NASA Technical Reports Server (NTRS) 20110016173: Electrostatic Beneficiation of Lunar Regolith: Applications in In-Situ Resource Utilization

7 MB·English

by  NASA Technical Reports Server (NTRS)

Electrostatic beneficiation oflunar regolith: Applications in In-situ Resource
Utilization
Steve Trigwell1, James Captain2, Kyle Weis2, andJacqueline Quinn3
SierraLobo, ESC-24,Kennedy SpaceCenter,FL32899
I
2QinetiQ North America, ESC-i5,KennedySpace Center,FL32899
NASA,NE-Si, KennedySpace Center, FL 32899
3
Abstract
Upon returningto the moon, or further afield such as Mars, presents enormous challenges in
sustaining life for extended periods oftime far beyondthe few days the astronauts experienced
onthe moon during the Apollo missions. A stayonMars is envisioned to last several months,
and it would be cost prohibitive to take all the requirements for such a stay from earth.
Therefore, future exploration missions will be required to be self-sufficientand utilize the
resources available at the mission site to sustain human occupation. Suchan exercise is currently
the focus ofintenseresearchatNASA under the In-situResource Utilization (ISRU) program.
As well as oxygen and waternecessaryfor human life, resources for providing building materials
for habitats, radiation protection, and landing/launchpads are required. All these materials can be
provided bythe regolith presenton the surface as it contains sufficient minerals and metals
oxides to meetthe requirements. However, before processing, it would be cost effective ifthe
regolith could be enriched in the mineral(s) ofinterest. This can be achieved byelectrostatic
beneficiation in which tribocharged mineral particles are separatedout and the feedstock
enrichedordepleted as required. The results ofelectrostatic beneficiation oflunar simulants and
actual Apollo regolith, in lunarhigh vacuum are reported in which various degrees ofefficient
particle separationand mineral enrichmentup to a few hundredpercentwere achieved.
1
•
Introduction
Future NASA manned missions ofexploration, whether back to the moon or further objectives
such as Mars or asteroids, will require the astronauts to spendconsiderable time is ahostile
environmentrequiringcrucial life-sustaining habitats and protection. To provide the required
materials to construct a base ofhabitats that will provide the necessary radiation and thermal
protection, along with future landing and launch pads for replenishment missions, and the
required oxygen and water, and bring all this from earth, would be extremely cost ineffective. It
is therefore necessary to identify the materials thatcan be utilized atthe mission sites to produce
the required end products to sustain life, i.e. living offthe land. This program is an areaof
intensive research at NASA and is collectively known as In-situ Resource Utilization (ISRU).
The lunar regolith covering the surface ofthe moon is a productofeons ofvolcanic activityand
micro-meteoritic bombardment, resulting in a covering offine dust composed ofminerals
containingoxides ofmajor metals such as aluminum, iron, titanium, magnesium, andsodium.
Also included are oxides ofelements such as silicon, calcium and potassium. These elements are
present in minerals such anorthite, olivine, spodumene, ilmenite, and various feldspars and glass
agglutinates. There are several processes in work to extractthe oxygen from the regolith; molten
oxide electrolysis (Sibille etat. 2009), the lunar waterresource demonstration (Moscatello et at.
2009), etc utilizing hydrogen for reduction, resulting in oxygen and water. The residual regolith
is an oxygen reduced alloy ofthe constituent metal elements that may provide sufficient
mechanical and thermal properties for construction materials. Similarly, research is being
undertaken on sintering lunar regolith to form tiles and pads for landing/launch sites (Hintze, et
at. 2009), and to prevent ejecta ofregolith bythe rocket plumes (Metzger et at. 2009, Immeret
al. 2011) thatcouldcause significantdamage to close-byhabitats or structures. Whatever the
2
application, buildingmaterials orwaterand oxygen, differentconcentrations ofthe composing
elements are required, and considerable cost in time and energywould be saved ifthe desired
minerals were more highly concentrated in the feedstock before processing. In the case of
oxygen production, the mineral ilmenite (FeTi03) is ofinterestas it is more energetically
favorable for extraction (Cameron 1992, Rao etal. 1979]. Ilmenite is the mostabundantoxide
mineral in lunarrocks (up to 20% in mare basalts), and can yield up to 10.5% oxygen by
hydrogen reduction ofthe contained FeO (Berggren etaI, 2011). Also the production ofiron and
titanium as side products is a bonus for constructionmaterials. In this case, enrichment ofthe
ilmenite concentration in the lunarregolith would be desirable.
