is of importance.
Wu & Murray 2003). It has never been demonstrated that
On the other hand, the characterization of the star sys-
the combination of the Kozai mechanism with tidal dis-
tems susceptible of hosting planets is underway and could
sipation may account for the existence of close-in planets
bring interesting constraints for the models, thus helping
with very low eccentricities. On the other hand, Kozai os-
our understanding of giant planet formation.
cillations are not likely to be currently at work for the
planets with short semimajor axes, in particular because
the Kozai mechanism is suppressed by general relativis-
tic effects. It is thus very unlikely that the low eccen-
Appendix A: The hypergeometric distribution
tricity of these planets may be due to the fact that they The hypergeometric distribution models the total number
are currently seen in their low-eccentricity phase. It seems of successes in a fixed size sample drawn without replace-
therefore difficult to explain the characteristics of all the ment from a finite population of N items of which G are
short-period planets orbiting in multiple stars systems by labelled success and (N − G) are labelled failure. The hy-
invoking the Kozai migration alone. pergeometric distribution is described by three parame-
ters: N , the size of the population; G, the total number
of items with the desired characteristics in the popula-
6. Conclusion
tion; and n, the size of the random sample drawn from
The characteristics of giant planets found in multiple star the population.
systems seem to be different from the ones of planets or- The probability distribution of the hypergeometric
biting single stars, at least for the short-period planets. random variable X, the number of successes in a random
The major differences are: sample of size n selected from the total population is:
8 A. Eggenberger et al.: Statistical properties of exoplanets III

ets drawn from the total population of planets is then

N −G
C G Cn−x P (X = 5) = 3.77 %.
P (X = x) = x N
Cn Acknowledgements. We thank the Swiss National Research
Foundation (FNRS) and the Geneva University for their con-
where x = max(0, n − (N − G)), . . . , min(n, G). This tinuous support to our planet search programmes. We thank
probability formula represents the ratio of the number of the anonymous referee for valuable suggestions regarding the
samples containing x successes and (n − x) failures to the statistics. This research has made use of the SIMBAD database
total number of possible samples of size n. and the VizieR catalogue access tool operated at CDS, France.
To test the statistical significance of the possible dif-
ference observed in the mass–period or in the eccentricity– References
period diagram, we considered a subsample of planets
drawn from the total population of extrasolar planets (i.e. Allen, C., Poveda, A., & Herrera, M. A. 2000, A&A, 356,
planets found in multiple star systems and around iso- 529
Artymowicz, P. & Lubow, S. H. 1994, ApJ, 421, 651
lated stars). A planet was labelled success if it was lo-
cated within the test zone of the diagram considered and Bodenheimer, P., Hubickyj, O., & Lissauer, J. J. 2000,
failure otherwise. We then computed the probability that Icarus, 143, 2
Boss, A. P. 1997, Science, 276, 1836
such a random subsample would give rise to a similar con-
figuration as the one actually observed for planets found Boss, A. P. 1998, AAS/Division for Planetary Sciences
Meeting, 30, 1057
in multiple star systems, namely a configuration with the
same number of planets within the test zone. Boss, A. P. 2000, Earth Moon and Planets, 81, 19
Boss, A. P. 2003, in Planetary systems and planets in sys-
tems, ed. S. Udry, W. Benz, & R. Vonsteiger, Space
A.1. The mass–period diagram Science Series of ISSI, Kluwer Academic Publishers, in
press
We give here more details regarding the statistical signif- Butler, R. P., Marcy, G. W., Fischer, D. A., et al. 1999,
icance of the lack of planets with M2 sin i ≥ 5 MJ and ApJ, 526, 916
P ≥ 100 days orbiting in multiple star systems (Fig. 1 Butler, R. P., Marcy, G. W., Williams, E., Hauser, H., &
and Sect. 3.1). As explained in Sect. 3.1, the hyperge- Shirts, P. 1997, ApJ, 474, L115
ometric distribution was used to compute the statisti- Campbell, B., Walker, G. A. H., & Yang, S. 1988, ApJ,
cal significance of the difference observed. The param- 331, 902
eters used were: N = 77, the number of planets with Cochran, W. D., Hatzes, A. P., Butler, R. P., & Marcy,
P ≥ 100 days and K ≥ 15.0 ms−1 ; G = 21, the num- G. W. 1997, ApJ, 483, 457
ber of planets with P ≥ 100 days, M2 sin i ≥ 5 MJ and Cochran, W. D., Hatzes, A. P., Endl, M., et al. 2002,
K ≥ 15.0 ms−1 ; n = 10, the number of planets with AAS/Division for Planetary Sciences Meeting, 34, 916
P ≥ 100 days and K ≥ 15.0 ms−1 found in multiple star Desidera, S., Gratton, R. G., Endl, M., et al. 2003, A&A,
systems; and x = 0, the number of planets with P ≥ 100 405, 207
days, K ≥ 15.0 ms−1 and M2 sin i ≥ 5 MJ found in mul- Duquennoy, A. & Mayor, M. 1991, A&A, 248, 485
tiple star systems. Given these parameters, we obtain a Eggenberger, A., Halbwachs, J., Udry, S., & Mayor, M.
probability P (X = 0) = 3.25 % to have no planet with 2003, in The environments and evolution of double and
P ≥ 100 days, K ≥ 15.0 ms−1 and M2 sin i ≥ 5 MJ among multiple stars, Rev. Mex. Astron. Astrofis., in press
a subsample of 10 planets drawn from the total population Els, S. G., Sterzik, M. F., Marchis, F., et al. 2001, A&A,
of planets. 370, L1
Fischer, D. A., Marcy, G. W., Butler, R. P., Vogt, S. S.,
A.2. The eccentricity–period diagram & Apps, K. 1999, PASP, 111, 50
Fischer, D. A., Marcy, G. W., Butler, R. P., et al. 2003,
More details concerning the statistical significance of the ApJ, 586, 1394
possible difference observed for short-period planets in Goldreich, P. & Tremaine, S. 1979, ApJ, 233, 857
the eccentricity–period diagram (Fig. 2 and Sect. 3.2) are Goldreich, P. & Tremaine, S. 1980, ApJ, 241, 425
given here. The parameters used to compute the hyper- Halbwachs, J. L. 1986, A&AS, 66, 131
geometric probability were: N = 25, the number of plan- Hale, A. 1994, AJ, 107, 306
ets with log P ≤ 1.6 and K ≥ 15.0 ms−1 ; G = 14, the Hatzes, A., Cochran, W. D., Endl, M., et al. 2003
number of planets with log P ≤ 1.6, K ≥ 15.0 ms−1 and Hauser, H. M. & Marcy, G. W. 1999, PASP, 111, 321
e < 0.05; n = 5, the number of planets with log P ≤ 1.6 Holman, M., Touma, J., & Tremaine, S. 1997, Nature, 386,
and K ≥ 15.0 ms−1 orbiting in multiple star systems; 254
and x = 5, the number of planets with log P ≤ 1.6, Holman, M. J. & Wiegert, P. A. 1999, AJ, 117, 621
K ≥ 15.0 ms−1 and e < 0.05 found in multiple star sys- Innanen, K. A., Zheng, J. Q., Mikkola, S., & Valtonen,
tems. The probability to have 5 planets with log P ≤ 1.6, M. J. 1997, AJ, 113, 1915
K ≥ 15.0 ms−1 and e < 0.05 in a subsample of 5 plan- Kley, W. 2001, in IAU Symposium 200, 511
 A. Eggenberger et al.: Statistical properties of exoplanets III 9

