Regolith Simulants

A regolith simulant is a terrestrial material manufactured to stand in for lunar or Martian soil during hardware testing. No simulant reproduces the target material on every axis at once, so a simulant is selected against the property that dominates the test: particle size distribution and density for mobility and excavation, mineralogy and evolved gas for resource extraction, particle shape and fines content for abrasion and seals [1]. Lunar mare simulants MLS-1 and JSC-1 were developed in the late 1980s, MLS-1 approximating an Apollo 11 high-titanium basalt milled from a coarse-grained gabbro and JSC-1 approximating an Apollo 14 low-titanium basalt obtained from a glassy volcanic ash. Both stocks are exhausted [2]. The root and derivative scheme now used defines a root simulant as a compositional end member, low-titanium mare basalt or high-calcium highland anorthosite, from which derivatives are made by addition of ilmenite, agglutinate glass or nanophase iron.

Fidelity accounting

The Figures of Merit method scores a simulant against a reference regolith by taking the scaled L1 norm of the difference between two property vectors, subtracted from unity, so that 1 is a perfect match and 0 is no match [3]. Six figures are computed: general composition, which bins particles into minerals, glasses, lithic fragments, agglutinates and other; mineral composition; particle size distribution; particle aspect ratio; particle angularity; and density, the last taken as a vector of minimum bulk density, maximum bulk density and specific gravity. Where distributional data on shape are not available, the aspect ratio and angularity figures fall back to the difference of the two ensemble means [3]. A separate open database now consolidates lunar regolith and simulant physical and geotechnical properties from mission reports and laboratory work into a single machine-readable compilation [12].

Four limits are common to every highland simulant characterized to date [4]:

• Grain size within particles. Crystal sizes in the terrestrial feedstocks used for highland simulants (Stillwater, Greenland, Shawmere) are millimeter scale, not micron scale, so simulants cannot reproduce lunar liberation behavior during comminution or beneficiation.
• Texture. Agglutinates, shattered particles, and complex 20 micron grain structures common in lunar samples are not reproduced [4].
• Minor mineralogy. Simulants carry quartz and pyrite at abundances that are too high, and lack maskelynite and armalcolite entirely.
• Space weathering. Nanophase and microphase iron cannot be produced in useful quantity, which caps the spectroscopic accuracy of a simulant and puts its maturity index far from lunar values. JSC-1A was measured to be very different from lunar samples in maturity [4].

Lunar simulants

Simulant	Source material and producer	Target
JSC-1, JSC-1A	Glassy basaltic volcanic ash, Merriam Crater, San Francisco Volcanic Field, Arizona [5]	Apollo 14 low-Ti mare basalt [2]
MLS-1	Milled holocrystalline coarse-grained gabbro [2]	Apollo 11 high-Ti mare basalt [2]
BP-1	Silt-sized washing waste from an aggregate quarry on the Black Point basalt flow, San Francisco Volcanic Field, Arizona [5]	Geotechnical only; not a compositional simulant [5]
NUW-LHT-5M	17.6 wt% norite and 37.7 wt% anorthosite from the Stillwater Complex, 4.7 wt% olivine, and 40 wt% An100 vesicular glass made for the purpose by Washington Mills [4]	Apollo 16 highland regolith [4]
GRC-1, GRC-3, GRC-3b	Blended commercial silica sands, Covia [7]	Geotechnical only; well-graded silty sand, USCS class SM [7]
LHS-1, LMS-1	Exolith Laboratory, University of Central Florida [6]	Lunar highlands and lunar mare [6]
LHS-1D, LMS-1D	Dust-sized fractions of the above	Lunar dust fraction
LHS-1-25A	LHS-1 with 25 wt% sintered anorthosite-iron agglutinates	Agglutinate-bearing mature regolith

BP-1 is made from an alkaline continental basalt with a TiO2 content of 2.2 to 2.3 wt% and a high Fe:Mg ratio [5]. The alkali content is too high for compositional work, though the Fe:Mg ratio is mare-like; its geotechnical properties resemble NU-LHT-2M and Chenobi [5]. Three metric tonnes of NUW-LHT-5M were purchased by NASA MSFC and NASA JSC for engineering use before the characterization report was published [4].

