Apr 10, 1974 - Electrical property measurements have been made on an Apollo 15 lunar soil ..... + tan (9). Wâ¢og'(w, T). The parameters ⢠and r' in the third term ...
VOL. 79, NO. 11
JOURNAL
OF GEOPHYSICAL
RESEARCH
APRIL 10, 1974
Electrical Propertiesof Lunar Soil Sample 15301,38 G. R. OLHOEFT •
LockheedElectronicsCompany,Houston, Texas 77058
A. L. FRISILLO AND D. W. $TRANGWAY: NASA JohnsonSpace Center, Houston, Texas 77058 Electrical property measurementshave been made on an Apollo 15 lunar soil samplein ultrahigh vacuumfrom room temperatureto 827øC for the frequencyspectrumfrom 100 Hz through ! MHz. The dielectricconstant,the total ac losstangent,and the dc conductivityweremeasured.The dc conductivity showedno thermal hysteresis,but an irreversible(in vacuum) thermal effect was found in the dielectric loss tangent on heating above 700øC and during the subsequentcooling. This appearsto be related to severaleffectsassociatedwith lunar glass above 700øC. The sample also showedcharacteristiclowfrequencydispersionin the dielectricconstantwith increasingtemperature,presumablydue to MaxwellWagner intergranulareffects.The dielectricpropertiesmay be fitted to a model involving a Cole-Cole frequencydistributionthat is relativelytemperatureindependentbelow 200øC and followsa Boltzmann temperaturedistributionwith an activation energyof 2.5 eV above 200øC. The dc conductivityis fitted by an exponentialtemperaturedistributionand becomesthe dominantlossabove700øC.
Over the past severalyears, severalexperimentalgroups have been measuring the electrical properties of returned lunar samples(seeChunget al. [1972]or Othoeftet al. [1973a] for references).Dielectricparametersare important in the interpretationof groundand orbital radar measurementsand in the interpretationof two Apollo 17 experiments.Both the surface electricalpropertiesexperiment [Simmonset al., 1972] and the Apollo lunar sounderexperiment [Brown, 1972] are relying on the uniqueelectricalnature of the lunar surfaceto probe the upper severalhundred metersof the moon. The dc conductivityinformation is being usedto interpret lunar surface and orbital magnetometerdata [Dyat and Parkin, 1972; Dyal et al., 1972; Sill, 1972; Sonnett et al., 1972]; the laboratory conductivitydata versusthe temperaturedata are being used to convert magnetometer-derivedconductivitydepth profiles of the moon into deep interior temperatures. There have, however, been only minimal attempts to parameterizeand modeltheoreticallythe electricalproperties of lunar sampleswith the goal of identifyingthe mechanisms behindthem. Part of the reasonfor so few attemptshasbeen
losseswere attributed to careful transfer of the samplein dry nitrogen to the vacuum chamber without exposure to atmosphere.Katsubeand Collett [1971], Chunget al. [1972], and Strangwayet al. [1972] have observedthe flattening of the dielectric constant frequency responsewhen samples were measuredin dry nitrogeninsteadof air. This reportpresentsthe resultsof measurements of the elec-
trical propertiesof lunar soil sample15301,38and analyzes them by using the wide temperature and frequency ranges covered. EXPERIMENT
the lack of adequatedevelopment in the applicationof stan-
Lunar soil sample 15301 is soil scoopedfrom station 7 at the Apollo 15 landing site on the northwest rim of Spur crater. It was returned from the moon in sealedApollo lunar sample return container 173 [Apollo 15 Preliminary Science Report, 1972]. The sample was processedin dry nitrogen in the Lunar ReceivingLaboratory, Houston, Texas, and 2.5 grams were transferred under a dry nitrogen tent in the NASA GeophysicsBranch into the electrical properties vacuum
dard dielectric theories to terrestrial
chamber.
rocks without
moisture.
It has long been known that moisture has an extremely strong effect on the electricalpropertiesof rocks [Keller and Licastro, 1959; Parkhomenko, 1967]. Baldwin and Morrow [1962] and Saint-Amant and Strangway [1970] have shown evidencefor the enormouseffect on dielectricpropertiesof moisture adsorbed from the atmosphere.Strangway et at. [1972] extended these investigationsto lunar samplesand terrestrial analogs of lunar materials. Measuring in good vacuum,they reportedconductivities2-3 ordersof magnitude lower than had previouslybeen reported [e.g., Chunget at., 1972]. Similarly, the sampleexhibiteddielectriclossnearly an order of magnitude below other determinations. These low • Now at Department of Physics,University of Toronto, Toronto, Ontario.
