NASA MSFC Lunar Regolith Terrain Field and Dirty Vacuum Chambers

NASA/Joe Kuner. Public domain (NASA / US government work).

A lunar surface mechanism cannot be qualified against air. Ordinary tribology assumes every metal surface carries an oxide and an adsorbed film, and the friction and wear a designer measures on a bench are properties of those films, not of the metal beneath them; pump the pressure down and the films are lost, junction adhesion rises, and the wear mechanism itself changes from oxidative to metal transfer, a shift documented for laboratory single crystals decades before any lunar hardware existed [4]. Add the dust that a wheel or a drill actually meets and the problem compounds: micron-scale regolith is abrasive, electrostatically active and, once loose in a chamber, a hazard to the pumps meant to hold vacuum around it. Marshall Space Flight Center’s answer is not one flagship chamber but a graded set of them, plus an outdoor field, so that a concept can be raised through fidelity in stages rather than built straight into the hardest test available.

Why a ladder of facilities

Marshall’s Space Environmental Effects laboratory presents its three chambers as a deliberately staged sequence: early materials work in the Lunar Environment Test System (LETS), payload and system level exposure in PLANET, and integrated qualification in the V20 chamber. The reason is optical and physical rather than organizational. Electron beam and ultraviolet sources used to reproduce the charged-particle and radiation environment only cover a spot of about 0.2 to 0.5 m before their uniformity and intensity fall off sharply, so a chamber built much larger than that spot cannot apply the space environments it is meant to simulate evenly across a full-scale test article [1]. A single large chamber would therefore have to trade combined-environment fidelity for size, which is why Marshall instead runs a 76 cm chamber for materials, a roughly 2 m chamber for payloads, and a 6 m chamber that gives up everything but vacuum and thermal control in exchange for room. The outdoor Lunar Regolith Terrain field sits outside that ladder entirely, at one gravity and Alabama ambient conditions, for work that needs open ground rather than a controlled environment.

Every rung shares one problem that ordinary vacuum tribology does not: keeping the dust itself in the chamber, a constraint discussed in more detail below under what the ladder does not reproduce. Choosing what to test the dust against is a second problem, and one that predates any of the three chambers. A quantitative scoring method for how closely a candidate simulant matches a reference lunar composition and particle size distribution was worked out as a companion to Marshall’s own simulant requirements program: the composition score is defined as one minus the normalized vector difference between a simulant’s constituent fractions and the reference’s, so a perfect match scores 1.0 and the framework itself reports no measured score for any real material, only the formula and illustrative example vectors [6]. The same framework sketches analogous scores for particle size distribution and density but only qualitatively, with the full derivation deferred to an unpublished companion requirements document [6]. Applying the composition score to the simulants actually in use found none scoring above 0.65 against a real Apollo core on composition, and worse on particle size, so whichever simulant goes into LETS or PLANET is a known, bounded mismatch to the material it stands in for rather than an unknown one [7].

Dust in vacuum is also why an ordinary friction number does not transfer to the Moon at all. Chemisorbed films control clean-surface friction under vacuum: half a monolayer of oxygen on a tungsten surface drops its friction coefficient from about 3.0 to 1.3, and graphite sliding against silver rises from its atmospheric value to roughly 0.5 once pumping strips the silver oxide the graphite otherwise transfers against, a mechanism worked out for brushes and slip rings well before any of Marshall’s dirty chambers were built [4]. That earlier work also warns that its own numbers are upper bounds rather than predictions: its specimens were clean single crystals prepared in situ, not engineering hardware carrying machining marks, plating and handling contamination, so a mechanism built to flight tolerances should be expected to do worse in vacuum than the reference values suggest, before any dust is added at all [4]. Marshall’s chambers add regolith to that already pressure-sensitive picture, which is one further reason a single environment cannot stand in for the surface: the dust changes what the vacuum itself does to a sliding contact.

