Sie sind auf Seite 1von 8

ICARUS

ARTICLE NO.

132, 196203 (1998) IS985900

Habitable Planet Formation in Binary Star Systems


Daniel P. Whitmire, John J. Matese, and Lee Criswell
Department of Physics, University of Southwestern Louisiana, Lafayette, Louisiana 70504 E-mail: whitmire@usl.edu

and Seppo Mikkola


Tuorla Observatory, Turku University, 21500 Piikkio, Finland Received March 26, 1997; revised December 24, 1997

1. INTRODUCTION Assuming current models of terrestrial planet formation in the Solar System, we numerically investigate the conditions under which the secondary star in a binary system will inhibit planet growth in the circumstellar habitable zone. Runaway accretion is assumed to be precluded if the secondary (1) causes the planetesimal orbits to cross within the runaway accretion time scale and (2) if, during crossing, the relative velocities of the planetesimals have been accelerated beyond a certain critical value which results in disruption collisions rather than accretion. For a two solar mass binary with planetesimals in circular orbits about one star at 1 AU, and a typical wide binary eccentricity of 0.5, the minimum binary semimajor axis which would not inhibit planet formation, ac , is 32 AU. If the planetesimals orbit the center of mass of the binary system, ac 0.10 AU, which is inside the tidal circularization radius. We obtain an empirical formula giving the dependencies of ac on the binary eccentricity, secondary mass, planetesimal location, and critical disruption velocity. Based on the distributions of orbital elements of a bias-corrected sample of nearby G-dwarfs, we nd that 60% of solar-type binaries cannot be excluded from having a habitable planet solely on the basis of the perturbative effect of the secondary star. This conclusion is independent of when the secondary star formed, nebula dissipative mechanisms, and the time scale for runaway planetesimal accretion, and is relatively insensitive to the mass of the secondary star, the critical disruption velocity, and the location of planetesimals within the circumstellar habitable zone. An earlier study of planet formation in binary star systems came to a different conclusion, namely that planet formation, even at Mercurys distance, is unlikely except in widely separated systems ( 50 AU), or when the secondary has a very low mass and near circular orbit as in the SunJupiter system. The discrepancy with the present numerical study is due in part to the different runaway accretion time scales assumed and the neglect in the earlier study of an exact criterion for crossing orbits. 1998
Academic Press

Key Words: planet formation; binary stars; life.


196
0019-1035/98 $25.00 Copyright 1998 by Academic Press All rights of reproduction in any form reserved.

Modeling the accretion of terrestrial planets in the Solar System (Wetherill and Stewart 1993) and around other single stars (Wetherill 1996) suggests that planet formation within a stars habitable zone may be common. We denote a terrestrial planet that forms within the circumstellar habitable zone (Kasting et al. 1993; Whitmire and Reynolds 1996) as a potentially habitable planet, recognizing that many other conditions may be necessary for life to actually evolve on such a planet. Since 2/3 of solar type stars are known members of multiple star systems it is of interest to consider the constraints placed on planet formation due to the presence of a secondary star. Two congurations will be considered: the internal-planetesimal geometry in which the planetesimals orbit the primary and the externalplanetesimal geometry in which the planetesimals orbit the center of mass of the binary system. We assume that the secondary star formed simultaneously with the primary. If this were not the case and the secondary star formed or was captured after the initial stages of planet formation were complete, then planet accretion could have proceeded as expected for a single star. In this case the only restriction would be the dynamical ejection of the planet itself (Graziani and Black 1981; Pendleton and Black 1983). The assumption of simultaneous star formation is conservative in the sense that if it is invalid the probability of habitable planet formation is greater than our analysis will suggest. The accretion of planetesimals from a turbulent gas and dust nebula is not well understood theoretically even in the case of our own Solar System. Nonetheless, dust grains managed to collect into planetesimals even in the parent asteroid belt where the perturbation due to Jupiter or proto-Jupiter would have been stronger than will be assumed in our binary star analysis. More direct evidence