Mars simulants

Simulant	Source material	Notes
JSC Mars-1, Mars-1A	Weathered volcanic ash from a late Pleistocene cinder cone on the flank of Mauna Kea, Hawaii, spectrally matched to the bright martian dust [8]	Loses 7.8 wt% at 100 C and 21.1 wt% at 600 C, far more than martian soil [8]
MMS (Mojave Mars Simulant)	Crushed Saddleback Basalt from an early Miocene outcrop in the western Mojave Desert near Boron, California, sourced by JPL	Loses about 1 wt% water below 450 C, against 2.0 plus or minus 1.3 wt% at Rocknest
JSC-RN (JSC-Rocknest)	65 wt% MMS dust and 35 wt% sub-2 mm MMS, plus sodium perchlorate, ferric sulfate, pyrite, goethite, granular ferric oxides and forsterite	Matched to Curiosity SAM, CheMin and APXS data for the Rocknest sand shadow; 1 to 3 wt% water evolved below 450 C, matching the flight measurement
MGS-1, MGS-1C, MGS-1S, JEZ-1	Exolith Laboratory, University of Central Florida [6]	Global, hydrated-clay, polyhydrated-sulfate and Jezero variants [6]

The Rocknest deposit itself contains roughly 35 wt% X-ray amorphous material and a crystalline assemblage of plagioclase feldspar, olivine and pyroxene, with 42.97 wt% SiO2 by APXS [8]. Approximately 800 kg of JSC-RN was produced by the USGS and shipped in 21 five-gallon buckets; thirteen were randomly sampled for evolved gas analysis and all 21 were subsampled for X-ray fluorescence to verify homogeneity. JSC Mars-1 carries far more water than any martian soil measured in situ, which is why it is unsuitable for ISRU water extraction testing.

Particle shape

Shape parameters measured on a common apparatus across eleven simulants [6]:

Simulant	Aspect ratio	Sphericity	Convexity	Perimeter, um	Equivalent circular diameter, um
LHS-1	0.655	0.561	2.089	128.51	25.63
LHS-1D	0.684	0.657	0.814	52.18	13.09
LHS-1-25A	0.665	0.608	0.997	72.01	16.37
LMS-1	0.652	0.554	1.088	90.92	21.21
LMS-1D	0.669	0.591	0.764	75.95	18.16
JSC-1A	0.655	0.545	1.705	87.53	19.10
BP-1	0.654	0.528	1.304	134.31	27.56
MGS-1	0.647	0.538	1.462	122.47	23.40
MGS-1S	0.665	0.585	0.941	79.68	17.96
MGS-1C	0.674	0.582	0.934	78.72	18.08
Glass beads, 1 mm (control)	0.960	0.858	0.981	3607.17	1.035e6

Reported aspect ratios for Apollo 12 soil run 0.725 to 0.855, and 0.551 to 0.744 for combined Apollo and Luna samples, so the simulants sit inside the returned-sample range on this parameter while remaining more spherical than lunar material is known to be at the fine end [6]. Martian regolith is subrounded to very rounded and equant, the consequence of impact generation followed by aeolian rounding, whereas basalt-derived simulants are angular because they are produced by fragmentation [11].

Particle size distribution

Source: [5]. Public domain (NASA).

Laser-diffraction distributions measured on a common instrument [6]:

Simulant	Mean size, um	D10, um	D50, um	D90, um
LHS-1	92	124.73	293.48	405.91
LHS-1D	30	18.27	32.96	37.46
LHS-1-25A	65	58.84	141.14	275.71
LMS-1	90	53.91	116.50	277.21
LMS-1D	35	84.53	211.50	428.69
JSC-1A	164	57.06	225.94	343.05
BP-1	198	92.10	228.91	319.80
MGS-1	90	109.06	232.36	384.08
MGS-1S	119	72.05	145.09	213.36
MGS-1C	25	73.83	190.86	342.97
JEZ-1	68	120.30	269.56	512.48

None of these simulants contains material coarser than 2 mm, by design [6]. GRC-3b and its variants are classified as well-graded silty sand under the Unified Soil Classification System, with a particle size distribution unchanged between the base material and the dust-suppressed variant [7].