Kozai, Y. 1962, AJ, 67, 591 Wu, Y. 2003, in Scientific Frontiers in Research on
Latham, D. W., Stefanik, R. P., Mazeh, T., Mayor, M., & Extrasolar Planets, ed. D. Deming & S. Seager, ASP
Burki, G. 1989, Nature, 339, 38 Conf. Ser., in press
Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Wu, Y. & Murray, N. 2003, ApJ, 589, 605
Nature, 380, 606 Wuchterl, G., Guillot, T., & Lissauer, J. J. 2000,
Lowrance, P. J., Kirkpatrick, J. D., & Beichman, C. A. Protostars and Planets IV, 1081
2002, ApJ, 572, L79 Zucker, S. & Mazeh, T. 2002, ApJ, 568, L113
Marcy, G. W., Butler, R. P., Fischer, D. A., et al. 2002, Zucker, S., Naef, D., Latham, D. W., et al. 2002, ApJ, 568,
ApJ, 581, 1375 363
Marcy, G. W., Butler, R. P., Vogt, S. S., Fischer, D., & Zucker, S. et al. 2003, A&A, to be submitted
Liu, M. C. 1999, ApJ, 520, 239
Mayer, L., Quinn, T., Wadsley, J., & Stadel, J. 2002,
Science, 298, 1756
Mayor, M., Udry, S., Naef, D., et al. 2003, A&A, in press
Mazeh, T., Krymolowski, Y., & Rosenfeld, G. 1997, ApJ,
477, L103+
McGrath, M. A., Nelan, E., Black, D. C., et al. 2002, ApJ,
564, L27
Naef, D., Latham, D. W., Mayor, M., et al. 2001, A&A,
375, L27
Naef, D., Mayor, M., Korzennik, S., et al. 2003, A&A, in
press
Nelson, A. F. 2000, ApJ, 537, L65
Nelson, R. P., Papaloizou, J. C. B., Masset, F., & Kley,
W. 2000, MNRAS, 318, 18
Patience, J., White, R. J., Ghez, A. M., et al. 2002, ApJ,
581, 654
Pilat-Lohinger, E. & Dvorak, R. 2002, Celestial Mechanics
and Dynamical Astronomy, 82, 143
Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996,
Icarus, 124, 62
Queloz, D., Mayor, M., Naef, D., et al. 2000a, in
Planetary Systems in the Universe: Observations,
Formation and Evolution, IAU Symp. 202, ed.
A. Penny, P. Artymowicz, A.-M. Lagrange, &
S. Russell, ASP Conf. Ser., in press, also available at
http://obswww.unige.ch/Preprints/Preprints/cine art.html
Queloz, D., Mayor, M., Weber, L., et al. 2000b, A&A, 354,
99
Rasio, F. A., Tout, C. A., Lubow, S. H., & Livio, M. 1996,
ApJ, 470, 1187
Santos, N. C., Mayor, M., Naef, D., et al. 2002, A&A, 392,
215
Tokovinin, A. A., Griffin, R. F., Balega, Y. Y., Pluzhnik,
E. A., & Udry, S. 2000, Astronomy Letters, 26, 116
Trilling, D. E., Benz, W., Guillot, T., et al. 1998, ApJ,
500, 428
Trilling, D. E., Lunine, J. I., & Benz, W. 2002, A&A, 394,
241
Udry, S., Eggenberger, A., Mayor, M., Mazeh, T., &
Zucker, S. 2003a, in The environments and evolution of
double and multiple stars, Rev. Mex. Astron. Astrofis.,
in press
Udry, S., Mayor, M., Naef, D., et al. 2002, A&A, 390, 267
Udry, S., Mayor, M., & Santos, N. C. 2003b, A&A, 407,
369, Paper I
Ward, W. R. 1986, Icarus, 67, 164
Ward, W. R. 1997, Icarus, 126, 261