HABITABLE PLANETS IN BINARY STAR SYSTEMS

197

that planetesimals can form from dust grains in binary star systems is given by Kalas and Jewitt (1997), who have observed a dust disk surrounding the BV5 binary system BD 31 643. Based on dust lifetime arguments they conclude that the dust must have planetesimal sources located at 1,000 AU. The observed binary separation is 200 AU. This system is an (external-planetesimal case) example of a binary in which planetesimals have formed at distances of ve times the observed stellar separation. If the binary orbit were circular, contrary to expectations, the ratio of planetesimal formation to binary semimajor axis would be 5 : 1. This is within a factor of about 2 of the dynamical limit for planetesimal ejection (of coplanar orbits) from the system and much closer than our analysis will require (typical ratio 30 : 1). Although the entire disk in this system may have more total mass and therefore potentially more dissipation than disks typical of solar-type binaries, the dust-to-planetesimal accretion distance is 1,000 AU from the center of mass of the binary, making it likely that the surface density is less than at 1 AU in a solar type disk. The most perturbation sensitive stage in the terrestrial planet formation process is the runaway planetesimal accretion phase (Wetherill and Stewart 1993, hereafter WS; Lissauer 1993). This phase is rapid because of the small relative velocities between the planetesimals and the growing embryo as a result of dynamical friction. The runaway accretion time scale is 2 104 yr (WS). This phase ends when the relative velocities of the growing embryos increase due to their mutual gravitational interactions. Subsequent growth by nondestructive embryoembryo collisions to a nal planet occurs on a much longer time scale of 107 yr. If the relative velocities at innity U of orbit-crossing planetesimals are accelerated beyond a certain critical value Uc then runaway accretion and thus planet formation will be precluded since collisions will then cause disruption rather than accretion. In a binary star system or a system containing a massive planet or brown dwarf the secondary body may accelerate planetesimals beyond this limit. For planetesimals bound by nongravitational forces alone (expected for planetesimal radii up to 100 km) the critical velocity is independent of mass and is 100 m s 1 according to laboratory measurements (Greenberg et al. 1977) and simple theoretical arguments (Heppenheimer 1978; Wetherill 1991). Including the gravity of the colliding planetesimals will U 2 V 2 and the increase both the impact velocity c es escape velocity from the merged pair. The additional relative kinetic energy due to free fall will result in the (two) collision fragments having nonzero velocities but, beginning with a velocity at innity Uc , these fragments cannot have a velocity greater than Ves , thus justifying the neglect of gravity in our conservative analysis. In a more realistic collision with a spectrum of fragment masses and velocities

it is possible for some fragments to achieve escape velocity even when U Uc , but most mass will remain bound. Depending on the mass of the planetesimals, velocity pumping by the secondary body will be countered by several dissipative processes, which include inelastic collisions, gas drag, viscous stirring, and dynamical friction. For growing planetesimals of size 10 km the effects of dissipative processes other than dynamical friction and inelastic collisions are usually assumed to be negligible. We conservatively neglect all damping mechanisms in our analysis since the net effect of these processes is to make planet formation more probable. Velocity acceleration due to mutual perturbations between large planetesimals and embryos is important in current models for planet formation in the Solar System and other single stars, causing growing embryos to collide and grow. The additional effect of the secondary star on the interactions between planetesimals will be taken into account directly in our numerical calculations discussed below. Internal-planetesimal formation in binary star systems was rst considered in some detail by Heppenheimer (1974; 1978, hereafter H78). That investigation was based on the then current GoldreichWard (1973) model of terrestrial planet formation and on the assumption that planetesimal collisions would lead to disruption rather than accretion if the relative velocities between planetesimals exceeded the critical velocity Uc 100 m s 1. The time scale for planetesimalplanetesimal accretion during which this value of Uc would be relevant was assumed to be 107 yr. The increase in relative velocity due to planetesimal eccentricity pumping and differential precession was then computed from secular perturbation theory. A reduction in eccentricity due to two different models of nebular gas drag was also considered in the analysis. The conclusion of Heppenheimers study was that planet formation even at Mercurys distance was not likely except for widely separated binaries ( 50 AU) or when the secondary mass was small and its orbit circular as in the case of the SunJupiter system. Even for this system it was assumed necessary to invoke the gravitational damping of eccentricities by the solar nebula to allow accretion of the terrestrial planets. If accurate, that study would imply that 2/3 of solar-type stars could be excluded as candidates for habitable planets. The present study comes to a different conclusion and we return to the explanation for this discrepancy after presentation of our analysis and results.
2. ANALYSIS

