Mars Orbit, Phobos & Deimos

A node, not a destination

Why Mars Orbit Is an Architecture Decision, Not a Place to Visit

Nobody goes to Mars orbit for its own sake. The question is where the vehicle that brought the crew waits while they are on the ground.

Two architectures split here. Under direct-to-surface, the interplanetary vehicle lands, refuels from a surface plant, and flies home — there is no node. Under the orbit-staged ferry, the transit habitat parks in the Mars system, keeps its shielding and its Earth-return propellant, and a small ferry carries crew down and back up. The second architecture needs 107 t of locally made propellant against 683 t for the first, and it needs a place for the vehicle to sit.

Three candidates: low Mars orbit, Phobos, and Deimos. The published field has argued about them on energy grounds for 30 years, and the energy argument turns out to be nearly empty. What the node actually has to supply is bulk shielding mass for a habitat that will sit there for years, a stable position that needs no propellant to hold, and a geometry that lets a crew in orbit supervise machines on the ground. Those three requirements point in different directions, and the data to settle them arrives in 2029 for one moon and never for the other.

Departure is flat

Trans-Earth Injection Δv from Low Orbit, Phobos and Deimos

The Oberth effect at low orbit almost exactly cancels the deeper gravity well. Node altitude is nearly free on the way out.

Departure Δv from the three candidate nodes differs by a few hundred meters per second, and the ranking even reverses with departure energy: low orbit is worse at low C3 and better at high C3, with the crossover around C3 13–15. Ascent from the surface is the opposite — a monotonic penalty for every kilometer of altitude, and a large one.

Read down the ascent row and the conclusion is immediate: node altitude costs 1.2 km/s to reach Phobos and 1.7 km/s to reach Deimos, every single sortie, and returns almost nothing on departure. The energy argument says put the node as low as its shielding and stability functions permit — which is an argument against both moons, and means the case for a moon has to be made on mass, volatiles and dwell time rather than on Δv.

Entry speed is a second altitude term that the field usually misses. A ferry descending from low Mars orbit meets the atmosphere at 3,612 m/s; from Phobos it arrives 25–40% faster, and convective heat flux scales roughly as the cube of velocity. The low node sees about a sixth of the thermal load, which is the difference between reusable metallic tiles and an ablative shield that must be replaced or manufactured.

The two moons

Phobos and Deimos: Two Captured Bodies, Two Different Jobs

Neither is a world. Both are low-density rubble with escape velocities a person could jump past.

The low densities are the interesting number: both bodies are 25–35% void, meaning either rubble-pile structure, substantial ice, or both. Which one decides whether a moon is a propellant source or only a sandbag. Phobos carries the 9 km Stickney crater, a scar from an impact that nearly destroyed it, and orbits below areostationary altitude — it circles Mars three times a sol and rises in the west. Its orbit is decaying, and in roughly 30 to 50 million years it will break up inside the Roche limit.

Deimos’s advantage is geometric and has nothing to do with composition. A 30-hour period against Phobos’s 7.6 means far longer continuous dwell over any given surface site — better teleoperation continuity and better comms for a node whose crew is supervising machines below. Phobos’s advantage is that it is closer, cheaper to reach from the surface, and about to have ground truth.

Anchoring

Working on a Body Where Escape Velocity Is 11 Meters Per Second

A 20 t machine weighs 114 newtons on Phobos. Drilling pushes back harder than the machine weighs.

Every operation on these moons is an anchoring problem. Coring at a modest 500 N for 10 seconds imparts about 5 m/s to a 20 t machine — roughly half Phobos escape velocity, from one hole. A torque of 50 N·m rotates a 1 t vehicle through 90° in 5.6 seconds. Reaction force exceeds weight by orders of magnitude, so the machine must be tied to the ground before it does anything, and tying it to the ground requires knowing the regolith’s strength, which nobody does.

The one in-situ datum is Philae, which bounced at comet 67P in November 2014 when its harpoons failed to fire, and came to rest in shadow. Hayabusa2 and OSIRIS-REx both avoided the problem rather than solving it, touching and retreating within seconds. No spacecraft has ever anchored to a small body and held.

There is a second, less obvious constraint on any excavation here. Material thrown at more than 11.3 m/s does not fall back — it enters Mars orbit at nearly the moon’s own orbital velocity and stays there, forming a debris torus along the path any vehicle parked at Phobos must fly. Open-pit mining at a node you also live at is self-defeating. Closed borehole excavation, with cuttings captured rather than ejected, is a precondition for the whole Phobos architecture rather than a refinement of it — and it does not exist as hardware anywhere.

MMX

Martian Moons eXploration: What Arrives in 2029, and What Never Does

Phobos gets ground truth. Deimos gets flybys. The asymmetry decides which moon can be evaluated at all.