Strength and density

Values below are quoted at comparable bulk density, since cohesion and friction angle in granular material are both functions of density rather than fixed material constants [7], and the same dependence holds across the Exolith lunar and Martian simulants [6].

Material	Minimum density, kg/m3	Maximum density, kg/m3	Cohesion, kPa	Internal friction angle
Lunar regolith (reference)	870	1930	0.1 to 1	30 to 50 deg
GRC-3b, literature	1520	1939	0	47.8 deg at 1839 kg/m3
GRC-3b, current batches	1581	1972	0	33.4 deg at 1792 kg/m3
GRC-3b-DST	1352	1971	11.5	35.5 deg at 1816 kg/m3

The 14 degree spread between the literature friction angle for GRC-3b and the value measured on current production batches at similar relative density is batch variation in a nominally unchanged product [6]. Apollo relative density at depths to 20 cm runs 60 to 80 percent, and is itself a derived quantity computed from specific gravity and the minimum and maximum index densities rather than a direct measurement.

GRC-3b-DST was produced by adding a dust-suppressing additive at 0.7 wt% to the GRC-3b formulation, 22 tonnes of it, to cut respirable silica [7]. The additive raised cohesion from zero to between 5 and 11 kPa and lowered the minimum density from 1581 to 1352 kg/m3, which widens the range of achievable soil states towards the lunar minimum of 870 kg/m3 without reaching it. Respirable silica exposure during bed preparation measured 9.1 to 12 ug/m3 against the OSHA permissible exposure limit of 50 ug/m3 over an eight hour day [7].

Cohesion rises exponentially with bulk density in every simulant measured. Over 1.32 to 1.97 g/cm3 the coarse lunar simulants span 0.185 to 1.672 kPa; the dust fractions span 0.116 to 1.289 kPa over 0.80 to 1.30 g/cm3; the Martian simulants span 0.179 to 1.406 kPa over 1.03 to 1.86 g/cm3 [6]. Internal friction angle across all samples and densities ranges from about 15 to 65 degrees, and does not vary monotonically with density: MGS-1C peaks near 29 degrees at 1.14 g/cm3 then falls to 25 degrees at 1.35 g/cm3, while MGS-1S rises from 34 degrees at 1.31 g/cm3 to near 53 degrees at 1.83 g/cm3. Adding 25 wt% agglutinates to LHS-1 raised cohesion at any given density through particle interlocking but cut the maximum achievable density from 1.80 to 1.75 g/cm3, and so reduced the attainable cohesion range by about 10 percent [6].

Apparatus, stress range and preparation state

A cohesion figure quoted for a named simulant is a property of the apparatus and the stress range as much as of the material. LHS-1 is reported at 0.311 kPa from an ASTM D3080 direct shear fit over normal loads of 0.098 to 0.67 kPa at bulk density 1.32 g/cm3 [4], and at 11.0 kPa in a compilation of values taken at approximately 90 percent relative density with the fit made over normal stresses of 25 and 50 kPa [16]. Cohesion in a Mohr-Coulomb fit is the intercept extrapolated from the normal stresses actually applied, so a fit over hundreds of pascals and a fit over tens of kilopascals do not return the same quantity for the same material. JSC-1A separates by a factor of 112 the same way: 0.1 to 2.5 kPa in the prior literature at density and method unstated [4], against 8.9 plus or minus 0.1 kPa in direct shear at 25 and 50 kPa normal stress and 11.2 plus or minus 0.1 kPa in unconsolidated undrained triaxial at 30, 60 and 90 kPa confinement, both on a worked sample at 90 percent relative density [16].

Preparation density is the second condition. Published triaxial values for JSC-1A run 41.87 degrees at 24.6 percent relative density, 46.48 degrees at 54.7 percent and 56.70 degrees at 84.6 percent, and BP-1 is quoted at 51.0 degrees at 85 percent relative density against 39.0 degrees with zero cohesion at 50 percent [16].

