Spotlight #69: Orbital Mechanics — Three-Body Problem, Lagrange Points & Roche Lobes

I. The Restricted Three-Body Problem in a Co-Rotating Frame

The full three-body problem — three point masses interacting through Newtonian gravity — has no closed-form solution in general. Henri Poincaré showed in 1890 that the motion is generically chaotic: nearby trajectories diverge exponentially, making long-term prediction impossible in principle. But one special case turns out to be exactly tractable and immensely useful: the restricted three-body problem (R3BP), in which two massive bodies (the primaries, mass M and m with M ≫ m) orbit their common centre of mass on circular orbits, and a third body of negligible mass moves in their gravitational field.

The key simplification is to switch to a co-rotating (synodic) reference frame that rotates with the same angular velocity ω as the primaries. In this frame, the primaries are stationary. A massless test particle experiences not only the gravitational attraction of both primaries, but also two fictitious forces that arise from the rotation: the centrifugal force (pushing outward from the rotation axis) and the Coriolis force (deflecting moving particles perpendicular to their velocity).

The equations of motion for the test particle in this frame are:

x'' - 2*omega*y' = dU*/dx
y'' + 2*omega*x' = dU*/dy

where the effective potential U* combines gravity and centrifugal terms:

U*(x, y) = -G*M/r1 - G*m/r2 - 0.5*omega^2*(x^2 + y^2)

r1 = distance from test particle to primary M
r2 = distance from test particle to primary m
omega = sqrt(G*(M + m) / a^3)    (orbital angular velocity, Kepler's 3rd law)
a    = separation between the primaries

The Coriolis terms (-2ωy' and 2ωx') on the left-hand side couple the x and y equations, making the system non-conservative in the usual sense. However, there is one conserved quantity: the Jacobi integral, often written C, which plays the role of total energy in the rotating frame.

C = 2*U*(x, y) - v^2

where v^2 = x'^2 + y'^2 is the speed in the rotating frame.

A particle at rest (v = 0) satisfies C = 2*U*(x, y).
This defines zero-velocity curves: contours of U* equal to C/2.
A particle cannot cross a zero-velocity curve (it would require v^2 < 0).

Zero-velocity curves (also called Hill curves) are the boundaries of the regions a particle can reach for a given value of C. High C values confine the particle tightly near one primary; as C decreases, the forbidden regions shrink and new regions open up, eventually allowing the particle to circulate freely around both primaries.

II. The Five Lagrange Equilibrium Points

Equilibrium points of the effective potential — where the gradient of U* vanishes and hence a stationary particle experiences zero net force — are the Lagrange points. There are exactly five, labelled L1 through L5. Joseph-Louis Lagrange found L4 and L5 analytically in 1772; the collinear points L1–L3 had been found earlier by Euler.

Collinear points: L1, L2, L3

These three points lie on the line connecting the two primaries. All three are saddle points of U*: stable along the line of primaries, unstable perpendicular to it. Because they sit at saddle points, any small perturbation sends a particle drifting away exponentially — they are Lyapunov unstable. Spacecraft stationed at L1 or L2 require periodic station-keeping manoeuvres every few weeks to remain in position.

L1: between the primaries, closer to the smaller mass m.
    Approximate position (mass ratio mu = m/(M+m) << 1):
    x_L1 ~ a * (1 - (mu/3)^(1/3))    from the larger primary

L2: on the far side of m from M.
    x_L2 ~ a * (1 + (mu/3)^(1/3))

L3: on the far side of M from m (nearly opposite m).
    x_L3 ~ -a * (1 + 5*mu/12)

    All three: unstable (saddle points of U*).
    Characteristic timescale of instability ~ orbital period / (3*mu)^(1/3)

L1 is the gateway between the two gravitational wells: matter flowing through L1 (in a binary star system, for instance) is transferring from the sphere of influence of one body to the other. This is precisely how mass transfer works in close binary stars and cataclysmic variables. L2, located on the opposite side of the smaller primary, is where the James Webb Space Telescope resides — more on this below.

Triangular points: L4 and L5

L4 and L5 sit at the vertices of equilateral triangles formed with the two primaries. They are maxima of U* (local energy hills) rather than minima, which naively suggests instability — but the Coriolis force saves them. When the mass ratio satisfies M/m > 24.96 (approximately), the Coriolis deflection of any small perturbation creates a closed epicyclic orbit around L4 or L5 rather than an escape trajectory. Both the Sun–Earth and Sun–Jupiter systems satisfy this criterion comfortably.

