NASA GRC Lunar Dust Adhesion Bell Jar

Current operating status unconfirmed. Every published description of this chamber dates from 2007 to 2011 [1][2][3], and it does not appear in the index NASA Glenn publishes of its laboratories and test facilities [4]. That index omits facilities Glenn is known to operate, so it does not establish that the chamber is gone [4]. Contact Glenn for current availability.

The Lunar Dust Adhesion Bell Jar is a small vacuum chamber at NASA GRC built around one question: what a lunar dust mitigation method does once the dust has been made lunar. Its distinguishing asset is not the chamber but the activation protocol run inside it. Simulant out of the bottle carries adsorbed water and organics on every grain, and a coating brushed clean in that state does not predict lunar behavior; the chamber bakes, thermally cycles and plasma-treats the simulant before a sample ever sees it [1].

It is the documented case where a dust mitigation technique passed in air and failed in the chamber. Nylon and PTFE strip brushes that cleared most of the dust from a radiator coupon on the bench recovered only about half of pristine thermal performance under simulated lunar conditions, which is what happened to the nylon brushes carried on the Apollo Lunar Roving Vehicle [3].

It is a small-sample facility. A sample is a 2.54 cm disk, runs are four to seven per material, and the deposition step cannot be commanded [2].

Laboratory data

Parameter	Value
Operator	NASA GRC [1]
Location	Cleveland, Ohio, United States
Commissioned	Facility paper published 2007 [1]
Type	Dust adhesion and thermal-optical property chamber
Floor area	Not published
Capabilities	Chamber, dust conditioning
Simulant or terrain	JSC-1AF and NU-LHT-1D, activated in situ [1][3]
Instrumentation	Quadrupole residual gas analyzer; computer-controlled x-y-theta microscope stage [1][2]
Ground truth	Dust coverage counted by microscopy after the fact, not commanded [2]
Fidelity limits	1e-7 torr against 1e-10 to 1e-14 torr lunar; no reduced gravity; no particle radiation [1]
Access	Not published
Cited by	dust testing, thermal vacuum testing

Capabilities

Bell jar chamber

Parameter	Value
Working volume	Not published. Samples up to about 7 cm diameter [2]
Test article limits	Coupons on a rotatable sample holder; 2.54 cm disks used in practice [2]
Vacuum	About 1e-7 torr with dust present; 7e-8 torr after bakeout [1][2]
Temperature	77 K liquid nitrogen cold wall; 30 K helium-refrigerated cold box in use [1][2]
Illumination	Xenon arc solar simulator, 200 to 1200 nm [1]
Simulant or terrain	Lunar simulant loaded into an in-chamber sieve [2]
Slope	Not applicable
Gravity offload	None [1]
Instrumentation	Type K thermocouples on sample backs; residual gas analyzer to 1e-11 torr partial pressure [1][2]

The pump is a liquid-nitrogen-trapped oil diffusion pump, chosen for robustness to dust ingress rather than for ultimate pressure, and the chamber is built around keeping dust out of it: a top-mounted pumping port, five alternating baffles, an offset antechamber, and glass sampling tubes between the baffles so dust transport can be monitored [1]. The residual gas analyzer detects pump oil at 1e-11 torr partial pressure, which is the check that the trap is working.

The cold wall sits centimeters from the xenon-illuminated zone, so a sample can hold an illuminated and a shadowed region a few centimeters apart, reproducing the 200 K contrast that occurs over a few centimeters on the lunar surface [1]. A 40 K cryobox for permanently shadowed crater conditions was planned but not installed at the time the facility was described [1]; the later measurement campaigns used a helium-refrigerated 30 K copper cold box [2].

Dust conditioning and deposition

Parameter	Value
Working volume	In-chamber sieve over the sample holder [2]
Test article limits	25 or 32 micron sieve mesh [2]
Temperature	Dust baked at 200 C, chamber at 60 C, overnight [2]
Instrumentation	Steel comb stirring the dust throughout pumpdown [1]

The activation sequence is the transferable part of this facility. Rough down to 1e-1 torr with the dust stirred continuously by a steel comb, which is a safety requirement rather than a preparation step because trapped gas in a cohesive powder can throw the charge; bake at 200 C at 1e-6 torr until the residual gas spectrum stops changing; three thermal cycles to about -150 C; an RF air plasma at 1e-4 torr for at least an hour to strip organics; then a hydrogen-helium plasma to re-reduce the grain surfaces and stand in for solar wind implantation [1][2]. Chamber pressure after that sequence is about 2e-7 torr [2].

The 200 C bakeout is above the roughly 125 C lunar surface maximum, so the treated simulant has been taken somewhere the destination does not go [1]. The authors state only that they hope the treatment produces the right surface defects and dangling bonds; no published source demonstrates that a plasma-treated simulant grain resembles a regolith grain [1].