2.1. Crossing Orbits We wish to determine an analytic criterion for the crossing of two coplanar planetesimal orbits given by

198 1 r1 1 r2 1 1 e1 cos ( p1 e2 cos( ) , p2

WHITMIRE ET AL.

(1) (2)

2.2. Relative Velocity Given that two elliptical orbits satisfy the above crossing criterion we determine their relative velocities U at the two crossing points (corresponding to the two general solutions for cos ) from (U )2 e1 sin( p1 p1 r r p2
2

where r1 (r2) is the radial position of planetesimal 1 (2), is their common polar angle at the point(s) of intersection, 1 2 is the longitude of periastron (i.e., the angle between the two periastron vectors), e1 (e2) is the eccentricity, and p1 (p2) is the semilatus rectum of the ellipse (1 e 2 )a1(2) . 1(2) At orbit intersection 1 r1 1 r2 0 1 p1 1 p2 e1 cos( p1

e2 sin p2 ,

(7)

e2 cos . p2

where U is divided by the orbital velocity at 1 AU 30 km s 1 and r, p are in AU, (3) r 1 p2 e2 cos p1 1 e1 cos ( . (8)

The intersection point angles are cos where A B C p2 p1 AB C C 2 B2 B2 C 2 A2 (4)

We note that even when orbits cross and the relative velocity is greater than 100 m s 1 an actual collision need not occur in 2 104 yr since the orbits are not entirely lled with planetesimals. The assumption that a destructive collision will always occur when orbits cross and the relative velocity is greater than Uc is therefore conservative. 2.3. Calculation of Orbital Evolution For the internal conguration case where the planetesimals orbit the primary star, the relevant secular perturbation theory equations for the time dependence of e and for planetesimals with initial eccentricities 0 are (H78) e(t) 5 a eB u t sin 2 aB 1 e 2 2 B 1 sin ut cos ut M a3/2 1 , 2 3/2 e B) m1/2 a3 B (9) (10) (11)

e1p2 cos e1p2 sin

e2 p1

(5)

Physical solutions (i.e., orbit intersections) occur when C2 B2 A2 0,

yielding the necessary criterion for orbit crossing (e1 p2)2 (e2 p1)2 2e1p2e2 p1 cos

( p)2,

(6)

tan( t) u

where p p2 p1 . Two isolated planetesimals will eventually collide if their 1/3 orbits come within a critical distance 2.4( 1 , 2) where is the planetesimal mass in solar units (Gladman 1993). For planetesimals of density 1 g cm 3, located at 1 AU, and having radii in the range 11,000 km the critical collision radius is 4 10 6 4 10 3 AU. As discussed below, if the separation between planetesimals at 1 AU is less than 10 3 AU the perturbation of the secondary star (not present in Gladmans analysis) will rapidly, within a few orbits, increase a to values 10 3 AU. The time in which the two nearly circular orbits are within the critical distance of each other is negligibly small for the binary parameters used in this analysis. The orbits can still cross and a collision can be inferred to have occurred.