L4: 60 degrees ahead of the smaller primary m in its orbit
L5: 60 degrees behind m

Stability criterion (Routh's condition):
    27 * mu * (1 - mu) < 1
    i.e. mu < 0.0385 (mass fraction of smaller primary)

    Sun-Jupiter: mu ~ 0.00095   --> stable
    Earth-Moon:  mu ~ 0.012     --> stable
    Pluto-Charon: mu ~ 0.11     --> UNSTABLE (no Trojans observed)

The stability of L4 and L5 is remarkable: these are not minima of the potential but the Coriolis force creates what is effectively a restoring force for small displacements. Particles trapped near L4/L5 execute slow epicyclic (tadpole) orbits around the equilibrium point with a period roughly 1/sqrt(27*mu) times the primary orbital period.

III. Roche Lobes, Mass Transfer & Zero-Velocity Curves in Practice

The Roche lobe of each primary is the region of space within which material is gravitationally bound to that primary rather than to the other. It is defined as the region enclosed by the zero-velocity curve that passes through the inner Lagrange point L1. The teardrop-shaped boundary that runs through L1 is called the Roche surface.

In binary star systems, the Roche lobe concept is central to understanding mass transfer. If one star expands (as in the red giant phase) until it fills its Roche lobe, matter at the L1 point feels equal gravitational attraction from both stars and begins to flow across to the companion. This process powers some of the most energetic phenomena in the Universe: novae, X-ray binaries, and Type Ia supernovae.

Roche lobe radius (Eggleton 1983 approximation):
    R_L / a = 0.49 * q^(2/3) / (0.6 * q^(2/3) + ln(1 + q^(1/3)))

    q = m / M  (mass ratio, secondary to primary)
    a = orbital separation

Example: Sun-like star + white dwarf, a = 1 solar radius
    q ~ 0.5 --> R_L ~ 0.38 * a ~ 0.38 solar radii
    As the donor star evolves off the main sequence and swells,
    once R_star > R_L mass transfer begins, often runaway.

The shape of the zero-velocity curves changes dramatically as the Jacobi constant C decreases. For very large C, two separate closed ovals surround each primary (the Roche lobes) with a forbidden region between and outside them. As C drops to the value at L1, the inner forbidden region pinches off and the two Roche lobes touch at L1. For still lower C, the zero-velocity curve passes through L2 and L3 in turn, opening up pathways for material to escape the system entirely — a mechanism implicated in the formation of common-envelope binaries.

IV. Real-World Examples: JWST at L2 and the Trojan Asteroids

James Webb Space Telescope at Sun–Earth L2

The Sun–Earth L2 point sits approximately 1.5 million kilometres from Earth, directly away from the Sun. Because L2 co-rotates with Earth, a spacecraft there sees the Sun, Earth, and Moon all within a 50-degree cone on one side — the entire other hemisphere of sky is permanently in shadow. This makes L2 an ideal location for infrared astronomy: the telescope’s sunshield can simultaneously block heat from the Sun, Earth, and Moon with a single five-layer foil structure.

JWST does not sit exactly at L2 (the exact point offers no station-keeping advantage and would periodically be eclipsed by Earth). Instead it orbits L2 in a large halo orbit with an amplitude of roughly 500,000 km, tilted so that Earth and the Moon are always outside the sunshield keep-out zone. The orbital period is about 6 months. Station-keeping burns of around 2–4 m/s per year are required to counteract the instability of the collinear Lagrange point.

Trojan asteroids at Jupiter–Sun L4 and L5

The Sun–Jupiter system satisfies Routh’s stability criterion with a comfortable margin, and the result is spectacular: more than 12,000 known asteroids (the Jupiter Trojans) librate around L4 (the Greek camp, 60° ahead of Jupiter) and L5 (the Trojan camp, 60° behind). They are trapped in tadpole orbits around the equilibrium points, librating with amplitudes of up to 30 degrees and periods of decades.

NASA’s Lucy mission, launched in October 2021, is the first spacecraft to explore the Trojan swarms. It will fly past eight Trojan asteroids between 2027 and 2033, using gravitational assists from Earth three times to reach the necessary trajectory. The Trojans are thought to be primitive remnants from the early solar system, possibly captured during the period of giant-planet migration described by the Nice model — making them time capsules of the conditions 4.5 billion years ago.

Earth’s own L4 and L5 are not empty: the asteroid 2010 TK7 librates around Earth’s L4, and a handful of co-orbital companions have been found at L5. Mars has several Trojans as well. Even Saturn and Uranus host small populations, though Neptune’s are particularly numerous — over 30 known, and models suggest the true population rivals Jupiter’s in number.

Try It Yourself

Closing Thought

The restricted three-body problem is one of the most productive “unsolvable” problems in physics. Unable to write down general solutions, mathematicians instead mapped the structure of the solution space: the topology of zero-velocity curves, the stability of equilibrium points, the existence of periodic orbits and invariant tori. That structural knowledge is precisely what engineers exploit when they send JWST to L2 or Lucy to the Trojans. The lesson is broader than orbital mechanics: understanding why a system behaves as it does — which conserved quantities exist, which regions are accessible, which equilibria are stable — is often more powerful than any specific solution. The equations are the map; the physics is the territory.