2Now at Departmentof Geology, Universityof Toronto, Toronto, Ontario.
Copyright¸
1974by the AmericanGeophysicalUnion. 1599
Vacuum
was maintained
at less than
10 -7 torr
throughout the measurements.A three-terminal (disc electrodeswith guard) electrodeassemblywas usedaccordingto the American Societyfor Testingand Materials [1970] standard procedures. The electrodes were manufactured from molybdenumand held by high thermal conductivity,low electrical conductivityberylia (BeO). General Radio 1620-AP and 1682capacitancebridgeswere usedto measurethe dielectric properties, and a Hewlett-Packard 4329A high-resistance meter read the dc conductance.The repeatability of dielectric measurementswas0.5%, with an absoluteaccuracyof 2%. An Aerovac 610 massspectrometermonitored the residual gas pressure, and, as a function of temperature, typical gas pressureswere comparable to those noted by Gibson and Moore [1972], who attributed the releasedgas primarily to solarwind products.Temperaturewasvariedin a seriesof 24hour cycles from room temperature to some maximum in 50øC incrementsand back to room temperature. Over a 14day period, successivethermal cycles reached maximum
1600
OLHOEFTET AL.: LUNAR SOIL SAMPLE15301,38
An exponential(sometimescalledWagner) temperaturedistribution is used for the dc conductivity data [Smith, 1942;
temperaturesof 25ø, 172ø, 200ø, 279ø, 457ø, 500ø, 538ø, 610ø, 647 ø, 700ø, 827ø, 306ø, and 163øC. Measurementswere taken
Adler, 1971]:
at each50ø intervalduring heatingand coolingafter thermal equilibrium had been established;a maximum temperature variation of 2ø was observedduring the recordingof data. The bulk densityof the sample was 1.47 + 0.06 g/cms. Measurementsof the electricalpropertiesare summarized at severaltemperatures andfrequencies in Table 1.We [01hoeft et al., 1973b]havecomparedthe roomtemperaturevalueswith those of other soils. At the density noted above the highfrequencydielectricconstantapproached3.0 and is in general agreementwith measurementsfor other soils by Gold et al. [1970, 1971, 1972], Bassettand Shackelford[1972], and Katsubeand Collett [1971]. It is slightly higher than the average value of 2.5 reportedfor severalApollo 15 soilsby Goldet al.
a = aoeAv
(2)
wherethe directlymeasuredparametersare •', the real part of the dielectric constant; tan b, the total electrical loss (see below); and a, the dc conductivity. All are functions of temperatureand (exceptconductivity)frequency. The followingparametersare derived from the fits of the above equationsto the data: Ko •© ro Eo a
dielectricconstantlow-frequencylimit; dielectricconstanthigh-frequencylimit; time constant of relaxation at infinite temperature; relaxationactivationenergy,electronvolts; distributionparameter(appearingas the slopeof the log losstangentversusthe log frequencyfar from the frequencyof peakloss)whoselimits are 1, an infinitely broad distribution of relaxation times [Fuoss and Kirkwood, 1941], and 0, a single Debye relaxation; ao conductivityat 0øK, mhos per meter;
[1973].
At room temperatureover the frequencyrangemeasured, the dielectricconstantis nearlyfrequencyindependent.As the temperature increases,the characteristiclow-frequencydispersionbeginsto appear.Figures1 and 2 displaythe dielectric constant as a function of frequency at severaltemperatures; the circles representthe data points taken in a 1-, 2-, 5frequencysequenceat fixed temperature,and the solid lines are the theoretical fit explained below. Below 10•' Hz and above 106 Hz the fits are extrapolated. Similarly, the loss tangent as measuredis shown in Figures3 and 4. In Figure 5 the losstangent at constantfrequency(100 Hz) is shownas a function of temperature.Three regionsof thermal activation can be recognized (below 200øC, at 200ø-450øC, and
A
reciprocalconductivitythermalactivationparameter.
The constantsarek = 8.6176 X 10-seV/øK and{0 = 8.854185 X
10-•' F/m; T is temperaturein degreesKelvin, and co= 2;rf rad/s.
