VOL. 78, NO. 29 JOURNAL OF GEOPHYSICAL RESEARCH OCTOBER 10, 1973

Lunar Powder Simulator under Lunarlike Conditions:

Dielectric Properties

I•OMiN ALV•Z

E•gi•eerit•g Geoscien, ce, U•iversity o] Cali/ornia, Berkeley, Cali]ornia 94720

The dielectric response of a powdered basalt simulating lunar fines is studied in the temperature range found in the lunar surface and under high-vacuum conditions. Two con- current analyses are made: one regarding moisture contamination effects and the other related to the response of the lunar regolith. The atmospheric moisture remaining in the pore system of the sample in the high vacuum produces a low-frequency dispersion, which is attributed to the presence of isolated water adsorption centers in which the local con- ductivity values are raised with respect to the conductivity of the basalt. It is suggested that, a change in the main electric conduction mechanism occurs between 300 ø and 370øK; close to 300øK, conduction seems to be dominated by adsorbed moisture, whereas at 370øK the main conduction process appears to be controlled by thermally activated carriers. The dielectric response of the lunar regolith during a lunarion is schematically described. It is found that during approximately seven tenths of a lunation the dielectric response of a 5- to 10-cm surface layer should be fairly constant; during the remaining three tenths it should undergo rapid changes. Changes in the dielectric properties of the surface layer should be controlled by temperature; in the subjacent layers such changes are thought to be con- trolled by the increasing density of the regolith.

The temperature on the surface of the lunar regolith varies from approximately 100 ø to 400øK during the lunar synodic period. The amplitude of the temperature wave is rapidly attenuated with depth [Robie and Hemi, gway, 1971; Hoyt eta[., 1971], and thus only the first few centimeters experience the full change in temperature. In the present study a terres- trial basalt simulating lunar powder has been subjected to temperature variations in the 100ø-373øK range in a high vacuum. The changes in the dielectric properties of the lunar simulator have been determined and are

thought to represent in a qualitative way the corresponding variations in the uppermost layer (i.e., 5-10 cm in depth) of the lunar regolith.

Although various measurements on the dieice- trio properties of actual lunar samples have been made [Chung eta/., 1970, 1971, 1972; Kat- sube and Co.llett, 1971; $tra•gw.ay et al., 1972; Gold eta/., 1970, 1971], few have gone to tem- peratures below 273øK; none of these measure- ments, however, have been made under high- vacuum conditions. Similar studies on terrestrial

samples [Strangway, '1969; Saint-Amant and Strangway, 1970; Campbell and Ulrichs, 1969]

Copyright • 1973 by the American Geophysical Union.

also have not combined a high vacuum and low temperatures.

To reach conditions similar to those in the

lunar surface, a terrestrial powder has to lose a considerable amount of the atmospheric mois- ture usually adsorbed to its surfaces [Alvafez, 1973b]; in so doing, the sample experiences changes in its dielectric properties. An analysis of these changes yields an insight into the reverse process, namely, that of dielectric vari- ations arising from moisture contamination; this process occurs in lunar samples that have been in contact with atmospheric moisture [Strangway eta[., 1972].

Low-temperature determinations are neces- sary to generate a complete theoretical descrip- tion of the dielectric behavior of lunar samples, as well as to assist in the explanation of elec- tric phenomena occurring in the lunar night. For studies on terrestrial rocks, low-tempera- ture measurements offer the advantage of freezing any water that may be present in the sample and thus effectively reducing conduction effects associated with it.

•ExPERIMENTAL TECHNIQUE

The experimental setup consisted of a Davies and Wilder high-vacuum system (model 616),

6833

6834 ALYAREZ' LUNAR SIMULATOR AND DIELECTRIC PROPERTIES

a cryogenic thermometer with a range of 1 ø- 300OK from American Magnetics, a thermistor calibrated in the temperature range of 77 ø- 400øK, a General Radio capacitance-measuring assembly (model 1610-B), and a guarded dec- trode system; all are commercial instruments except the last one.

A detailed description of the guarded elec- trode system will be given elsewhere; here we shall briefly outline its characteftstics. A vacuum-sealed stainless steel box (Figure 1) is used as a liquid nitrogen reservoir inside the

L.N. , • '-• OUt

© Fig. 1. Schematic view of the guarded elec-

trode system. The whole assembly is maintained high-vacuum chamber. The box is externally in the high vacuum' low and high temperatures fed by means of adequate feed-throughs. A are obtained when liquid nitrogen (L.N.) fills, heavy copper strap is welded to the-base of the or heated air flows through, the reservoir. The liquid nitrogen reservoir; as long as there is sample exchanges heat with the reservoir via liquid nitrogen in the reservoir, the strap main- the copper bar and the guarded electrode. 1,

upper electrode; 2, guarded electrode; 3, guard; tails the liquid nitrogen temperature. 4, sample or sample holder; 5, liquid nitrogen