Laboratory data

Parameter	Value
Operator	NASA MSFC, Space Environmental Effects laboratory [1]
Location	Huntsville, Alabama, United States
Commissioned	LETS operational since 2008; PLANET customer access from mid-2024
Type	Outdoor analogue terrain field and three dirty vacuum chambers
Floor area	Not published for the field or for the laboratory
Capabilities	LRT field, LETS, PLANET, V20
Simulant or terrain	Not published as an inventory. JSC-1A has been run in the LETS tray
Instrumentation	Residual gas analysis, particle imaging velocimetry, electron gun
Ground truth	Not published
Fidelity limits	No chamber spans the problem; simulant must be baked before pumping
Access	Not published for the field. PLANET customer access planned from mid-2024
Cited by	Break the Ice Lunar Challenge excavation demonstrators, tested in V20 [2]

Capabilities

Lunar Regolith Terrain field

Parameter	Value
Temperature	Not controlled. Alabama outdoor conditions [1]
Slope	Not published
Gravity offload	Not applicable. Outdoors at 1 g

Marshall’s outdoor lunar analogue site. No area, simulant inventory, bed depth, terrain preparation, buried target set, instrumentation list or support arrangement is published for it.

LETS

Parameter	Value
Working volume	76.2 cm (30 in) diameter chamber housing a regolith box
Temperature	Cryogenic shroud; range not published
Illumination	A 100 keV electron gun provides the charged particle environment
Simulant or terrain	JSC-1A in the tray, under a non-airtight cover
Instrumentation	Particle imaging velocimeter; 100 keV electron gun

Source: [1].

The Lunar Environment Test System is the smallest and oldest of the three chambers, operational since 2008 [1]. Its simulant tray has been run with JSC-1A under a non-airtight cover designed to minimize turbulent flow across the surface material during initial pumpdown. It has been used for regolith charging and lofting studies and for exposing material coupons. Choosing JSC-1A at all rests on a standardization effort that predates the chamber, held here in the record of a January 2005 MSFC workshop [5]. That workshop established a root-and-derivative simulant framework because the earlier reference materials, JSC-1 and MLS-1, were by then exhausted, poorly characterized, and had never been quality-controlled as reference stock, and because Apollo hardware failures, from LRV slip and sinkage to seal failures on every rock box, argued that surface hardware needed a traceable standard material to be tested against at all. It also tabulated the geotechnical envelope any simulant is implicitly judged against, a bulk density of 1.4 to 2.2 g/cm3, a cohesion of 0.1 to 1 kPa, and a friction angle of 30 to 50 degrees, and ranked grain size and its distribution as the property a simulant must reproduce first, ahead of particle density and glass content [5]. The report is explicit that this is a recommendation rather than a record, however: the simulant family it proposed did not exist yet when it was written, and its projected demand of 125 to 250 metric tons over four years is a planning estimate rather than a delivered quantity. Later scoring work built on that framework also records a chemistry mismatch bearing on oxygen production: the NU-LHT simulants run about 4 weight percent FeO against a 5 percent lunar reference, while JSC-1 runs closer to 11 percent, a gap that matters to any ISRU test run against either material [7].

PLANET

Parameter	Value
Working volume	2 m diameter x 3 m cylinder, domed doors both ends; bed 1.22 x 2.44 m, 0.2 m deep
Test article limits	Floor designed for roughly 1200 kg including the bed structure
Vacuum	At least the 10^-7 Torr regime empty; scroll roughing, cryopump for high vacuum
Temperature	-180 C on liquid nitrogen to radiant heating by quartz lamps
Illumination	Near and vacuum ultraviolet sources
Simulant or terrain	About 800 kg of simulant in the bed; product not published
Instrumentation	Residual gas analyzer; quartz crystal microbalance under consideration

Source: [1].

