Lagrange Points

The restricted three-body setup

Picture two bodies — say, the Sun and Earth — orbiting their common center of mass at angular velocity Ω = 2π/(1 year) = 1.99 × 10⁻⁷ rad/s. Drop a third, much smaller body into this system: a satellite, a dust grain, an asteroid. Its motion in the inertial frame is a complicated dance because the gravitational field is not static. Switch instead to a reference frame that co-rotates with the two large bodies at angular velocity Ω. In this rotating frame the two big bodies are at rest, but two pseudo-forces appear: centrifugal Ω²r outward from the rotation axis, and Coriolis 2Ω × v perpendicular to the velocity.

An equilibrium of the small body in the rotating frame exists where the net force on it (gravity from both large bodies plus centrifugal pseudo-force) vanishes. The Coriolis force does not contribute because v = 0 at equilibrium. Joseph-Louis Lagrange showed in 1772 that there are exactly five such equilibria, and they form the famous L1–L5 configuration.

The effective potential

In the rotating frame, the small body's motion follows from an effective potential that combines gravity and the centrifugal pseudo-force:

U_eff(r) = −G·m₁/|r − r₁| − G·m₂/|r − r₂| − ½Ω²(x² + y²)

where r₁ and r₂ are the positions of the two large bodies, r is the small body's position, and the rotation axis is z. The first two terms are the usual Newton gravity; the third is the centrifugal "potential" associated with the rotating frame (note the sign — it lowers U away from the axis).

Equilibria are points where ∇U_eff = 0. Five such points exist:

• L1, L2, L3 — collinear with the two masses. L1 sits between them, L2 beyond the smaller mass, L3 on the far side of the larger. All three are saddle points of U_eff: gradient zero, but the second-derivative matrix has one negative eigenvalue. Small perturbations along the unstable direction grow exponentially.
• L4, L5 — apex of the two equilateral triangles built on the line between the masses. L4 leads the smaller mass by 60° in its orbit; L5 trails by 60°. Both are local maxima of U_eff, which sounds unstable, but the Coriolis force in the rotating frame turns small displacements into closed loops when the mass ratio is small enough.

Worked example: where exactly is Sun-Earth L1?

Define the dimensionless mass ratio µ = m_E / (m_S + m_E). For Sun (m_S = 1.989 × 10³⁰ kg) and Earth (m_E = 5.972 × 10²⁴ kg), µ = 3.003 × 10⁻⁶. Place coordinates with the Sun-Earth distance R = 1 AU = 1.496 × 10¹¹ m as length unit and angular frequency Ω as time unit. The collinear equilibria satisfy a quintic equation

x − (1−µ)·(x+µ)/|x+µ|³ − µ·(x−1+µ)/|x−1+µ|³ = 0

where x is the dimensionless coordinate along the Sun-Earth line. For L1 we look for the root just inside Earth's orbit. For µ ≪ 1 the asymptotic solution is

x_L1 ≈ 1 − (µ/3)^(1/3) · [1 + small corrections]

So L1 is a distance r ≈ R · (µ/3)^(1/3) inside Earth's orbit. Plug in µ = 3.003 × 10⁻⁶:

(µ/3)^(1/3) = (1.001 × 10⁻⁶)^(1/3) ≈ 1.0003 × 10⁻²
r_L1 ≈ 1.496 × 10¹¹ × 0.01000 ≈ 1.496 × 10⁹ m ≈ 1.50 million km

About 1.5 million km Sunward of Earth. The same formula gives L2 ≈ 1.50 million km on the opposite side. This is roughly four times the Earth-Moon distance and about 1% of the Sun-Earth distance. The Hill sphere of Earth has radius r_H = R(µ/3)^(1/3) too — L1 and L2 sit at the Hill-sphere radius along the radial line.

For NASA's SOHO observatory at Sun-Earth L1, this distance is exactly right: SOHO sees the Sun continuously without ever being eclipsed by Earth, and is positioned to give Earth ~1 hour of warning for any solar wind disturbance traveling at ~400 km/s.