3 2 (1

Where aB and eB are the semimajor axis and eccentricity of the binary star system and M and m are the masses of the primary and secondary star, respectively. The secular perturbation approximation is valid only in the limit (a/aB) 1 for the internal-planetesimal conguration. This approximation is also inaccurate in the limit eB 0 since Eq. (9) predicts that the planetesimal eccentricity e 0 in this limit. Test calculations using a numerical N-body code and an integration time of 2 104 yr resulted in systematically larger values of the critical binary semimajor axis ac (beyond which planet accretion would not be inhibited) compared with the secular approximation. It

HABITABLE PLANETS IN BINARY STAR SYSTEMS

199

was also found that planestesimal semimajor axis variations were not negligible as assumed in the secular approximation. The secular approximation formulae are obtained by averaging over both the planetesimal and binary orbits and therefore information about variations in the orbital elements within a single binary orbit are lost. The numerical calculations showed that typically these variations were dominant during the short 2 104 yr integration time of interest. Applying our conservative theme, and because the numerical results are more accurate, we used the numerical calculations to obtain the present results and employed the secular approximation only as a test of the N-body code in an appropriate limit. The N-body code used for these calculations is based on the symplectic mapping method of Wisdom and Holman (1991). It was developed by S. Mikkola and K. Innanen and has proved itself in earlier works (e.g., Mikkola and Innanen 1995). For the present application several additional tests were performed to ensure that the program was functioning properly, including comparison with the analytic secular perturbation equations in the limit (a/aB) 1, large binary semimajor axis, and moderate eccentricity. In another test we compared the numerical output with an analytically derived Jacobi constant. For this test eB 0 and the planetesimal was given an arbitrary eccentricity and inclination. The Jacobi constant was found to be conserved to the expected high accuracy at each computational step. The initial conditions of the simulated 4-body systems were chosen as follows. Two planetesimals of negligible mass were started in coplanar circular orbits around the primary star in the invariant plane of the binary system and orbiting in the same sense as the binary. The initial positions of the two planetesimals were 90 from the binary semimajor axis and the secondary star was started at binary periastron. This conguration was chosen because it was found empirically that it resulted in the largest perturbation (relative velocities) during crossing and is therefore the most conservative initial conguration. This is likely due to the fact that this conguration results in the greatest initial differential torque on the planetesimals. The orbital elements of the planetesimals were recorded only at binary periastron since it was found empirically, as expected, that the orbital phase corresponding to the maximum perturbation of the planetesimals (as measured by relative velocity) was near periastron. Comparing the orbital elements at xed time intervals (and therefore at random orbit phases) systematically resulted in smaller relative velocities, and therefore smaller values of ac , during the xed integration time. From Eqs. (6) and (7) it is seen that in general crossing and relative velocity during crossing depend on the osculating values of a, e, and . For our initial conditions both eo and o 0. Experimentation showed that, for

the binary parameter space of interest, values of ao 0.001 AU resulted in the osculating a increasing to 0.001 AU or larger within a few binary orbital periods. It was also found that values of ao signicantly greater than 0.01 AU never produced crossings within 2 104 yr. Our simulations were carried out for three values of ao , 0.001, 0.003, and 0.01 AU. It was found that the maximum relative velocity obtained at orbit intersection was essentially independent of ao for ao 0.01 AU. On the other hand, orbit crossing itself was typically dependent on ao . For a given set of binary parameters, if crossing and a given maximum relative velocity were achieved for ao 0.001 AU, the principal change when the simulation was re0.003 and 0.01 AU was that crossings peated with ao occurred for a shorter time interval at comparable maximum relative velocities. The recorded planetesimal orbital elements were inserted into Eqs. (6) and (7) to determine if the orbits crossed and, if they did, what the relative velocities were at the points of intersection. As noted earlier, even when orbits cross an actual collision need not occur in 2 104 yr since the orbits are not entirely lled with planetesimals. The assumption that a collision will necessarily occur is thus conservative.
3. RESULTS

