How to change the relative inclination in a triple-star system

Naive numerical experiments with tidal dissipation
Staffan Söderhjelm, Lund Observatory
Turku Few-body workshop, July 2005

• • •
Algol 90 deg problem j-change in hierarchical triple star systems Primordial Algol with and without Kozaicycles (varying the apsidal motion k(2)) • Tidal dissipation+Kozai-cycles shrinks close orbit and changes j • No explanation for 90 deg j, but simple theory works in other cases

Original inspiration, Algol
• Söderhjelm(1975): orbits co-planar because
observed inclination-changes small (90 deg-case deemed implausible) • Lestrade et al(1993): orbits really at about 90 (±10) deg relative inclination • From the unchanging eclipse depths, j=90±3 deg (dΩ/dt ~ cos j) • Is there some mechanism favouring exactly 90 degrees??

Some parameters in a hierarchical triple star system
• • • •
Period-ratio Pc/Pw Ratio of orbital angular momenta G1/G2 Relative inclination between orbit-planes, j Eccentricities ec(variable) and ew

Fundamental formula for the relative inclination
Total (orbital) angular momentum is the (vector) sum of the individual momenta C2= G12+G22+2 G1G2 cos j or cos j =0.5 [((C/G2)2- 1)/(G1/G2)-G1/G2] that is, for constant C and G2, the relative inclination is determined by G1

Causes of G1-variation
The orbital angular momentum in the inner orbit is G1=M1M2/(M1+M2) [G(M1+M2)ac(1-ec2)]0.5 and typically, G1 diminishes during episodes of high eccentricity and/or when ac is shrunk by tidal dissipation

The j-variation depends strongly on G1/G2
• In most hierarchical triples, G1/G2 <<1, that is,
the wide orbit dominates the angular momentum budget • When the third mass is small relative to the other two, and P1/P2 is not too small, G1 may become dominant, however, which has interesting consequences

A typical, almost symmetric case initial G1/G2=0.05

Same plot with a/a0 on abscissa G1/G2=0.05

At initial G1/G2=0.5, there is a marked asymmetry

and at G1/G2=1.5, for retrograde orbits, the angle between the orbital planes may increase markedly with diminishing G1

Present-day Algol post large-scale mass-exchange, how did it start out?
• Trials with rapid binary evolution code (BSE, •
Hurley et al 2002) gives typical start parameters M1=3.26, M2=1.14, P=1.95 days With reasonable k(2) stellar structure constants, the apsidal motion from the tidal deformation dominates over the 3-body effects, and there were no high-e Kozai cycles

Typical rapid+slow mass-transfer

Period (3.17 d) a bit too long (real Algol 2.87)

With k(2)=0 (mass-point case) at j=95 deg, there are extreme Kozai cycles with e reaching 0.999

With realistic k(2) (around 0.005), the apsidal motion is fast and the eccentricity stays low

To show the transition from 3-body dominated (mass-point) motion to deformation-dominated, one may make numerical integrations with arbitrary (unphysical) k(2)-values.

k(2) 0 0.00001 0.00003 0.0001 0.0002 0.0004 0.0008 0.00100 0.00112 0.00125 0.00158 0.002 0.005

P_e 0.38e6 0.43e6 0.42e6 0.43e6 0.43e6 0.44e6 0.49e6 0.68e6 0.7e6:: 0.33e6 0.10e6 0.066e6 0.022e6

e_max 0.998 0.83 0.78 0.69 0.62 0.51 0.31 0.17 0.010 0.0025 0.0015 0.0012 0.0011

P_omg 0.70e6(circ) 0.86e6(circ) 0.85e6(circ) 0.85e6(circ) 0.83e6(circ) 0.86e6(circ) 1e6: (circ/lib) Irr (circ/lib) Irr (circ/lib) 0.36e6(circ/lib) 0.20e6 (circ) 0.13e6(circ) 0.044e6(circ)

P_aps(D) 21.9e6 6.92e6 2.19e6 1.09e6 0.55e6 0.27e6 0.22e6 0.20e6 0.18e6 0.14e6 0.11e6 0.044e6