At. the top of the copper strap are located reservoir; 6, copper bar; 7, liquid nitrogen feed- the guarded electrode and the guard ring. The through' 8, vacuum chamber' 9, cryogenic ther- former is a copper disk held against the strap toometer; 10, superinsulation layers; 11, electri- by copper screws. The latter is a copper ring cally shielded cable. electrically insulated from the copper strap by a Teflon layer. The sample (in the case of solid one in contact with the guarded electrode. rock specimens) or the sample holder (in the Measurements were made at temperatures of case of powders) is placed above the guard 100 ø, 298 ø, and 373øK. The powdered sample and guarded electrodes. Heat is directly ex- was packed to densities of 1.8, 2.0, and 2.2 changed between the sample and the guarded g/cm '• in the working volume (area of 17.9 cm • electrode, and thereby the sample in the high and thickness of 0.5 cm) of the sample holder. vacuum is cooled. The third (upper) electrode Once a series of determinations was finished clamps the sample from above; adjustments with one sample, it was discarded and replaced in the position of this electrode are made by by a fresh one. Increasing densities were at- means of a threaded guide. tailed by compressing increasing amounts of

Heat. losses by radiation from the liquid powdered sample in the constant volume of the nitrogen reservoir and sample are minimized sample holder; the uncertainty in the density by means of several layers of superinsulation determinations is estimated to be ---+0.01 g/cm • (i.e., aluminized Mylar sheets) that cover the of the values given above. The sample holder electrode assembly in the evacuated chamber. is made out of a good-quality dielectric material To reach the higher temperatures (i.e., 373øK), (National Electrical Manufacturers Association pressurized air, or nitrogen, after being heated G-10) and has suitable mechanical character- to the appropriate temperature, is made to flow istics in the temperature range of interest. through the reservoir.

A set of measurements in the frequency range TABLE 1. Source and Sizes of Basalt Components of Lunar Simulator

of 30-10 '• Hz, at one temperature, iavolved periods of 6-8 hours; of these, 3-4 hours were s• Opening, P ..... t Finer necessary to reach thermal quasi-equilibrium so .... Si .... by •eight conditions, and the remaining time was dedi-

Knippa 12 1.68 95 cated to the actual dielectric measurements. It Knippa 50 0.297 79

Knippa 100 0.149 66 must be noted that with the present system Berkeley 200 0.074 50 it is not possible to reach complete thermal equilibrium, since heat exchange with the sam- The basalt coup .... ts .... the same grain size distribu-

tion as the samples from Apollo 12 core tube S/N 2013 ple occurs only at one of its ends, namely, the (section B).

ALYAREZ' LUNAR SIMULATOR AND •)IELECTRIC 1:)ROPERTIES 6835

The sample was packed in the sample holder at atmospheric conditions and set in place in the electrode assembly. Evacuation started slowly with the mechanical pump; in approxi- mately 45 min the sample was exposed to the high vacuum. At this time the vacuum was usually between 6.0 X 10 -7 and 3.0 X 10 -6 t orr, depending on the packing density of the sample and the atmospheric humidity at which the sample was prepared. It was left to outgas in the high vacuum for approximately i hour more; in this time the pressure decreased to between 3.0 X 10 -• and 3.0 X 10 -• torr. Liquid nitrogen was admitted to the liquid nitrogen reservoir, and cooling of the sample started.

The temperature of the guarded electrode reaches 82øK in 20 min after the first filling of the liquid nitrogen reservoir in a vacuum of 10-8-10 -• torr. The sample grains directly in contact with this electrode are expected to attain such a temperature in approximately the same time. We observed, however, that, at places within the sample that were not directly in contact with the electrode, heat propagation was extremely slow under high-vacuum condi- tions, possibly owing to a heat conduction mechanism in which radiation predominates ICrcreefs, 1972]; whether this possibility is the actual case cannot be ascertained from our

experiments. For calibration purposes, several

temperature determinations were made inside the powder sample. It was found that tempera- ture gradients of approximately 100øK/cm in the direction of the symmetry axis of the sample holder existed in the lunar simulator of density of 1.80 g/cm • after 90 min of the first filling of the liquid nitrogen reservoir when the vac- uum was in the range of 10-•-10 -• torr.