Triboelectrificationand electrostatic beneficiation ofminerals using a parallel-plateseparator is a
well known technology, and has been successfullyused to separate coal from minerals (Inculet et
al 1982), quartz from feldspar, phosphaterock from silica sand, and phosphorus from silicaand
iron ore (Trigwell 2002). Actual lunar regolith is a powderydust with a grain size range of40 to
800 j.!m, with mostfalling between 45 to 100 j.!m(Heiken etal. 1993). Figure 1shows an image
ofa lunar mare soils sample containing a certain fraction offree mineral particles, including
ilmenite andpyroxene. The particles in this image are approximately50- 100 j.!m in size showing
suitabilityfor beneficiation. Lunar regolith also has a low electrical conductivityand dielectric
losses, and combinedwith lackofmoisture preventingparticles from sticking togetherand lack
ofatmospheric gaseous breakdown in the lunarenvironment, lends itselffor ideal
triboelectrification andelectrostatic separation (Trigwell etal. 2006, Captain etal. 2007,
Trigwell etal. 2009).
In this study, the goal was to enrich the ilmenite concentration in lunarregolith for processing.
Forthis, lunarsimulants JSC-IA(OrbitalTechnologies Corp, Madison, WI), and NU- LHT-2M
3
(US Geological Survey, Denver, CO) were used to optimize the process at high vacuum. Actual
beneficiation with Apollo 14 14163 lunar regolith was then performed.
Experimental
The apparatus used to perform the electrostatic beneficiation is shown in Figure 2. A vibrating
feeder was used to provide a steady flow ofregolith into the static charger which consisted ofa
series ofbaffles ofvaried materials so that the regolith would encountermany contactswith the
chargermaterial as it passed through the mixer by gravity. The materials used for the different
experiments were aluminum, copper, and polytetrafluorethylene (PTFE), with work functions of
4.28 eV, 4.65 eV, and 5.75 eV, respectively(Trigwe1l2002). The regolith contains mineral
particles ofdifferent compositions that will tribocharge when in contact with a different material.
Differentparticles will therefore charge up with different polarities and quantityofcharge,
depending upon the work function difference between the mineral and the contacting material.
The regolith then fell into the separatorchamber between two copper plates at voltages ofup +/
30kV. Particles were deflected towards one plate orthe otherdepending upon the quantityand
polarity ofthe charge acquired during the tribocharging and the strength ofthe electric field
(Figure 3). The apparatus was designed for use in a vacuum chamber so that testingand
6
optimization were performed at I x 10- mbar. Forvacuum testing, the regolith was pre-dried in a
thermal vacuum oven so it was as dry as possible before tribocharging.
Inorderto test the viability ofthe beneficiation apparatus, a test mixture was made up, composed
ofthe minerals representing those found in the lunar regolith, namely; feldspar
(Na,K,Ca)AISi308;SiOz), spodumene [LiAI(Sb06)] - pyroxene group, olivine (Mg,Fe)zSi04),
and ilmenite (FeTi03) (Reade Advanced materials, Rl). A summary ofcompositions oflunar
regolith from the literature show the major minerals presentas plagioclase; 20-50 wt%,
4
pyroxene; 40-65 wt%, olivine; 2-15 wt%, and ilmenite; 2-15 wt%. A mixture was made up in a
weight ratio of4:4:1:1offeldspar, spodumene, olivine, and ilmenite. This is the same ratio as
that reported byAgosta (1984,1985) who achieved successful separation ofilmenite from a
mixture of anorthite (feldspargroup), augite (pyroxene group), olivine, and ilmenite in vacuum.
Afteroptimization ofthe beneficiator, lunar simulants JSC-1A andNU-LHT-2M, and finally,
Apollo 1414163 lunarregolith were tested.
The particles were collected in a trayat the bottom ofthe separator that was split into 3or7bins
(the 7 binversion is shown inFigure 3), and analyzed for composition byX-rayphotoelectron
Spectroscopy (XPS). Forthe matrix and numberofsamples analyzed, this technique was
considered efficient enough to allow for quick turnaroundofsamples and provide statistically
valid data(Trigwell etat. 2011). The compositions were comparedto those taken for control pre
separated samples.
The XPS analyses were performedon a Thermo Scientific K-AlphaX-ray spectrometer, using a
Al Ka source at 1486.6 eVata backgroundpressure of2 x 10-9mbar. Ten samples ofeach
separated fraction were analyzedusinga 400 I!m spot size to encompass as manygrains as
possible. The collecteddata were referenced to the CIs peakto 284.6 eV. Quantificationwas
performedbymeasuring the areas under the peaks ofthe detectedelements inthe XPS spectrum,
and normalizingusing sensitivityfactors supplied bythe instrumentmanufacturer to produce
relative atomic % concentrations. The sensitivityfactors are a function ofthe ionization cross
section and the escapedepth ofthe photoelectrons for the elements. The escape depth ofthe
photoelectrons is in the orderof50-100 A, and so XPS is a very surface sensitive technique.
