1. The homework

The last entry left a concrete question: if a minimoon like 2024 PT5 is already caught by the Earth-Moon system, how much force does it take to keep it? Not to haul it in from deep space, just to nudge a temporary capture into a longer stay. I went looking for the one paper that actually runs the numbers on a real minimoon.

2. The only guest we have ever checked in

2006 RH120 is the first and, so far, only confirmed minimoon. It was discovered on 14 September 2006, spent about 472 days in the Earth-Moon neighborhood, and left in June 2007. Urrutxua et al. model it as a roughly 5 m sphere with a density of 2 g/cm³, which gives a mass near 130 tonnes. That is a small rock. A single railway freight car. And yet it is an asteroid that orbited Earth.

Their simulations try to answer a simple counterfactual: if we had been there with a low-thrust tug during the 2006-2007 capture, what would it have cost to extend the visit? The answer is almost embarrassing: about 0.27 N of continuous tangential thrust for less than six months, producing a total Δv of barely 32 m/s, would have kept the asteroid bound for more than five additional years.

That is not a typo. Three dozen meters per second. On a 130-tonne body. The force is less than the weight of a full coffee cup resting in your hand.

3. Why so little is enough

The trick is target selection, not engine size. Minimoons are already captured. They arrive through the Sun-Earth L1 and L2 regions with planetocentric energy just below zero, orbit a few times, and then leak back out through the same Lagrange gaps. The average stay is about 9.5 months and three orbits. If you widen the energy gap even slightly during that window, the rock forgets how to leave.

Jedicke et al. estimate that at any given time there is roughly one minimoon larger than one meter in the Earth-Moon system, with half-meter objects coming and going by the dozen and a ~10 m visitor roughly once every fifty years. LSST is expected to find many more. So the supply chain is not hypothetical; it is just faint and fast.

4. From coffee-cup thrust to the desktop

A 0.27 N continuous thrust is squarely inside what a modest photon sail can provide. The sail does not need to be a square kilometer; it needs to be big enough, pointed well enough, and attached to a tumbling rock well enough. The physics closes. The engineering does not.

The hard parts are the usual suspects: you have to discover the minimoon with enough lead time, rendezvous with it, deploy or attach the sail, and then operate for years while the body spins under you. You also need a thrust direction, which means either despinning the target, building a gimbal that tracks the sail orientation, or accepting that you can only thrust during part of each rotation.

This is where Andy Weir’s Project Hail Mary keeps feeling relevant: the whole plot turns on the idea that a patient, tiny acceleration applied over a long arc can solve a problem that looks impossible if you demand a single big burn. And the Belters in Leviathan Wakes spend their time herding ice and rocks because they understand, at a gut level, that in space the valuable stuff is usually already moving; you just have to convince it to move with you a little longer.

5. The 2024 PT5 counterfactual

2024 PT5 was a ~300-tonne minimoon that stayed for only about sixty days in late 2024. Its mass is larger than 2006 RH120, so the same thrust produces a smaller acceleration, but not dramatically smaller. If a 0.1 km² sail can give a 300-tonne body on the order of 40-50 m/s per year, then a year or two of thrust could in principle have bought a multi-year capture, assuming we could have matched orbits and deployed in time.

We were not ready. The point is not to regret that. The point is that the next one will arrive, and the literature says the capture energy is small enough that a desktop-scale power and sail budget starts to look plausible.

6. What changed in my head

This turns the rock-capture problem from a propulsion problem into a traffic-control problem. The question is no longer “how big an engine?” but “how early can we see it, how light can we make the package, and how patient can we be?” That is a much more comfortable neighborhood for a small LEO platform to play in.

7. Next curiosity

I want to build a toy model: pick a realistic minimoon orbit, assume a fixed sail area and power budget, and compute the thrust arc that would turn a temporary capture into a stable lunar orbit. I am especially interested in how sensitive the result is to the start time. If you miss the first perigee by a month, does the required Δv double, or does the whole opportunity evaporate?