Thus the time necessary to reach thermal quasi-equilibrium conditions appeared to be un- duly long. To reduce' the time, we increased the pressure in the vacuum chamber by means of dry nitrogen (i.e., 99.99% N.,), since it does not induce changes in the electric response of 'resistive' rock specimens [Alvafz, 1973b]. It was observed that the thermal conduction of

the sample improved markedly when the N• pressure was between 0.1 and 1.0 torr. Al- though higher pressures only slightly improved thermal conduction in, the sample, they con- siderably increased the overall thermal losses of the system. With the sample at a pressure of 1.0 torr of N• the times necessary to reach thermal quasi-equilibrium were reduced to the 3- to 4-hour periods mentioned above. The thermal gradients in the direction of the axis of the sample holder were approximately 20øK/cm when the dielectric responses were determined.

The sample being in thermal quasi-equilib-

5.0 I I I

4.0

--IO0•K A-O - 0 ................

2.0 I I I I0 I0 2 I0 3 104 105

FREQUENCY (Hz)

Fig. 2a.

Fig. 2. (a) Dielectric permittivity against frequency and (b) loss tangent against frequency for the sample of packing density of 1.8 g/cm 8. Temperatures followed the sequence 100øKA, 298øKA, 373øK•, and 298øK•. The pressures ranged from 2.2 X 10 -8 torr at 100øK• to 1.2 X 10 -7 torr at 373 o K•.

6836 ALVAREZ' LV• S•VL•Og •V D•ELECTR•C P•orsg•ms

1.0

e0 O.Ol o

I ! i

298øKA , A A

•98øK8'• ..... .,• ......

I

I0 10 2 I

10 3 FREQUENCY (Hz)

I

10 4

Fig. 2b.

io 5

rium conditions, the N2 was removed by a re- evacuation, usually to a pressure in the range of 10-•-10 -7 torr, before the dielectric determina- tions were made. The high temperature (i.e., 373øK) was attained by following the same sequence of evacuation, N•. admission, and re- evacuation previously described while heated air was being injected into the liquid nitrogen reservoir (Figure 1). After a cooling or heating process the sample was left in a N2 atmosphere of around 1.0 torr for 8- to 12-hour periods in which room temperature was attained.

DIELECTRIC BEHAVIOR

A description of the material used as the simulator of core S/N 2013 from Apollo 12 appears in Table 1. The dielectric response in the vacuum as a function of frequency, tem- perature, and packing densities of 1.8, 2.0, and 2.2 g/cm • appear in Figures 2, 3, and 4, re- spectively. The experimental error limits have been estimated for K' as --+0.01 at all frequencies and temperatures and for tan 8 as --_+0.001 (at 30 Hz) or better for values of tan 3 less than 0.04, -+0.005 (at 30 Hz) or better for values of tan 3 between 0.04 and 0.09, and --+0.01 (at 30 Hz) or better for values of tan 3 greater than 0.09. Each curve corresponds to one of the temperatures: 100 ø, 298 ø , or 373øK. For a given density a sequence of determinations starts at 100øK and continues to 298 ø and

373øK, and then measurements are repeated at some of those temperatures. To distinguish the order in • sequence, we use a subindex (A, B, or C) in the temperature value. This informa- tion is relevant to the correlation of moisture

losses with dielectric responses. The subindex shall be dropped when generic reference is made to • temperature.

The samples were frozen to 100øKA as de- scribed in the previous section; from the rates of outgassing we inferred that • large portion of the original moisture content of • sample was lost in the high vacuum prior to freezing. At 100øK, one expects the remaining moisture in the sample to be condensed in the form of ice. The dielectric responses of the samples at this temperature have been discussed elsewhere [Alvafez, 1973a]. When the sample was allowed to thaw and reached 298øKA, some extra water losses occurred, owing to additional outgassing in the high vacuum. Heating the sample to 373øK• increased the thermal energy of the water molecules remaining in the pore system of the sample, raising the pressure and forcing additional water to be lost. Therefore, when the sample cooled off to room temperature (i.e., 298øKB), the total moisture content in the pore system was less than it was at 298øK•.

A preliminary idea of how moisture is affect- ing the dielectric response of the sample is gained by comparing the dispersions at 298øK•

ALV^REZ' LU•^R SIMULATOR AND DIELECTRIC PROPERTIES 6837

5.0 I I !

• - /o :2.0

• 4.0 • •373 o

•.o I I I IO IO • IO • IO4 IO5

FREQUENCY (Hz)

•tg. 3•.

for •he sampte of packing de•s•y of 2.0 •/cm •. Temperatures followed •he sequence Z00ø•,, 298ø•,, 373ø•,, ]00ø•, •o 2.5 X Z0

with those at 298øKB; the latter are markedly reduced (Figures 2, 3, and 4), such a reduction being attributed to the lesser amount of mois- ture. After the first cycle of heating, a second cycle (i.e., heating to 373øKB and cooling to 298øKc, as in Figure 3) produced little further change in the dielectric response of the sample,

corresponding to comparatively smaller water losses during the second heating. Our observa- tions agree with other observations [Howell a•d Licastro, 1961; Strangway et al., 1972] that have shown that small amounts of moisture

affect mainly the response of the sample at the 'lower frequencies' (the range of these lower fre-

I.O I i I

?: 2.0

(n O.OI o

o.ool IO 10 2 10 3 10 4 10 5

FREQUENCY (Hz)

Fig. 3b.