The ’transition’ is rapid: red curve for k(2)=0.00100, blue k(2)=0.00112

With large-e episodes, the close orbit may be shrunk by tidal dissipation, and it is not unrealistic to start with e.g. Pc=4 days. Now there are Kozai cycles with physically realistic k(2), but the transition to D-dominated motion is very similar to the 1.95 day case.

k(2) 0 0.00125 0.0025 0.005 0.01 0.02 0.04 0.045 0.05 0.055 0.06 0.08 0.16

P_e 94000 90500 89400 89000 88600 92000 113500 135000 2e5 irr 56000 38000 20000 7800

e_max 0.96 0.78 0.74 0.68 0.61 0.48 0.27 0.19 0.08 0.0135 0.0116 0.0106 0.0102

P_omg 94000(lib) 90500(lib) 89400(lib) 89000(lib) 88600(lib) 92000(lib) 113500(lib) lib/circ lib/circ 112000 76000 40000 15600

P_aps(D) 1.90E+06 9.40E+05 4.70E+05 2.40E+05 118000 59000 52000 47000 43000 39000 29000 14500

At realistic k(2)=0.005, we now still have Kozai cycles

with nice libration of the apsidal line

and only at 10 times higher values we get the final rapid transition: k(2)=0.045(red), 0.050(green), 0.055(blue)

At large k(2) (=0.08), omega(green) circulates regularly, and the eccentricity (x100, red) is small

With large-e episodes, the close orbit may be shrunk by tidal dissipation
• Simple Kisseleva et al (1998) model,
parametrized by (theoretically poorly known) lambda • Runs with different lambda show final outcome to be very insensitive to strength of tidal friction

The eccentricity decreases with a speed proportional to lambda (1e-4, 1e-5, 1e-6)

a diminishes faster than e, to a value almost independent of lambda (P=4d -> 1.9 d)

Diminishing ac, means diminsihing G1 angular momentum, and we may use the simple original formulae to estimate the change in relative inclination

Parameters correspond to G1/G2=0.11, with 95-deg curve largely horizontal, thus little j-change expected

as is also the result of the numerical integrations

Preliminary results for larger initial Pc
• The large Kozai-cycles make for even
more effective orbit-shrinkage, but counter-intuitively with larger effect for smaller lambda • The smaller the lambda, the larger is the orbit-shrinkage

Final vs init Pc for different lambda
1e-10(lowest), 1e-9, 1e-8 (black), 1e-7, 1e-6(blue), 1e-5,1e-4(red)

Thus, for ’primordial Algol’:
• Original period may have been anything between 2
and 50(?) days, with a relative inclination close to 90 deg • The high eccentricity in the Kozai cycles shrunk the orbit by tidal dissipation to around 2 days, where rapid apsidal motion stabilized the eccentricity close to zero • The relative inclination did not change a lot in this phase, and the close to 90 deg value is a coincidence • (Then, there has been large-scale mass-transfer, producing the present system, still with circular close orbit and fixed relative inclination)

Examples with larger change of the relative inclination
• G1/G2 increased by decreasing M3 and/or the • • •
outer period Still 4d close orbit with preAlgol masses Model A: M3=0.2, Pw=275 d, start j=100 deg Model B: M3=0.1, Pw=175 d, start j=110 deg

Model A(1): qualitatively similar e-decrease

Model A(2): a decreases to same value independent of lambda (a/a0~0.69)

Model A(3): The relative inclination diminishes(=increases)

Model A(4): close agreement with the theoretical value expected at G1/G2=1.07

Model B(1): again similar e-decrease

Model B(2): a-decrease to 0.69 a0

Model B(3): relative inclination (by chance) stopping close to 90 deg

Model B(4): ..following the theoretical prediction for G1/G2=2.36

• Changes in the relative inclination needs a shrinking • • • • •
close orbit Tidal shrinkage of the close orbits needs Kozai cycles, that is, not too short initial Pc and an initial j>40 deg The inclination-change is greatest if the third star has small mass (G1>G2) If j<90 originally, the angle between the orbit planes decreases For retrograde orbits, the angle between the orbit planes may increase There is no special significance of a 90 deg relative inclination