Considering that tribocharging ofthe particles takes place inthe surface region, optimizationof
beneficiation is dependent upon surface composition. Due to the fact that the simulants and
5
Apollo 14regolith had been exposed to ambient, and in the case ofthe simulants, handled is an
assortment ofconditions, quantification was normalized without the carbon peak contribution.
Results and Discussion
Test lunarmixture
Separated lunar mixture samples were collected from each ofthe 7 bins at the bottom ofthe
separator after passingthrough an electric field with the plates at+/- 15 kV, and analyzedby
XPS. The data was then compared to XPS analysis ofthe mixture before separation, and the data
was reported as apercentage increase ordecrease in relative concentrations in Figure 4. Onlythe
majorelements are presented to clarify the plot. From this data, itcan be clearlyseen that the Fe
and Ti concentrations are significantlyhigher in bins 3and 4, while a significant increase in the
Ca and Mg concentrations was observed towards the left hand negative plate. From the known
compositions ofthe minerals used in the test mixture, the spodumene and olivine are more
concentrated towards the negative plate, and the ilmenite more in the middle indicating it did not
acquire as high a charge and so was notdeflected as much. The feldspar and spodumene powders
were an off-white color, the olivine was greenish-white, and the ilmenite was distinctivelydark
grey. When mixed, the result was a light grey powder. In Figure 5, the as-made mixture is shown
on the left, and after separation, the 1 bin (Left paper) is a much Lighter, and bin 7 (right paper) is
sl
also lighter, while the dark grayofthe ilmenite is obviouslyseen between the two in bin 3 ,
showing an obvious correlationwith the XPS data. Theresults here onthe mineral mixture
showed significantenrichment separation ofthe individual mineral compositions after
beneficiation and proofofconcept.
JSC-1A lunarsimulant
6
Previous experiments had been performed byXPS (Trigwell, 2009) to determine ifthere was a
compositional variation as a function ofparticle size inJSC-1A. Little compositional variation
was observedbetween sieved size fractions ofJSC-1A, and so the 50-75 Ilm size fraction was
used as this is near the centerofthe majority ofthe lunarregolith size range of45-100 Ilm
(Heiken etaI, 1993, Inculet and Criswell 1979). The simulantwas run through the separatorwith
the plates at+/-15 kY. In these experiments, only 3bins were used to collectthe separated
regolith fractions, oneunderthe positiveplate, one under the negative plate, and everything else
was collected in the middle labeled bottom tray. This was to limitthe large matrix ofsamples
that would have beencollectedwhile optimizingthe separator if7 bins were used. Table 1shows
the XPS beneficiation data for the JSC-1A simulant as a function ofchargermaterial. The first
observations are thatthe separationconcentrations are notas dramatic as for the test mineral
mixture. This was to be expectedas although ilmenite has beenreported by x-raydiffraction
analysis to be present in JSC-1A (RayetaI, 2010), itwas reported to be at a very low
concentration of0.1 molar% whenanalyzed by SEM/EDS (SchraderetaI, 2010). However,
Shraderetal (2010) also report the Fe andTi are presentas Ti-magnetiteratherthan ilmenite.
There are some distinctdifferences in the Fe and Ti concentrations between the separated
fractions for all three types ofchargers above the 5% errorthat indicatedsome degree of
separation ofminerals. The most significantchanges being for the Naconcentrations for the Cu
and PTFE chargers which mayshow the presence ofthe differentplagioclase phases.
NU-LHT-2Mlunarsimulant
NU-LHT-2M is a newer simulantrepresenting the highland regions on the moon, developedto
help inthe development ofregolith-moving machines and vehicles to be used in future lunar
missions (Zenget aI, 2010). However, although itdoes contain ilmenite, the concentration is still
7
low (0.2 molar%, Schrader etai, 2010). Due to this low concentration, and subsequent
uncertaintyofquantification ofbeneficiated regolith, 5% by weight ofilmenite was added to the
NU-LHT-2M. Separation runs for unsieved NU-LHT-2M were also performed for comparison.
Again as for the JSC-lA, due to the large numberofruns performed, only 3 bins were used to
collect the separated fractions. Theresults are summarized in Table 2. In these results,
enrichment in Fe and Ti were clearly observed in the bottom tray for both the unsieved and75
100 /lm size fraction. This was asimilarresult to that for the test mixture in which the ilmenite
was observed to be enriched in the centerbins rather than towards the positive ornegative plate.