The five points compared

Point	Location	Stability	Distance from Earth (Sun-Earth system)	Famous occupant
L1	Between Sun and Earth	Saddle (unstable)	1.50 × 10⁶ km Sunward	SOHO, ACE, DSCOVR
L2	Beyond Earth from Sun	Saddle (unstable)	1.50 × 10⁶ km anti-Sunward	JWST, Gaia, Euclid, Planck (retired)
L3	Behind Sun from Earth	Saddle (unstable)	~2 AU (always hidden)	None (no spacecraft)
L4	60° ahead of Earth in orbit	Stable for µ < 0.0385	~1 AU around the orbit	None for Earth (asteroid 2010 TK7 close)
L5	60° behind Earth in orbit	Stable for µ < 0.0385	~1 AU around the orbit	None for Earth
Sun-Jupiter L4/L5	60° ahead and behind Jupiter	Stable	—	~12,000 Jupiter Trojans

The five Sun-Earth points are by far the most economically important; the Sun-Jupiter L4/L5 swarms host the largest natural population of Lagrange-resident objects. Earth-Moon Lagrange points (separate system) are also accessible and have been considered for crewed deep-space outposts.

L4/L5 stability and the 1/24.96 bound

L4 and L5 are local maxima of U_eff — gravitationally they are uphill in every horizontal direction. Yet the rotating frame's Coriolis force can stabilize them. Linearizing the equations of motion around L4 in the rotating frame and demanding all eigenvalues to be purely imaginary gives a polynomial in µ whose discriminant changes sign at

µ_critical = ½(1 − √(23/27)) ≈ 0.03852  ≈  1/24.96

For mass ratios µ < 0.03852 the two large bodies are different enough that L4/L5 are stable; for µ > 0.03852 they are unstable like L1/L2/L3. Some examples:

• Sun-Jupiter: µ = 9.5 × 10⁻⁴ ≪ 0.03852 → L4/L5 stable. Result: ~12,000 Jupiter Trojans.
• Sun-Earth: µ = 3.0 × 10⁻⁶ → L4/L5 stable. Result: just one confirmed Earth Trojan asteroid (2010 TK7) so far.
• Earth-Moon: µ = 0.0123 → L4/L5 stable. Result: dust clouds at L4/L5 (Kordylewski clouds), confirmed in 2018.
• Pluto-Charon: µ = 0.103 > 0.03852 → L4/L5 unstable. No Trojan moons of Charon, consistent with theory.

Around stable L4/L5 the orbits are tadpole (small libration around the point) or horseshoe (larger libration that loops around L3 to the other side of the orbit). The asteroid 3753 Cruithne is a famous horseshoe of Earth.

Halo and Lissajous orbits

An unstable Lagrange point cannot host a stationary spacecraft for more than weeks before runaway sets in. Mission designers solve this by placing the spacecraft on a periodic halo orbit or quasi-periodic Lissajous orbit around the unstable point. These exist as a one-parameter family of solutions to the linearized equations: by choosing the right initial velocity perpendicular to the Sun-Earth line, you can put the spacecraft on a closed three-dimensional curve that loops around L1 or L2 with amplitudes from ~100,000 km up to ~1 million km perpendicular to the Sun-Earth line.

JWST orbits a halo of about ±800,000 km in the y-direction (perpendicular to Sun-Earth line in the ecliptic) and ±450,000 km in z (out of ecliptic), with a period of 6 months. The halo is chosen so JWST is always in Earth's solar shadow penumbra (no direct sunlight on its sunshade-shaded side), and Earth is never in the science instruments' field of view. Station-keeping requires ~2 m/s of Δv per year — over 20 years, that is ~40 m/s, well within the fuel reserves the launch precision left.