PLANET, the Planetary, Lunar and Asteroid Natural Environments Testbed, sits between the other two and is the most explicitly combined-environment of the three. It is designed to combine high vacuum, charged particle radiation with electrons from 50 eV to 100 keV and protons from 200 eV to 30 keV, near and vacuum ultraviolet, and thermal extremes from -180 C on liquid nitrogen to radiant heating by quartz lamps [1]. For comparison the lunar exosphere ranges from 10^-9 Torr by day to 10^-12 Torr at night. Radiation exposure is nonetheless not done inside the chamber itself: samples are irradiated separately, at a Pelletron or x-ray facility, and then transferred in, a break in the combined-environment argument that the chamber’s own designers flag rather than resolve.

Any moving stage built into that bed, for the regolith feed or mechanism testing described below, still faces the pressure-dependent friction problem that clean vacuum tribology already documents: what a bearing or a slip ring does at PLANET’s vacuum level is not what it does in air, regardless of the dust [4].

The chamber is split into two zones along its length, each fed by a group of four 10 in ConFlat overhead ports focused on the test article plane at a slightly different focus height, so that a hot bright zone and a cold dark zone can be run at once to emulate the terminator [1]. As of the design description, the vessel, pumps and automated controls were still in procurement for late 2023 delivery, with environments and instrumentation to follow afterward and customer access planned from mid-2024; no measured chamber performance, field uniformity or commissioning result had been published [1].

V20

NASA/Joe Kuner. Public domain (NASA / US government work).

Parameter	Value
Working volume	6 m (20 ft) diameter x 8.5 m (28 ft) long
Temperature	Cryogenic shroud with radiant heating; range not published
Illumination	None. No space environments other than vacuum and thermal
Simulant or terrain	A bed large enough for full-scale hardware; product and depth not published

Source: [1].

V20, outfitted as the Lunar Surface Simulator, is the large end of the sequence: a thermal vacuum chamber at the Marshall Environmental Test Facility, converted recently as of 2023 to handle regolith and dust, with a bed that can hold full-scale hardware, outfitted with a cryogenic shroud and radiant heating but no other space environments [1]. Containing the dust at that scale is a facility problem of its own, and V20’s answer was a large clean tent around the chamber entrance. The Lunar Surface Simulator outfitting of this chamber is described only in abstracts and presentations, so no shroud zone count, temperature range, pump configuration or cart and bed dimension is recorded above. A much larger dirty chamber exists elsewhere in the NASA system for comparison: Glenn’s Space Power Facility at Plum Brook Station runs a 30.5 m thermal vacuum chamber alongside base-shake, acoustic and electromagnetic test capability added between 2007 and 2011, but that facility qualifies whole launch payloads rather than lunar surface hardware, and its published capability is a clean-chamber envelope with no regolith handling of its own [15]. Its acceptance-level numbers are a useful contrast in scale: a 34,000 kg test article on a three-axis shake table, a 2860 m3 acoustic chamber reaching 163 dB, and a thermal vacuum chamber reaching 2e-6 Torr within eight hours, all published while acceptance testing was itself still incomplete, so most of that envelope is a design capability rather than a demonstrated one [15].

Instrumentation

PLANET is designed around two additions of its own. A regolith application and distribution system is to meter known quantities of simulant from above a test article, so that thin layers can be deposited for later exposures or introduced periodically from a hopper for wear or additive manufacturing work; the designers call the problem non-trivial, because the most damaging particles are 10 to 20 microns across and therefore invisible to the eye, making both precision and verification hard [1]. Mechanism and seal testing for wear, abrasion and leak resistance is the other, with motion stages, load cells and sample fixturing being designed for it. Real time chemical monitoring is by residual gas analyzer, with a temperature-controlled quartz crystal microbalance under consideration [1].

