Radiation shielding: how much regolith over the habitat
Study 003 · Draft · Concept study, not a built system.
Study 003 · Draft · First published 2026-10-13 · Revised 2026-10-13
Corrected 2026-10-13: was 2 m and 480 t of cover; the solar-activity case was misread and one paper was credited to the wrong authors. See corrections.
Question
How much soil does a habitat need overhead to cut the radiation dose?
Assumptions
- Surface dose0.64 ± 0.12 mSv per day, Curiosity RAD at Gale crater (Hassler 2014)
- Cruise dose1.84 ± 0.30 mSv per day inside the MSL spacecraft (Zeitlin 2013)
- LimitNASA career limit of 600 mSv effective dose (NASA-STD-3001, 2022)
- Depth modelRöstel 2020, Table 3, average solar modulation; equivalent dose in a 15 cm water sphere, which the authors call an upper estimate
- Loose regolith1.5 t per m³ (our assumption)
- Habitat6 × 12 m footprint, 2 m margin each side, flat cover, no side berms (our assumption)
| Target | Andesite (2.8 g/cm³) | Sandstone (2.2) | Sulfur concrete (2.0) | 10% water (2.4) | 50% water (1.4) |
|---|---|---|---|---|---|
| 100 mSv/yr | 1.67 m | 2.15 m | 2.48 m | 0.87 m | 0.87 m |
| 50 mSv/yr | 2.05 m | 2.65 m | 3.05 m | 1.37 m | 1.64 m |
Depths from Röstel et al. 2020, Table 3.
The numbers
Unshielded surface: 0.64 mSv/d × 365 ≈ 234 mSv/yr → 600 mSv in ≈ 2.6 yr Cruise (2 × 180 d): 1.84 × 360 ≈ 662 mSv (Zeitlin 2013) Areal density for 50 mSv/yr, dry rock (Röstel 2020): 205 cm × 2.8 g/cm³ ≈ 574 g/cm² 265 cm × 2.2 g/cm³ ≈ 583 g/cm² Depth in loose regolith at 1.5 g/cm³ (our approximation: areal-density scaling): 574 / 1.5 ≈ 383 cm ≈ 3.8 m Footprint (6 + 4) × (12 + 4) = 160 m² Cover 160 m² × 5.74–5.83 t/m² ≈ 920–930 t Hydrated cover: 50% water 164 cm × 1.4 ≈ 230 g/cm²; 10% water 137 cm × 2.4 ≈ 329 g/cm² → 40–60 % less mass than dry rock for the same target
Result
Modeled: keeping a buried habitat under 50 mSv a year needs about 575 g/cm² of dry cover: about 2.1 m of dense rock, or about 3.8 m of loose soil, roughly 920 t over a small habitat. Ice-rich ground cuts the mass by roughly half. The round trip alone, about 662 mSv (Zeitlin 2013), already exceeds NASA’s 600 mSv career limit with MSL-like shielding.
Trade-offs
| Shield | Areal density for 50 mSv/yr | Notes |
|---|---|---|
| Dry rock or regolith | about 575–585 g/cm² | free on site. Thin cover can be worse than none: equivalent dose peaks about 30–40 cm down |
| Water-rich regolith (10–50% water) | about 230–330 g/cm² | hydrogen moderates neutrons |
| Water tanks around sleeping quarters | not computed here | doubles as storage and a storm shelter |
Open problems
- Table 3 is a model (GEANT4/AtRIS) validated against RAD at the surface only; there is no measurement 2 m down.
- Scaling depth by areal density across materials is our approximation, not a published result.
- Doses are higher at solar minimum. Side shielding is not modeled.
- The rate and energy to move about 900 t robotically before the crew arrives are unknown.
What would change the answer
- Put long-occupancy rooms at the centre under the deepest cover.
- Use local ice or water in the overburden once extraction works.
Sources
- Hassler et al., “Mars’ Surface Radiation Environment Measured with the Mars Science Laboratory’s Curiosity Rover,” Science 343 (2014)
- Zeitlin et al., “Measurements of Energetic Particle Radiation in Transit to Mars on the Mars Science Laboratory,” Science 340 (2013)
- NASA-STD-3001 Vol. 1 Rev B (2022): career effective dose limit 600 mSv
- Röstel, Guo, Banjac, Wimmer-Schweingruber, Heber, “Subsurface Radiation Environment of Mars and Its Implication for Shielding Protection of Future Habitats,” JGR Planets 125 (2020)
- Zhang, Guo, Dobynde, Wang, Wimmer-Schweingruber, “From the Top of Martian Olympus to Deep Craters and Beneath: Mars Radiation Environment Under Different Atmospheric and Regolith Depths,” JGR Planets 127 (2022)
Status: Sketch is first-order arithmetic that may move by ±50%; Draft has full structure but uses an approximation we name; Study has every input sourced. Figures are rounded. Where a value is our assumption, the table says so. Errors are logged on Corrections; report one on Discord.