3.1. Internal-Planetesimal Case We focus on planetesimal accretion in the habitable zone of a 1 M primary. The habitable zone for a 1 M star depends on the evolutionary state of the star and the habitability time scale of interest (Kasting et al. 1993; Whitmire and Reynolds 1996): The narrowest habitable zone considered was the 4.5 Gyr continuously habitable zone. Using the most restrictive climatic assumptions this zone extended from 0.951.15 AU. Less restrictive climatic assumptions or shorter time scales result in somewhat larger habitable zones. The presence of a second star of luminosity equal to or less than that of the primary will not signicantly alter the location of the habitable zone for the range of semimajor axes of relevance for the internal-planetesimal case. Figures 14 show the dependence of the critical binary semimajor axes ac on the binary eccentricity eB , the mass of the secondary star m, the mean radius of the planetesimals a, and the critical disruption velocity Uc . Habitable planet growth will not be inhibited in binaries with semimajor axes greater than ac . Except when specied otherwise, the secondary star mass m 1 M , the binary star system eccentricity eB 0.5, the average planetesimal semimajor axes a 1.0 AU, and the critical planetesimal disruption velocity Uc 100 m s 1 0.003 VKepler . Fitting the numerical data in Figs. 14 gives an empirical function relating ac to the system parameters:

200

WHITMIRE ET AL.

FIG. 1. Critical binary semimajor axis ac beyond which planet formation at 1 AU will not be inhibited by a 1 M secondary star, as a function of binary eccentricity eB . Nonexclusion requires a binary periastron distance q 16 AU.

FIG. 3. Critical binary semimajor axis ac beyond which planet formation at 1 AU will not be inhibited by a 1 M secondary star, as a function of the mean planetesimal semimajor axis, a.

ac(eB , m, a, Uc)

16

1 1 eB
0.80

m 1M

0.31

(12)
1 0.30

a 1 AU

100 m s Uc

AU.

Fixing the other three parameters, these dependencies can be understood qualitatively as follows. The eB dependency corresponds to a constant periastron distance (for two 1 M stars this constant is 16 AU), which implies that the dominant effect is, as expected, the maximum perturbation at periastron. Larger values of a with the same periastron distance are equally effective in inhibiting

accretion (over the range of semimajor axes studied). Larger values of the mass of the secondary star and larger values of the average planetesimal radius a will increase the perturbation and thus increase ac . Larger values of the critical disruption velocity Uc means that a larger relative velocity at innity U is required for disruption, thus reducing ac . The critical binary semimajor axis is seen to be insensitive to secondary mass and critical disruption velocity. It is somewhat sensitive to the mean planetesimal semimajor axis, but since we are interested in habitable planet formation the actual range of a is restricted to 0.951.15 AU or somewhat larger depending on climatic assumptions and time scales of interest, as noted above. The critical semima-

FIG. 2. Critical binary semimajor axis ac beyond which planet formation at 1 AU will not be inhibited by a 1 M secondary star, as a function of secondary star mass m.

FIG. 4. Critical binary semimajor axis ac beyond which planet formation at 1 AU will not be inhibited by a 1 M secondary star, as a function of disruption velocity, Uc .