LETS carries a particle imaging velocimeter and a 100 keV electron gun alongside its cryogenic shroud, which is what makes it the charging and lofting chamber of the three [1]. The requirement those instruments answer to is stated most explicitly outside Marshall, in NASA Glenn’s own dust-simulation work: cohesion effects between grains appear once pressure falls below a few torr, but reproducing surface adhesion, the property that governs dust sticking to a mechanism or a suit, needs vacuum many orders of magnitude better, around 10^-7 Torr or finer [3]. That distinction is also why simulant cannot go into a chamber straight from the bottle. A dust activation protocol used elsewhere in NASA’s lunar dust work runs continuous stirring during pumpdown to prevent dust explosions, a bakeout near 200 C until the residual gas spectrum stops changing, thermal cycling to about -150 C, and an air-plasma strip followed by a hydrogen-helium plasma re-reduction meant to mimic solar wind implantation on the grain surfaces [3].

None of Marshall’s three chambers is built around testing a dust mitigation device, but the vacuum condition its chambers hold is what such devices are tested against elsewhere in NASA. A canted-spring PTFE rotary seal of the type flown on a Mars rover’s instrument arm was screened against JSC-1A and a highlands simulant at about 4e-7 Torr and kept simulant out of the seal-shaft interface in every one of seventeen trials, though only to 10,000 cycles against a service expectation of millions [9]. An electrodynamic dust shield built from transparent electrodes recovered a dust-loaded solar panel from as low as 11 percent of its clean output to above 90 percent within about two minutes once activated, tested at a rougher 1 kPa rather than high vacuum [11]. Passive surface treatment against a different simulant cut measured dust adhesion by roughly a quarter on treated quartz and by nearly two thirds on treated Kapton, but ultraviolet exposure then increased adhesion again on the untreated control surfaces, evidence that a mitigation result taken at one point in a chamber’s illumination and pumping cycle does not necessarily hold at another [10].

None of these three mitigation results was taken at a lunar-representative vacuum plus dust plus gravity combination at once. The seal screening ran only four completed vacuum trials before the vacuum motor itself failed, and the shaft roughness changes recorded across the test matrix were internally inconsistent with the measured seal weight loss in a way the authors flag without resolving [9]. The dust-shield trial reported the adhesion force at an assumed single particle diameter rather than a measured one, and the authors state plainly that the panel’s electromagnetic compatibility with the rest of a spacecraft was left unresolved [11]. Gravity was absent from all three, which matters because the physical requirement for a dust adhesion facility elsewhere in NASA is stated just as plainly: a designer needs vacuum near 1e-7 Torr or better to see adhesion at all, reduced gravity is deliberately not simulated even there because the facility’s own designers judged it not a prudent use of resources, and they note that omission could by itself invalidate a mitigation strategy that depends on dust falling off a surface under its own weight [3]. The dust-shield result also came from only four electrode-spacing configurations, with no stated count of repeated trials per configuration, so its two-minute clearing time is a demonstration rather than a characterized rate [11]. The surface-treatment result, in turn, discarded its highlands-simulant runs entirely, since grinding and sieving that material created uncharacterized larger particles the authors judged uninterpretable, leaving only the mare-simulant substrates as usable data [10]. And the seal screening is explicitly preliminary against a service target of millions of cycles, single-seal per test condition in most of its matrix, at constant speed with no start-stop or side load [9]. All three were tested on small coupons or commercial parts rather than flight hardware, at a single particle size fraction in the shield trial, and none reports a repeated dust-loading and clearing cycle count that would speak to long-term durability [11].

What it does not reproduce

Simulant that has not been baked is not the same material as simulant that has. Accepted practice across the NASA dirty chambers, PLANET included, is a thorough bakeout of the simulant above 100 C before pumping to high vacuum, to drive off adsorbed water that would otherwise boil off as the pressure falls; hotter bakes reach structural water but can change other chemical properties as well [1]. Any mechanical result taken on unbaked simulant at ambient humidity describes a different material from the one that goes into vacuum.

