Mars EDL: The Seven Minutes of Terror and the Engineering of Soft Landing

Mars has an atmosphere. This is both the fundamental advantage and the fundamental problem for landing there.

Earth’s atmosphere is dense enough that a heat shield and parachute can decelerate any spacecraft to near-zero velocity well above the surface. The Moon has no atmosphere at all — pure rocket-powered descent from orbital velocity. Mars sits in an uncomfortable middle ground. Its atmosphere is thick enough to cause significant aerodynamic heating during entry, requiring a heat shield. But it is far too thin — surface pressure approximately 600 Pa, less than 1% of Earth’s — to decelerate a spacecraft to a safe landing speed using aerodynamics alone.

At the parachute deployment point in a Mars entry, a spacecraft is still travelling at 400–500 m/s. The parachute, limited in size by the aeroshell diameter and in effectiveness by the thin Martian air, can only reduce velocity to approximately 80–100 m/s by the time the spacecraft is 1–2 km above the surface. That remaining velocity must be killed by rockets. The question is how.

This question has produced some of the most inventive engineering in the history of robotic exploration — and the answer that worked for a one-tonne rover has no obvious extension to the ten-tonne crewed landers that any serious Mars human mission architecture requires.

Key parameters

Parameter	Value
Mars atmospheric density at surface	~0.02 kg/m³ (1.4% of Earth)
Entry velocity (Curiosity)	5.9 km/s
Peak heating (Curiosity)	~215 W/cm²
MSL aeroshell diameter	4.5 m
Supersonic parachute diameter	21.5 m
Skycrane propellant (MMH/NTO)	~390 kg
Touchdown velocity	<0.75 m/s

The Atmosphere Problem in Numbers

Mars entry vehicle aerodynamics are governed by the same equations as any hypersonic entry (see atmospheric re-entry thermodynamics), with one critical difference: the destination is not the ground floor. Mars entry begins at approximately 125 km altitude; landing is on a surface 20–25 km below “sea level” in the thin atmosphere of Hellas Planitia, or at 0–3 km elevation for most exploration targets.

The entry vehicle enters at 5.5–6.5 km/s (Mars approach velocity minus Mars escape velocity of 5.03 km/s) at a flight path angle of approximately -15 to -16 degrees. Peak deceleration reaches 8–12 g during the entry corridor. Peak heating rate at the stagnation point for Perseverance: approximately 145 W/cm² — much lower than lunar return re-entry due to the lower approach velocity, but significant.

The aeroshell — a 70° sphere-cone geometry with a diameter of 4.5 m for Perseverance — uses PICA (Phenolic Impregnated Carbon Ablator) heat shield material, the same class of material developed for Stardust and used on Orion. The backshell is covered in SLA-561V (Super Light Ablator), a silicone-based material that handles the lower-heating backshell environment.

Mach 2 parachute deployment: 12 m diameter supersonic disk-gap-band parachute, generating approximately 65,000 N of drag. The parachute decelerates the system from approximately 420 m/s to 80–100 m/s over 1–2 km of altitude.

The Skycrane: An Engineering Absurdity That Works

The problem facing Curiosity’s designers in the mid-2000s was specific: a 900 kg rover had to be placed on the Martian surface without airbags (too heavy, too unreliable for the mass), without legged landers (unacceptably large footprint and retrorocket contamination of landing site), and with rockets that could not fire directly downward at landing (exhaust plumes would contaminate surface science samples and potentially destabilise the rover).

The solution, proposed by JPL engineers and initially viewed by many as implausibly complex, was the skycrane: a separate descent stage carrying eight retrorockets would slow to a hover at approximately 20 m altitude, then lower the rover on a 7.6 m bridle while all three systems — descent stage, bridle, and rover — descended together to the surface. At touchdown, pyrotechnic cutters severed the bridle and the descent stage flew away to crash at a safe distance.

The system demands that the rover’s wheels touch the surface, sensors detect the slack in the bridle indicating contact, trigger the cutters, and transmit this information to the descent stage before the slack propagates into catastrophic tangle — all within approximately 0.8 seconds.