Thermal property extraction

Parameter	Value
Working volume	One coupon at a time under the in-chamber lamp
Test article limits	AZ-93 paint and silvered or aluminized FEP on aluminum and graphite composite [2]
Illumination	Collimated unfiltered xenon lamp inside the chamber; about 10 suns as used [3]
Instrumentation	743-node, 1840-conductor Thermal Desktop and SINDA/Fluint model [2]

Absorptance is fitted from the heating curve and emittance from the cooling curve, by minimizing the weighted average temperature difference against the model to under 1 percent [2]. The pristine values that anchor everything else, each an average of four to seven runs: AZ-93 on aluminum at emittance 0.886 +/- 0.024 and absorptance 0.173 +/- 0.029; AZ-93 on K-1100 graphite composite at 0.833 +/- 0.027 and 0.196 +/- 0.006; AgFEP on aluminum at 0.719 +/- 0.041 and 0.073 +/- 0.006.

What it does not reproduce

Deposition is not repeatable. Activated simulant is too cohesive to sieve evenly. A 60 s sieving run sometimes deposited more than a 600 s run, and sinusoidal, jarring and vibrating sieve motions all failed to give consistent coverage [2]. Dust coverage in this facility is measured after the fact, never commanded, and the resulting data set is a scatter of whatever coverage happened to occur. A sample that looked clean to the eye was 5 percent covered under the microscope [2].

Coverage measurement is coarse. Fifty of 641 frames per sample are sampled by a computer-controlled x-y-theta stage at 100x with random frame and angle selection, gray-scale thresholded, with a finite population correction applied [2]. The 95 percent confidence interval on coverage is still 9 to 13 percentage points wide, worst near 50 percent coverage. A conclusion drawn from a coverage difference smaller than that is not supported.

Lamp intensity was never independently calibrated. The detector saturated, so the 29.1 kW/m2 figure, about 21.5 suns, was back-solved by assuming literature absorptance and emittance for the AZ-93 on aluminum baseline [2]. Every subsequent extraction inherits that assumption, and all experimental error is deliberately swept into absorptance and emittance rather than apportioned [2].

Vacuum, by three to seven orders of magnitude. About 1e-7 torr against a lunar surface value of 1e-10 to 1e-14 torr [1]. The facility paper argues the gap rather than closing it: clean-body cohesive behavior appears below a few torr, an inference drawn from the 1971 MSFC Apollo LRV wheel dust tests where LSS-4 simulant clumped and held together until impact below about 5 torr, while surface adhesion work needs 1e-7 torr or better [1].

Gravity, and particle radiation. Reduced gravity was explicitly ruled out as not a prudent use of resources, which the facility paper itself notes could invalidate any mitigation strategy that relies on dust falling off [1]. X-rays, gamma rays and energetic particles are not simulated; the hydrogen-helium plasma stands in for a solar wind ion flux of about 3e8 ions/cm2/s rather than reproducing it [1].

Campaigns run there

Thermal control surface degradation, reported 2010. A full monolayer of activated lunar simulant raises the ratio of solar absorptance to infrared emittance of a thermal control surface by about a factor of three, and that holds across AZ-93 paint and silvered FEP on both aluminum and graphite composite substrates [2]. The consequence is that a radiator expected to be dusted must be sized about three times larger. The figure is an extrapolation from sub-monolayer points with R squared 0.72 and one excluded outlier, not a measurement at monolayer coverage [2].

The same campaign overturned the common assumption that dust changes absorptance and not emittance. At sub-monolayer coverage AZ-93 emittance degraded to a projected 80 percent of clean at a monolayer, while AgFEP emittance was enhanced by about 20 percent [2]. That matters most where a radiator views shade rather than sun.

Brushing as a mitigation strategy, reported 2011. Ten brushes spanning nylon, PTFE, Thunderon, carbon and fiberglass bristles in strip, fan and round geometries, first on the bench in air and then in the chamber with plasma-cleaned, vacuum-baked, hydrogen-helium reduced JSC-1AF [3]. Strip brushes that cleared over 90 percent of the dust from AZ-93 in ambient bench tests restored only about half of pristine absorptance-to-emittance in the chamber, matching the Apollo experience. Longer, more pliable fan and round brushes did work: the Escoda nylon fan and the Zephyr fiberglass round brush both restored over 80 percent of pristine performance in 20 strokes and over 90 percent in 200 [3].

In absolute terms, AZ-93 went 0.22 pristine, 0.65 dusted, 0.30 after 20 strokes and 0.26 after 200; AgFEP went 0.12, 0.40, 0.21 and 0.15 [3]. Brushed AgFEP therefore beats brushed AZ-93 by 1.6 to 1.9 times despite retaining more dust: residual coverage after 200 strokes was 3 percent on AZ-93 with the Zephyr brush and 10 percent with the Escoda, against 29 to 31 percent on AgFEP for both. Brushing speed made no difference over a factor of ten in stroke rate.

Three qualifications the campaign states about itself. The bench stages used NU-LHT-1D highland simulant and the chamber stages JSC-1AF, because funding ended the plan to run both, and the assumption that brush ranking is simulant-independent is not shown [3]. Only four AZ-93 coupons were available for the in-chamber stage, and in-situ dust removal efficiency spreads by about a factor of two, so the difference between the two winning brushes is inside the error on most surfaces. Brushing visibly scratched both AZ-93 and FEP, and no cumulative damage or life assessment was made.