Pumping a granular bed is a facility problem in itself. The measures the Marshall team records for keeping simulant out of the pumps are tortuous paths, filters, placement of the pump inlet high in the chamber, a non-sealed lid over the simulant box to damp viscous and turbulent flow during rough pumping, and a preference for cryopumps over turbomachinery in the high vacuum regime, since a particle reaching a turbopump is a catastrophic mechanical failure. Those same measures cut pumping speed in the molecular flow regime [1].

Scale against fidelity. No one chamber spans the problem, which is why there are three. LETS gives radiation and charging over a tray; PLANET gives the combined environment over a 1.22 x 2.44 m bed; V20 gives room for full-scale hardware but only vacuum and thermal control, with no other space environments [1]. The stated progression is materials in LETS, payloads and systems in PLANET, integrated qualification in V20 [1].

Gravity. None of the facilities offloads. The field is outdoors at 1 g and the chambers hold their beds under Earth weight [1].

A simulant is not the regolith it stands in for, and no facility here corrects for that. Marshall’s own simulant-scoring work finds no candidate scoring above 0.65 against a real Apollo highland core on a composition figure of merit, and worse on particle size, because only one simulant family carries anything standing in for lunar agglutinates at all [7]. A pilot three-body abrasion program using two of these simulants found that grain-scale work elsewhere shows real lunar plagioclase to be two to three times weaker in compression than its terrestrial simulant counterpart, so abrasion measured against simulant in any of these chambers is expected to overstate wear against real regolith. That same program is a pilot by its authors’ own label, run at one load, one speed and one duration in ambient air with no vacuum or temperature control, so it supports only a material ranking under that one condition rather than a life prediction for any mechanism [8]. Within that one condition, alumina abrasive removed the most material from every specimen tested and hardened 1045 steel resisted all four simulants and sands tried against it, while soft materials, aluminum and PMMA, wore comparably whether the abrasive was simulant or ordinary sand [8]. And a field trial of a regolith feed mechanism found the same auger that worked reliably with JSC-1A ran inconsistently with a different simulant of nominally similar size, and needed a 10 degree tilt to feed at all, evidence that a mechanism qualified against one simulant cannot be assumed to behave the same way against another. That trial also had to develop its own regolith-tolerant valve seals with a vendor over six to eight months, because few vendors would take on a custom design for abrasive, hot spent material at all, and it states plainly that no wear or lifetime data exist for those valves beyond the ten-day field campaign that produced this result [12]. Oxygen production in that same campaign is reported only as a rate equivalent inferred from batch operation, not a sustained measurement, and a planned heat-recuperation improvement was deferred to a later phase the paper does not report [12].

The published fit-to-use guidance is a composition and size extrapolation, not a mechanical measurement. Marshall’s own simulant user’s guide scores nine simulants against a single Apollo highlands reference and then states, twice, that no cohesion, friction angle, bearing strength, shear, sinkage, abrasion rate or drilling energy was measured to produce its recommendations; the guidance for which simulant suits excavation, drilling or abrasion work is the authors’ extrapolation from composition and particle size to expected behavior [7]. Because that reference core came from one highlands site chosen for a polar-outpost architecture, the same guide states that a mare or high-titanium application would be judged wrongly by its scores. A test run in LETS or PLANET against a simulant recommended this way is therefore a test against a material whose mechanical behavior was inferred, not confirmed, to resemble the lunar soil at hand.

Campaigns run there

Break the Ice Lunar Challenge follow-on testing, July 2025. A rover from a Centennial Challenges finalist, second overall at the June 2024 live demonstration and finale, was installed in the V20 chamber at the Environmental Test Facility over a concrete slab it would operate on, lifted in on the rail cart by the overhead crane [2]. The Break the Ice challenge asked competitors to design, build and demonstrate robotic technologies that could excavate and transport icy lunar regolith. The record held here of this campaign is a single image-library caption, dated 30 July 2025 and naming the rover, the challenge placement and the chamber, and it gives no test parameters, duration or result beyond that [2]. The photographs of the rover on its rail cart and of the crew maneuvering it toward the chamber door are drawn from that same NASA public-domain release [2].