It worked on 6 August 2012 at 05:17 UTC, placing Curiosity within 2.4 km of its targeted landing ellipse (itself 20 km × 7 km — a precision that would have been considered extraordinary for any previous Mars landing). It worked again on 18 February 2021 for Perseverance, landing within 2.1 km of the target point in Jezero Crater with a landing ellipse of only 7.7 km × 6.6 km — the smallest ever achieved at Mars for a multi-tonne spacecraft.

Perseverance’s EDL included two innovations not present on Curiosity: Range Trigger (adjusting parachute deployment timing based on real-time navigation to optimise landing ellipse reduction) and Terrain Relative Navigation (TRN), which used an onboard map and camera to identify safe landing zones, steering the descent stage to avoid hazards. TRN reduced the final landing error to 62 metres — the first time a Mars lander had autonomously targeted its own landing site.

Why This Doesn’t Scale

Every component of the Curiosity/Perseverance EDL system was designed around a ~900 kg landed mass. The physics of Mars EDL impose fundamental scaling limits.

Parachute: The drag area of a parachute scales with cross-section, which scales with diameter squared. To slow a 20-tonne lander (the approximate minimum for a crewed Mars mission) by the same factor as a 900 kg rover requires a parachute roughly 5 times the diameter of Perseverance’s — approximately 60 metres. Supersonic deployment of a 60 m parachute is a qualitatively different engineering problem from a 12 m parachute, with different dynamics, material loading, and failure modes. No supersonic parachute of this scale has been tested at Mars.

Aeroshell: The heat shield must fit inside a launch vehicle fairing. Perseverance used a 4.5 m aeroshell inside a 5 m Atlas V fairing. A 20-tonne lander requires an aeroshell of approximately 10 m diameter — beyond any existing launch vehicle and beyond the current SLS envelope.

Retrorockets: A crewed lander must land with propellant remaining for ascent. The rocket propellant mass required for terminal descent from 80 m/s at 20 tonnes — approximately 10–15 tonnes of propellant — must be carried through entry, increasing the entry vehicle mass and making the problem circular.

This is why SpaceX’s Starship Mars landing architecture abandons the traditional EDL approach entirely. Starship is designed to enter Mars atmosphere nose-first using aerodynamic drag from the vehicle body (no separate aeroshell), then rotate to vertical using thrust and aerodynamic control surfaces (the “belly flop” manoeuvre), and perform a powered vertical landing using the Raptor engines. The vehicle carries 1,200 tonnes of propellant at launch — enough that the Mars landing propellant mass is a relatively small fraction of the total.

Whether this approach works at Mars, under Martian atmospheric conditions that produce aerodynamic forces and heating on a 9 m diameter stainless steel vehicle with different characteristics than a conventional aeroshell, is a question that cannot be fully answered before the first test landing attempt. The EDL environment at Mars remains one of the most challenging in solar system exploration, and the engineering solution for human-scale landing has not yet been demonstrated.

The Communication Delay and Full Autonomy Requirement

A detail that often escapes non-specialists: every Mars EDL sequence is fully autonomous. The one-way light travel time from Earth to Mars ranges from 3 minutes (closest approach) to 22 minutes (farthest). Entry, descent, and landing takes 6–7 minutes total.

By the time the signal indicating “EDL has begun” reaches Earth, the spacecraft has already landed — successfully or otherwise. Mission controllers watch telemetry with a 7–22 minute delay, informed observers of an outcome that has already been decided by software and hardware alone.

The entire EDL sequence — from atmospheric entry to surface contact — is executed by the spacecraft’s onboard computer running a pre-loaded sequence, with conditional branches for off-nominal events but without any possibility of human intervention. Every Mars landing in history has been, in this sense, the spacecraft landing itself.

For the propulsion approaches being considered for human missions that might eventually attempt this landing, see electric propulsion: ion drives and Hall thrusters.