From both the unseived feedstock and the 75-100 /lm size fraction, samples from the bottom tray
showing the highest increase in Fe and Ti concentrations, were submitted to the Johnson Space
Centerfor water production from the regolith. The elemental compositions byXPS and the
corresponding amount ofwater produced from these samples are presented in Table 3 where Run
A and Run B represent the two runs that were performed for the water production. Forthe
original unsieved feedstock and the 75-100 /lm feedstock, 65-70 mg ofwaterwas generated. It
was observed that this rose significantly for the Fe and Ti enriched separated fractions. The
highestamount ofwater generated (192 mg) was from the highest Ti enriched fraction ofthe 75
100 /lm size range. In this case, no significant increase in the Fe concentration was observed
which is interesting as oxygen recoveryfor watergeneration from lunarregolith is generally
consideredto come from the iron oxides presentin the regolith (Berggren et aI, 2011, Allenet aI,
1994). However, the increase in the Naand AI, indicating an enrichmentofthe plagioclase, may
have contributed.
Apollo 14 14163 lunar regolith
8
----- -_._----------_..__._-------~------------------------------
The Apollo 14regolith contains about 1.3 vol.% ofilmenite which is low comparedto the
Apollo 11 (6.5 vol.%) and Apollo 17 highland regolith (12.8 vol.%) (Heiken etaI, 1993). Due to
the small quantityofApollo 14regolith available, the testing was limited to the sieved 50-75 /lm
size range. The beneficiationwas optimized at+/- 15kVapplied to the plates ofthe separator
using an Al chargerand the separated fractions were collected in 3bins. Three separate runs
were performedand the mean values for the elemental relative compositions ofthe separated
fractions were calculated. Theresults are summarized inTable 4. As observed intable 4, the
bottomtray showed the highest increase in both Fe and Ti concentrations comparedto the
originalregolith composition. This data shows remarkable consistencywith thatofthe test
mixture and the NU-LHT-2M beneficiationresults in that the Ti containing mineral tends notto
be chargedand falls through straightthrough the separatorafter tribocharging, and that the other
minerals presentacquire highercharge, both positive and negative, and are deflected towards the
plates. As for the NU-LHT-2M simu1ant, the increase in theTi concentrationwas significantly
higher than for the Fe.
Futurework
Otherwork (Berggren et aI., 2011, Tayloretal 1992)have shown success using magnetic
separation due to the magnetic susceptibilityoflunar regolith. Also, particle size plays an
importantpart. The research in this study showed tribochargingand electrostatic beneficiation
performed optimallywith aparticle size around 50-100 /lm. Smallerparticles were found to
charge up significantly and stick to the chargers and walls, inhibitingregolith flow and
preventingotherparticles from impactingthe chargers. However, it is yet to be determined ifa
fme grainconcentrate wouldbe preferred for processing. Therefore optimization ofthe
beneficiationtechnique utilizing particle sizingand magnetic separation components are being
9
examined. Studies are also planned on performing multiple passes ofthe most enriched fraction
from each pass to determine the ultimate feedstock enrichment.
Furtherstudies were also undertaken performing the beneficiation experimentsusing the NU
LHT-2M simulant in areduced gravityenvironment (1/6 g), and are the subject ofasecond
paper.
Conclusions
Electrostatic beneficiationwas demonstrated to enhance the desired mineral concentrates, i.e.
iron oxides and ilmenite, in single pass separation tests in vacuum using a test mixture and
simulantNU-LHT-2M. SimulantJSC-IA was not thatsuccessful, but was due to its lackof
ilmenite and that itdid notcontain individual separate mineral particles. The results were
particularlyencouraging for the Apollo 14regolith.
These results indicate that tribocharging undera gravity feed, flowed bya twin-plate charge
separator is a viable method for enrichmentoflunar regolith for ISRU processing.
References
Agosta, W.M. (1984). "Electrostatic separation and sizingofilmenite in lunar soil simulants and
samples". Proc. LunarPlanetarySci. XV, Houston, TX. 1-2.'
Agosta, W.M. (1985). "Electrostatic concentration oflunar soil minerals", inLunarBases and
SpaceActivities ofthe 2pt Century, W.W. mendall, Ed. Houston, TX, LunarPlanetary!nst. 453
464.
Allen, C.C., Bond, G.G., McKay, D.S. (1994). "Lunar oxygen production- A maturing
technology". Engineering, Construction, andOperations in SpaceIV, ASME, 1157-1166
Berggren, M., Zubrin, R., Jonscher, P., and Kilgore, J. (2011). "Lunarsoil particle separator",
Proc. 49th AIAA AerospaceSciences Meeting, Orlando, FL. AIAA 2011-436.
10