Where Lagrange points show up

• JWST at Sun-Earth L2 (operational since 2022). The James Webb Space Telescope sits in a halo orbit at L2, 1.5 million km from Earth on the anti-Sun side. The halo orbit has Y-amplitude ±800,000 km and a 6-month period. Sun, Earth, and Moon are all on the Sun-side of the giant 21.2 m × 14.2 m sunshield, allowing the primary mirror to cool below 50 K passively. Station-keeping is performed roughly every 21 days at 0.1 m/s/burn; the original 10-year mission floor has been revised upward to 20+ years thanks to launch-trajectory accuracy.
• SOHO, ACE, and DSCOVR at Sun-Earth L1. Three solar-monitoring spacecraft sit ~1.5 million km Sunward of Earth, providing 30-60 minutes of warning for incoming solar wind and CMEs (coronal mass ejections). DSCOVR is responsible for the operational space-weather alerts NOAA issues to power-grid operators; SOHO has been continuously imaging the Sun since 1995 with extension after extension, having long outlived its 2-year design life.
• Gaia at Sun-Earth L2 (2014–2025). The European Space Agency's Gaia astrometric mission produced a catalog of 1.8 billion stars with positions and distances measured to microarcsecond precision. Its halo orbit at L2 gave it a stable thermal environment and uninterrupted view of the celestial sphere. Final Gaia data release (DR4) is scheduled for 2026; DR5 with full mission data for 2030.
• Jupiter Trojan asteroids at Sun-Jupiter L4 and L5. Two swarms of ~12,000 known asteroids share Jupiter's orbit. The L4 swarm (the "Greek camp") leads Jupiter by 60°; the L5 swarm (the "Trojan camp") trails by 60°. NASA's Lucy mission, launched 2021, will visit eight Trojan asteroids between 2025 and 2033. The largest, 624 Hektor, has dimensions ~370 × 200 km.
• Earth-Moon L1 Lunar Gateway (under development). NASA's planned Gateway space station orbits Earth-Moon L2 in a Near-Rectilinear Halo Orbit, with launch of the first elements (HALO + PPE) scheduled for 2027. The orbit has perilune ~3000 km and apolune ~70,000 km; the long apolune-side residence allows ~6.5-day crew access from Earth via Orion. Gateway is the staging point for Artemis crewed lunar surface missions.

Variants and extensions

• Elliptic restricted three-body problem. Drop the assumption that the two large bodies move on circles. The Lagrange points become time-varying solutions; their structure depends on the eccentricity of the binary orbit. Important for Sun-Mars (e = 0.093) and any binary star system.
• Solar-radiation-pressure-perturbed Lagrange points. Light pressure on a spacecraft acts as an effective reduction in solar gravity, shifting L1 and L2 closer to or further from Earth depending on the area-to-mass ratio. The DSCOVR spacecraft has its operational L1 set 1.4 million km from Earth instead of 1.5 million km because of solar-pressure perturbations on its instrument panels.
• Hill's problem. The limit of small mass ratio (µ → 0) where one keeps only leading-order terms in µ. Universal in scaling — every planetary Lagrange-point structure follows the same dimensionless equations.
• Weak-stability boundary transfers. Mission-design technique exploiting the fact that the unstable manifolds of L1 and L2 in a Sun-planet system extend out to capture and ballistic transfer trajectories. Hiten (1991) and Genesis (2001) used these "low-energy" tubes for fuel savings of 30–50% over Hohmann transfers.
• Co-orbital configurations. Tadpole and horseshoe orbits around L4 and L5 are part of a richer family of co-orbital phenomena including quasi-satellites and exchange orbits. Earth has at least 8 known co-orbital companions, including 3753 Cruithne (horseshoe) and 469219 Kamoʻoalewa (quasi-satellite, the proposed origin of the Yarkovsky-launched lunar fragment).
• Generalized Lagrange points in N-body systems. In the Sun-Earth-Moon-Jupiter system there is no exact Lagrange-point geometry, but quasi-equilibria persist where solutions are anchored by overlapping stability of the various two-body sub-problems. Numerical exploration is the modern tool; the analytical five-point picture is a good starting point.

Common pitfalls

• Treating "L1 is where gravities cancel" as the definition. L1 is where gravities and centrifugal pseudo-force balance in the rotating frame. The pure gravity-cancellation point is much closer to Earth (~260,000 km) and is not an equilibrium because of the rotating frame.
• Assuming L4/L5 stability always. The 1/24.96 bound matters. Sun-Jupiter L4/L5 (µ = 0.001) is stable; Pluto-Charon L4/L5 (µ = 0.10) is unstable. Always check the mass ratio before claiming a generic Lagrange point can hold dust or spacecraft.
• Forgetting that L1/L2/L3 require active station-keeping. Missions that ignore this fact run out of station-keeping fuel and tumble out of the Lagrange neighborhood within months. Lifetime budgets are dominated by Δv required, not by instrument or thermal lifetime.
• Confusing the two-body Lagrange points with multi-body extensions. Every two-body system has its own L1–L5. Sun-Earth, Sun-Jupiter, Earth-Moon, Saturn-Titan are all different. A spacecraft "at L2" must be qualified — at L2 of which pair?
• Ignoring the Coriolis force when explaining L4/L5. Without Coriolis, L4 and L5 would be simple maxima of effective potential and unstable like a marble on a hilltop. With Coriolis, displacements get bent perpendicular to velocity, turning hill-roll-off into closed loops. Ten textbooks miss this; the Coriolis force is the entire reason Trojan asteroids exist.