Apparatus type alone produced a 19 degree spread when four methods were run on one soil during the Apollo Lunar Roving Vehicle wheel program at the Waterways Experiment Station [17]. Crushed Napa basalt at 0.9 percent moisture and 29 percent relative density gave a friction angle of 38.5 degrees by vacuum triaxial at 0.85 psi normal stress, 34.0 degrees by in situ plate shear at 1.06 psi, 29.0 degrees by bevameter ring shear and 19.8 degrees by Cohron sheargraph, with apparent cohesion on that same soil of 0.03 psi by trenching, 0.14 psi by bevameter ring shear and 0.32 psi by sheargraph [17]. The same vacuum triaxial series returns 38.5 degrees at 0.85 psi and 37.0 degrees at 1.65 psi, so the angle is not constant within one apparatus at one density either.

Absorbed atmospheric water is a first-order error term in any test run at ambient conditions. LHS-1D holding roughly 0.4 wt% absorbed water differed in cohesion by about 15 percent from the same sample baked at 140 C for 24 hours [6]. Vacuum conditioning of a simulant bed for polar testing is therefore part of the test setup rather than an incidental step [13].

Behavior under vacuum

Gas trapped in pore structures and water adsorbed on particle surfaces are released when a simulant bed is pumped down, and the release can be violent enough to loosen and weaken the bed, changing its consolidation state before the test begins. In Vacuum Facility 13 at NASA GRC, a chamber 3.66 m tall and 1.5 m in inner diameter, a one tonne bed of GRC-3 or LHT-3M 0.64 m deep and 0.914 m square showed disruptions that could be suppressed by reducing pump rate, with no disruptions observed below 2.5 torr regardless of rate [9]. Reaching 1e-5 torr took a full week because of slow outgassing [9].

Electrostatic behavior

Terrestrial atmosphere provides a charge dissipation path that neither the lunar surface nor the Martian surface provides, so tribocharging measurements made in air are not transferable. Charging of JSC Mars-1 against Teflon and fiberglass cylinders has been measured in a 10 mbar CO2 atmosphere using an aerodynamic electrometer with the dust driven onto the insulator by an impeller fan, alongside the mineral constituents Ottawa sand, Fe2O3, Al2O3 and MgO tested separately [10]. Because simulants do not reproduce lunar nanophase iron or agglutinate texture [4], charge transfer measured on a simulant bed is a bound on the flight case rather than a prediction of it.

Flight surprises traced to simulant fidelity

InSight HP3 mole. The mole was required to reach 3 m depth in the representative simulants MMS below 2 mm and WF-34 quartz sand, and to execute at least 20,000 strokes; the flight unit met both requirements and life tests on flight-equivalent units exceeded 60,000 strokes with no loss of hammer force [11]. The design relied on hull friction from cohesionless soil to balance hammer recoil. Sharp-edged, high friction angle Syar sand had been dropped from the test set earlier in development because no Mars data supported it as representative, and a redesign for high self-cohesion and high friction angle materials was judged beyond the available schedule and budget. The mole reached about 40 cm [11]. Slope stability analysis of the resulting pit walls gave a duricrust cohesion of 5.8 kPa at an assumed 30 degree friction angle, and penetration resistance implied 4 to 25 kPa for friction angles of 30 to 40 degrees, against Martian sand cohesion of 1 kPa or less. The landing site conditions fall on the 4 m zero-active-pressure isoline, against a verified test domain bounded near 0.5 m [11].

Terrestrial gravity compensation. The same analysis establishes that a terrestrial test bed should use soil of higher cohesion than the target body to compensate for Earth gravity when reproducing the active soil pressure regime a penetrator sees. For MMS sand tested on Earth, the zero-active-pressure layer thickness matches that of mildly cohesive sand on Mars, which is why MMS was an acceptable analogue for the insertion phase and not for the phase that followed [11].

Phoenix icy soil delivery. On the first two attempts to deliver an icy-soil sample to the Thermal and Evolved Gas Analyzer, the sample congealed in the scoop and stuck to it during the dump, most likely because sunlight fell directly on the material. A third sample, scraped off the surface so that it carried less ice, did not congeal and was delivered successfully [15].

Apollo Lunar Roving Vehicle. Dust thrown by the wheels was controlled by fenders, and where a rear fender extension was knocked off, on Apollo 16 and 17, dust covered the crew and the equipment [14]. Dust deposited on the rover radiators could not be brushed off and insulated them: on Apollo 15 the rover batteries ran 68 to 78 F above their expected temperatures [14].