PLANET procurement and build, 2022 onward. The vacuum system was in procurement in late 2022, with the vessel, pumps and automated controls due for delivery in late 2023, outfitting with environments and instrumentation to follow over the months after, and customer access planned from mid-2024 [1]. No published full paper found here reports the commissioning results.

MSFC’s own dirty-chamber heritage predates the current ladder by decades. In May 1971, before any of the three current chambers existed, Marshall ran a full-scale wire mesh wheel and fender through 65 vacuum runs at one-sixth g aboard a C-135A aircraft, to settle whether the Lunar Roving Vehicle’s fender design threw a dangerous dust plume at a following crew member [13]. The program found the original fender inadequate, that an added flap gave the minimum acceptable control, and that of five distinct mechanisms by which a driven wheel throws dust, only the forward plume in a 120 to 60 degree arc was a real hazard; it also dismissed two feared failure modes, finding the mesh wheel did not fling gravel and that the fender was not a power-costing dust pump [13]. The quantitative instrumentation for that campaign failed in flight, so its dust and velocity findings rest on film interpretation rather than measurement, but it is the only reduced-gravity vacuum wheel test in this facility’s history and the origin of the fender design whose Apollo 17 failure is cited by nearly every modern lunar dust paper. Its own power data are called suspect by the report’s authors, who describe the slip and normal-force behavior recorded during the runs as a confused picture, so even the one number the campaign did trust, that the fendered wheel drew less power than the unfendered one, carries that qualification [13]. Parabolic flight also bounded the campaign to only seconds of one-sixth g per run, far short of the sustained reduced gravity a Marshall dust-adhesion facility would need to answer the same question today [3]. And the test soil was a terrestrial analog rather than lunar regolith, so the 1971 report itself settles nothing about the cohesion question it set out to study [13].

Some numbers used elsewhere to justify dirty-chamber testing at all trace back further still, to the geologic record rather than to any test program: lunar regolith thickness by site, grain size and its maturity dependence, and the impact flux that produces the material in the first place come from a 1970s survey of Apollo and Luna samples together with Surveyor and Lunar Orbiter photogeology [14], a source that itself notes irreconcilable factor-of-two disagreements between seismic methods at the same Apollo site, and that the impact flux feeding every model is tied to the present epoch only by the assumption that its exponent held constant for the last three and a half billion years, an assumption the same source reports another research group concluding does not hold. The survey’s own sample base is narrow: nine sites, six of them mare, with no pure highland area far from mare or KREEP terrain sampled by the time it was written, so it characterizes neither highland nor polar regolith, the terrains the current simulant standard and Marshall’s own test program are built around. Every number in it also predates post-Apollo remote sensing, so it carries no mechanical property at all, no cohesion, no friction angle beyond a slumping threshold, no bearing strength, only the geologic constants a 1970s track and rare-gas laboratory could measure [14].

What is not established

No published source gives a floor area, simulant inventory, bed depth or instrumentation list for the outdoor Lunar Regolith Terrain field, and none states its access terms. PLANET had not been commissioned at the time of the design description held here: the vacuum vessel was still in procurement, no measured chamber performance, uniformity or campaign result exists for it, and its combined-environment claim already has a known gap, since radiation exposure happens outside the chamber and is transferred in rather than applied in situ. V20’s Lunar Surface Simulator outfitting is documented only in abstracts and presentations, so its shroud zone count, temperature range, pump configuration and bed dimensions are not recorded anywhere citable [1]. No chamber in the ladder offloads gravity, for the reasons already given above. Every mechanical number produced against a simulant in any of these chambers, abrasion, wear, feed behavior or otherwise, carries the standing caveat that no simulant scores above 0.65 against real lunar regolith on the properties that have actually been measured, that the published fit-to-use guidance for which simulant suits which test is an extrapolation from composition and size rather than a mechanical measurement, and that the gap is structural rather than a matter of better mixing [7].