HABITABLE PLANETS IN BINARY STAR SYSTEMS

201

jor axis was found to be essentially independent of the assumed runaway accretion time scale for integration times between 2 104 and 1 105 yr. This can be understood as being due to the dominance of the intraorbit perturbations over this time scale. For longer integration times such as the 107 yr accretion assumed in H78 the differential precession of the orbits would dominate and the results would depend more on time (as well as the initial a). 3.2. External-Planetesimal Case Because of the broad distribution of solar-type binary star orbital elements (Duquennoy and Mayor 1991, hereafter DM91), the external-planetesimal case is much less important for habitable planet considerations. Observations show that solar-type binaries with periods less than 11 days (semimajor axis 0.12 AU for two 1 M stars) have circular orbits (DM91), presumably the result of premain-sequence tidal torques. The narrow habitable zone depends on luminosity so for two 1 M stars it moves out to 1.4 AU. Setting a 1.4 AU, ac was found to be 0.11 AU. Since this is less than the tidal circularization radius no additional eccentricities were simulated. For the more typical case where the secondary is signicantly less massive than the primary a 1.0 AU due to the strong sensitivity of luminosity to stellar mass. Assuming the internal-planetesimal parameter dependencies apply, ac 0.10 AU for m 0.5. 3.3. Statistics of Habitable Binaries Duquennoy and Mayor (DM91) give the distributions of the orbital elements in an unbiased sample of nearby G star binaries. The median binary eccentricity is 0.5 after correction for observational bias. The observed median mass ratio is no greater than 0.5 and we take m 0.5 M as a conservative limit for the median secondary mass. Using this mass and the median eccentricity we obtain from Figs. 1 and 2 a median value of ac 26 AU. The median orbital period of the sample is 180 yr, which corresponds to a semimajor axis of 36 AU. Figure 5 gives the distribution of aB , based on the period distribution given by DM91 and the median secondary mass. The right vertical dashed line at aB 26 AU (log(aB) 1.415) denotes ac for the median system. Statistically, the fraction of internal-planetesimal binaries which could have accreted a planet at 1 AU is the relative area under the curve to the right of this line, which is 57%. The left vertical dashed line is the external-planetesimal ac 0.10 AU. The fraction of binaries with semimajor axes less than 0.10 AU is the area to the left of the dashed line 2%. Thus the total fraction of solar-type binaries for which habitable planet accretion cannot be excluded is 59%. There is some uncertainty in the median eB for the internal-planetesimal case. If the median eB were random and 0.7, rather than the

FIG. 5. Observed bias-corrected distribution of binary semimajor axes, aB , assuming m 0.5 M . The left (right) vertical dashed line denotes the critical semimajor axis for the external-planetesimal (internalplanetesimal) case.

observed bias-corrected value of 0.5 (DM91), the total nonexcluded fraction would be reduced to 52%. The uncertainty in the median eB is our least conservative assumption.
4. COMPARISON WITH EARLIER WORK

Our results are signicantly different from those found in the earlier analysis of Heppenheimer (H78). That study, which considered only the internal-planetesimal conguration, concluded that planet growth even at Mercurys distance of 0.39 AU would require binary semimajor axes 50 AU, or small mass secondaries in circular orbits like that of the SunJupiter system. Even for the SunJupiter system it was assumed necessary to invoke nebula disk gravitational damping of planetesimal eccentricities for consistency with the existence of terrestrial planets in the Solar System. For the typical internal-planetesimal binary eccentricity of 0.5 and secondary mass 0.5 M , our analysis allows planet growth at 1 AU to occur for semimajor axes 26 AU. The discrepancy between Heppenheimers and the present study can be traced to three factors: (1) The different time scales appropriate for the planet accretion models used, (2) the absence of an exact criterion for crossing orbits in Heppenheimers study, and (3) the application of the secular-average approximation in that analysis. It is primarily differences (1) and (2) that explain our smaller values for ac . In Heppenheimers analysis the planetesimal accretion time scale, during which the critical disruption velocity was set equal to 100 m s 1, was taken to be 107 yr. This is in

202

WHITMIRE ET AL.