The propagationfactor •, for electromagneticenergyin homogeneousmaterial is [Stratton, 1941]
above450øC),and the thermalhysteresis causedby heating above 700øC is apparent. The dc conductivityis shown in Figure6 with the solidline exponentialtemperaturedistribution fit. Also shownare the error barsoverseveralcyclicmeasurementsand the residuallossesof the sampleholder and
where • - •0 is the permittivity,# is the magneticpermeability, a is the dc conductivity,andj = (-1) •/•'. On theassumption thatthereis a complexdielectricpermit-
measurement system.
tivity, i.e.,
•/•' = #•co •' + jlzaco
• = •o(•' - j•")
THEORY
In
this
section
we
consider
(3)
a theoretical
model
that
and that both the conductivityand the magneticpermeability are real and without imaginarylosstermsof their own [Miles et al., 1957; Fuller and Ward, 1970; Olhoeft, 1972], the parametercharacterizingthe loss of the material is then defined as the imaginary part of the propagation factor divided by the real part and is called the losstangent:
parameterizesthe observeddielectricproperties[01hoeftet al., 1973a].The model followsthe developmentsof Cole and Cole [1941] and Gevers [1945], which use distributionsof Debye singlerelaxationswherea singlerelaxationalonehasbeenfound to be insufficient.A Cole-Cole frequencydistributionwith a Boltzmanntemperaturedistributionis usedfor the dielectric
tan O = tan (b/2) = lm 'y/Re 'y
data:
(5)
(It is relatedto the phaseangle0, as is shown.)We thus see that there are two distinct terms that are separable and
1 Jr-(jcoroeeø/kv) '-" TABLE 1.
(4)
Lunar Sample 15301,38 Electrical
Data-Dielectric
Constant
Temperature, øC
Frequency, kHz 0.1
25 3.24
(0.010) 1.0
10. 100.
3.20
(0.0058) 3.18 (0.0033) 3.17
(0.0032) 1000.
3.17
(0.001) The loss tangent
110
200
330
409
500
610
700
827
......
3.86
4.82
7.38
21.4
47.
......
(0.11)
(0.27)
(0.68)
(0.69)
(1.0)
'''
3.49
3.85
5.28
7.54
20.
ß ''
(0.31) 3.82 (0.12)
(0.56) 4.75 (0.28)
3.24
(0.0086) 3.20 (0.0043) 3.19
(0.0009) 3.03
(0.0008)
is given in parentheses.
3.33
(0.009) 3.20 (0.0044) 3.19
(0.0008) 3.16
(0.0008)
(0.046) 3.35 (0.017) 3.29
(0.0052) ......
......
(0.12) 3.47 (0.043) 3.35
(0.014)
3.44
(0.038) 3.14
(0.015)
3.80
(0.12) 3.55
(0.052)
(0.7) 9.4 (0.60) 4.8
(0.26) 3.5
(0.09)
'''
... • 30. (1.1) 10.
(0.53) 4.4
(0.2)
OLHOEFTET AL.' LUNARSOIL SAMPLE15301,38
1601
1.0'
•40
]
o ø zo
700oc
O.Ol
•
OOOC
,.o, •o
4o9
ß
330
ø
o.ool ,
•
25 ø
ß o.oool
Io 254•• 0
I
I
I0 •
I
I0 2
I
I0s
I
104
I
I0 8
I
I0" Hz
I
I01
I
I0 2
I
I0 •
I
104
I
I0 •
I01 Hz
Fig. 1. Dielectric constant versus frequency for several temperatures.The circlesrepresentthe data, and the solidlinesrepre-
Fig. 3. Total loss tangent versus frequency at several temperatures. The circlesrepresentthe data,andthe solidlinesrepre-
sent the theoretical
sent the theoretical
fit.
fit.
characterizethe loss due to dielectricand conductivity Debyelikeat highertemperatures, in generalagreementwith mechanisms:
tan • --
/c"(co,T) •oe • d• • ½o, T) •O•o•(co, T)
(6)
Applying.these equations to thedata,we areableto fit the experimentaldata with:
this expression. Although the losstangentwas frequencydependentat all temperatures,it was nearly temperatureindependentbelow 200øC,and it wasfound necessary to add a third term to the expressionabovefor the losstangent:
tanc• /=
ao = 0.6 X 10-•7 mho/m;
•(oor/) • sin
1+
(8)
A = 0.0237 øK- •'
Ko= K©= ro = •o =
where/• = 0.33 (the frequencydistribution breadth for this term), and the completeloss tangent is thus
6700; 3.0; 2.56 X 10- •' s; 2.5 eV.
AT
tan
•'(w,D + W•og'(w,T) + tan
(9)
Thesearethefitsillustrated bythesolidlinesin Figures1'-6of the experimental data. It was alsofoundnecessary to make
The parameters• and r' in the third term cannot be deter-
the frequencydistributionparametera (sometimescalledthe activationenergydistributionparameter[Whitehead,1946]) temperaturedependentin the followingapproximatemanner
our range of measurement.The constraintsfor these values are/• = 0.33 and (2a'r')-• < 10•',where the first constraintis
[Olhoeftet al., 1973a]' 1 - a = [ec