JAXA’s Martian Moons eXploration mission launches on an H3 from Tanegashima, arrives in the Mars system in late 2027, enters a quasi-satellite orbit at Phobos, and lands in 2029 to collect at least 10 g of regolith, returning it to Earth in 2031. It carries the CNES and DLR IDEFIX rover, the first wheeled vehicle to operate on a body this small, and the MEGANE gamma-ray and neutron spectrometer, which measures elemental composition including hydrogen — the water answer.

After MMX, Phobos will have bulk composition, water content, regolith mechanics, surface strength and thermal behavior: everything needed to evaluate it as a node, a shield source and a propellant source, plus the substrate data the anchoring question has been waiting on since the 1990s. Deimos gets flybys on the outbound spiral and nothing else — better imagery and spectra, which say essentially nothing about regolith strength, anchoring, or extractable water.

Two consequences worth stating plainly. The Phobos water answer arrives in 2029 at the earliest and the definitive one in 2031, so any architecture committing before then is committing without it. And the Deimos characterization gap — a small orbiter or hosted payload returning gravity, thermal inertia and neutron-spectrometer data — is a named, fundable gap that appears on nobody’s roadmap at any agency.

Low Mars orbit

The Case for the Lowest Node the Mission Allows

Two penalties usually charged against low orbit turn out to be small, and one advantage nobody counts turns out to be large.

Low Mars orbit is routinely dismissed on two grounds, and both are weaker than stated. Station-keeping at a 1-sol periapsis orbit costs on the order of 0.2 m/s per year — noise against any propellant budget on this page. And the claim that Mars blocks the sky for a parked habitat cuts the other way: the planet occults 27.7% of the sky from low orbit against 3.4% from Phobos’s radius, which is 27.7% of the radiation environment removed for free by a body that also removes part of the solar flux. A shield you do not have to mine.

Against that, the moons offer something low orbit cannot: bulk mass in place. A transit habitat parked for years wants meters of shielding, and hauling that mass up from the surface at 4.10 km/s is absurd while a rubble pile sits at the node. That is the entire case for a moon, and it rests on regolith that can be excavated without ejecting a debris torus, which is the unresolved item.

One further consideration pushes toward low orbit and was derived from an unrelated argument. If Mars ends up with two settlements chosen for decorrelation rather than for score, they will sit at different latitudes — and a moon node is inclination-locked within about 1° of the equator. Serving a latitude-separated pair from a moon costs a plane change of order 0.7–1.0 km/s per sortie, both directions, every sortie. Low Mars orbit has no such lock, so the two-settlement requirement pushes the node down, or pushes the program toward two nodes.

The ferry

The Ferry Ledger: Propellant, Power and the Aeroshell Cap

A ferry beats landing the interplanetary vehicle at every crew size, and by more as the crew grows — but the seat count is capped by aeroshell diameter, not by propellant.

The power column is the finding. A ferry moves the surface reactor requirement from “invent a megawatt-class space fission plant” to “use the 40 kWe hardware already under contract,” and it does it by never lifting Earth-return Δv off the surface in the first place.

The seat count has a separate ceiling. At 50 seats a ferry’s entry mass reaches 58.8 t, which needs roughly a 14 m decelerator at a ballistic coefficient of 250 — and the largest rigid aeroshell that fits inside a Starship is about 9 m, capping the vehicle near 10 seats at that ballistic coefficient and about 20 at Starship-class terminal velocity. Bigger ferries also make evacuation lumpier: a 20-seat ferry burns 69.3 t carrying 20 people and 60.1 t carrying 10, because the mass is dry-mass dominated, so a 100 t propellant farm funds one sortie and evacuates 20 rather than the 29 a linear reading suggests. The optimum is two mid-size ferries, which is also the redundancy answer.

Relay and areostationary

Mars Orbit as Communications Infrastructure

Every surface mission since 2004 has talked to Earth through somebody else’s orbiter.

Mars orbit already carries an operating relay network, assembled by accident from science orbiters: NASA’s Mars Odyssey, in orbit since 2001 and the longest-serving spacecraft at another planet; Mars Reconnaissance Orbiter; MAVEN; ESA’s Mars Express and Trace Gas Orbiter, which relayed for NASA’s rovers under a standing cross-support agreement; and China’s Tianwen-1 orbiter, which relayed for the Zhurong rover and continues its own science mapping. The relay function is a side job on all of them, and the fleet is aging.

An areostationary orbit exists at 17,032 km altitude, where the period matches the 24 h 37 m sol and a satellite holds station over one longitude — the Martian equivalent of geostationary, and the obvious home for a purpose-built relay and navigation constellation. It has never been used. A crewed program brings continuous coverage requirements that no science-orbiter side job can meet, and surface positioning has no equivalent of GPS at all.

None of this removes the light-time floor: 6 to 44 minutes each way, whatever the orbital architecture. Relay improves availability and bandwidth, never latency. The deep-space side of the link — the Deep Space Network, ESTRACK and China’s deep-space stations — is system-level infrastructure serving every regime and is covered under Substrates rather than here.

Record

Provenance and Data Sources