contrast to our use of current terrestrial planet accretion models in which this phase is only 2 104 yr (WS93). During Heppenheimers much longer time scale the relative velocities of planetesimals tended to achieve their maximum values due to differential precession of their orbits. The differential precession time scale depends on the planetesimal separation a, which was somewhat arbitrarily taken to be 0.01 AU. For this value of a the precession time scale was 105 yr for the typical binary (which was not specied). Thus, in the assumed planetesimal accretion time scale, there was more than enough time for the orbits to become randomized, in which case Umax 100 m s 1 for the typical binary. This is not emaxVKepler the case for the much shorter runaway accretion time scale appropriate to current accretion models. In Heppenheimers analysis there was no assurance that when the relative planetesimal velocity exceeded 100 m s 1 the orbits actually crossed, allowing for the possibility of a collision. A critical relative velocity is a necessary but not sufcient condition for disruption. Our analysis computed the relative velocity only if the crossing criterion was met. As noted earlier, the secular approximation used by Heppenheimer averages over the binary orbital period and therefore cannot predict the intraorbit excursions that are the dominant effect causing orbit crossing and maximum relative velocities during the short integration time of relevance in current accretion models. The secular approximation, when applied to the accretion time of 2 104 yr, was found to systematically lead to smaller values of ac for a given set of parameters compared to our numerical calculations. Thus, although Heppenheimers secular calculation for his longer time scale was different from our numerical calculations for the shorter time scale, this is not why our values of ac are signicantly less than those of Heppenheimers study. Since our approach and assumptions have been conservative, the calculated function ac(eB , m, a, Uc) may be too large (small) for the internal-planetesimal (externalplanetesimal) case and therefore, in reality, a greater fraction of solar-type binaries may have accreted a habitable planet. This is in contrast to the negative approach of Heppenheimer which, for example, required additional assumptions about the physics of dissipative mechanisms (an important mechanism in current planetesimal accretion models is dynamical friction, which was not considered in that study) and roughly simultaneous formation of both stars.
5. DISCUSSION

Embryoembryo collisions occur over time scales of 107 yr (WS93). Internal-planetesimal numerical simulations of this phase of planet formation showed that when

crossing occurred the relative velocity was comparable to its maximum value (Umax emaxVKepler) since the orbits are essentially randomized. This Umax is typically much greater than 100 m s 1. However, the requisite disruption velocity for embryos is also much greater. Embryos are gravitationally bound and have sizes between those of the Moon and Mercury (WS93; Wetherill 1996), whose escape velocities are 2 and 4 km s 1 respectively. Assuming 3 km s 1 for the escape velocity of a single embryo, we increase this by a factor of 2, obtaining 4.2 km s 1, to take into account the fact that disruption requires escape from the gravitational force of two embryo masses. For all eB 0.9 the maximum relative velocity is less than 4.2 km s 1 for aB ac , as determined in the planetesimal disruption analysis. Therefore the results for ac based on the planetesimal disruption analysis will not be subsequently invalidated by embryoembryo collisions except possibly in cases with very large binary eccentricities. In our simulations we have assumed (as did H78) that the binary and planetesimal orbits were coplanar. It is likely that this is approximately true for binary semimajor axes less than 30 AU as indirect observations (Hale 1994) and our own Solar System suggest. All planets, with the exception of Mercury (which lies in the Suns equatorial plane) and Pluto, lie close to the invariant plane, which is essentially Jupiters orbital plane, in spite of the fact that the Suns equatorial plane is tilted 7 relative to this plane. The observed large comets/planetesimals in the Kuiper belt continue this trend to over 40 AU. For a 1 M secondary it is likely that the accretion disk would relax to the invariant binary orbital plane prior to planetesimal accretion to distances equal to or even closer than Mercurys for the internal-planetesimal case and out to distances much further than 40 AU for the external-planetesimal case. In binaries with aB greater than ac it is conceivable that over 107 yr time scales relative velocities sufcient to disrupt embryos could develop due to differential nodal precession if the relative inclination were sufciently large. Should this occur, the small debris from such collisions would itself relax through inelastic collisions to the invariant binary plane and the entire solid body accretion process would begin again, though without the full gas nebula present. Although not considered binary star systems, the newly discovered extrasolar planets and brown dwarfs can be treated as the secondaries in our model and we can inquire whether a habitable planet could accrete in these systems assuming the secondaries formed in situ in their present orbits. These primary stars are solar-type and have masses near 1 M . The secondaries have masses ranging from 0.47 MJ to 60 MJ , where MJ is the mass of Jupiter 0.001 M . Using data for the 18 conrmed planets/ brown dwarfs compiled by Schneider (1997), we nd that the following six systems, all of which are the external-

HABITABLE PLANETS IN BINARY STAR SYSTEMS

203

planetesimal type, cannot be excluded from having a habitable planet at 1 AU: 51 Peg, And 55, Cnc, Boo, HD98230, and HD283750. As an additional illustration we simulated the SunJupiter system assuming (contrary to conventional theory) that Jupiter (or proto-Jupiter) formed before planetesimal accretion at 1 AU. Our model was consistent with the existence of a planet at 1 AU but not at the inner edge of the asteroid belt at 2.2 AU.
6. CONCLUSIONS

Gladman, B. 1993. Dynamics of systems of two close planets. Icarus 106, 247263. Goldreich, P. M., and W. R. Ward 1973. The formation of planetesimals. Astrophys. J. 183, 10511061. Graziani, F., and D. C. Black 1981. Orbital stability constraints on the nature of planetary systems. Astrophys. J. 251, 337341. Greenberg, R., D. R. Davis, W. K. Hartman, and C. R. Chapman 1977. Size distribution of particles in planetary rings. Icarus 30, 769 779. Heppenheimer, T. A. 1974. Outline of a theory of planet formation in binary systems. Icarus 22, 436447. Heppenheimer, T. A. 1978. On the formation of planets in binary star systems. Astron. Astrophys. 65, 421426. Kasting, J. F., D. P. Whitmire, and R. T. Reynolds 1993. Habitable zones around main sequence stars. Icarus 101, 108128. Lissauer, J. 1993. Planet formation. Annu. Rev. Astron. Astrophys. 31, 129174. Mikkola, S., and Innanen, K. 1995, Mon. Not. R. Astron. Soc. 277, 497 501. Pendleton, Y. J., and D. C. Black 1983. Further studies on the criteria for the onset of dynamical instability in general three-body systems. Astron. J. 88, 14151419. Schneider, J. 1997. Extrasolar Planets Encyclopedia. Available http:// www.obspm.fr/planets. Wetherill, G. W. 1991. Occurrence of Earth-like bodies in planetary systems. Science 253, 535538. Wetherill, G. W. 1996. The formation and habitability of extra-solar planets. Icarus 119, 219238. Wetherill, G. W., and G. R. Stewart 1993. Formation of planetary embryos: Effects of fragmentation, low relative velocity, and independent variation of eccentricity and inclination. Icarus 106, 190209. Whitmire, D. P., and R. T. Reynolds 1996. Circumstellar habitable zones: Astrophysical considerations. Circumstellar Habitable Zones: Proceedings of the First International Conference (L. R. Doyle, Ed.), pp. 117 142. Travis House Publications, Menlo Park, CA. Wisdom, J., and Holman, M. 1991. Symplectic maps for the N-body problem. Astron. J. 102, 15281538.

Assuming current terrestrial planet accretion models, the critical binary semimajor axis, beyond which habitable planet formation will not be inhibited by the presence of the secondary star, has been calculated as a function of binary eccentricity, secondary star mass, planetesimal location, and critical disruption velocity. Using the distributions of orbital elements of an unbiased sample of nearby solartype binary stars, we nd that 60% of these systems cannot be excluded from having a habitable planet solely on the basis of the gravitational perturbations of the secondary star. Future astronomical searches for habitable planets, as well as targeted SETI searches, need not exclude binary star systems whose semimajor axes, eccentricities and masses lie above the curves in Figs. 13.
ACKNOWLEDGMENTS
We thank Ray T. Reynolds and Patrick G. Whitman for advice and suggestions.

REFERENCES
Duquennoy, A., and M. Mayor 1991. Multiplicity among solar-type stars in the solar neighborhood. II. Distribution of the orbital elements in an unbiased sample. Astron. Astrophys. 248, 485524.

Das könnte Ihnen auch gefallen