6838 ALVAREZ' LUNAR SIMULATOR AND DIELECTRIC PROPERTIES

5.0

>4.0

• 3.0

ø1 •.0 I0

I I I

298"KA •

/o :2.2

2:98 ø KB..,•...... I O0 ø K

IO0øKB

102 I

103 FREQUENCY (Hz)

Fig. 4a.

I

io 4 io 5

Fig. 4. (a) Dielectric permittivity against frequency and (b) loss tangent against frequency for the sample of packing density of 2.2 g/cm •. Temperatures followed the sequence 100øKA, 298øKA, 373øKA, 298øK,, and 100øK,. The pressures ranged from 2.5 X 10 -• torr at 100øK, to 4.0 X 10 -7 torr at 373øKA.

quencies depends on the amount of moisture in the sample and may go as high as I MHz). Notice the small but measurable effect that

moisture release has on the sample at 100øK; Figure 4 shows measurements at 100øKa and at 100øK•, the latter being obtained after a cycle of heating (to 373øKa) and cooling took place.

One must bear in mind in analyzing these results that the same amount of moisture in

the pore system of the sample contributes differently to conduction and dielectric response at different temperatures. At 100øK the mois- ture is frozen, whereas at room and higher tem- peratures the moisture will become water or water vapor depending on the actual pressure

1.0

I- z b.i

z

I-

o 0.01

o. ool

IO

i i i

37 3øKA --

298øKB ,,.., ..,..,,. •,•... •. • -- fO0 ø "' .,...•,.,..•j•

•oc• ø

I0 2 I0 3 i0 4 10 5 FREQUENCY (Hz)

Fig. 4b.

ALVAREZ' LUNAR SIMULATOR AND DIELECTRIC PROrERTIES 6839

I.OO

0.80

0.60

0.40

0.20

I I I I I I I

- .0:5 37 3 • K A, p = 1.8

.O8

.5

K A, p

• p=l.8 •

2.60 2.80 3.00 3.20 3.40 3.60 3.80 4.00 4.20 4.40

K /

Fig. 5. Plots of K' against K" for the distributions at temperatures of 298øKA and 373øKA and packing densities of 1.8 and 2.0 g/cm 3. The distributions are of the Cole-Davidson type, and the sets of parameters describing them appear in Table 2. The numbers by the data points in the distribution of 373øK• and of the density of 1.8 g/cm 3 represent frequencies in kilohertz; to avoid overcrowding, only a few frequencies are indicated.

in the inner pores of the sample. The dielectric and conduction properties of ice and water at the frequencies analyzed are totally different.

Interpretation of dielectric relaxations is fa- cilitated by plotting the data in curves of against •', where •" = •' tan It. In Figure 5 we have plotted such curves for densities of 1.8 and 2.0 g/cm • at temperatures of 298øK• and 373øKA. The plots correspond to Cole-Davidson distributions [Davidson and Cole, 1951] given by

K* = K•'-]-[(•o' --•')/(1 d-jcor) t•] (1) where •* is the complex (relative) dielectric permittivity, •o' and •' are the real parts of •* at zero and infinite frequency, o• is 2•r times the frequency, T is a relaxation time,/9 is a param-

eter that can assume values between 0 and 1, and j = (--1)•/2; the imaginary part of • is denoted as •<". In Table 2 are summarized the

parameters of the Cole-Davidson distributions obtained for temperatures of 298øKA, 373øK•, and 373øKR and for the three densities.

The data in Table 2 suggest that the relaxa- tion time T is controlled mainly by the tempera- ture (in the temperature range of 298ø-373øK), although, as will be seen, moisture also has an effect on ß at 298øK. These data show that

increasing densities yield larger •o' and •' values, as was expected for a given temperature, and suggest that increases in the parameter • occur for increasing temperatures, although the data for the latter observation do not seem to be

conclusive. In addition, decreasing amounts of

TABLE 2. Parameters of the Cole-Davidson Distributions at 298øK A, 373øK A, and 373øK B

Temperature g/cm • sec 8 •0' • , A•' * A•" *

298øK A

373 ø K A

373øK B

1.8

2,0

2.2

1.8

2.0

2.2

2.0

0 OO4

0 004

0 004

0 010

0 010

0 010

0 009

0.25 3.11 2.51 0.03 0.02 0.28 3.88 2.60 0.04 0.05 0.31 4.10 2.83 0.07 0.07 0.35 4.25 2.46 0.02 0.07 0.28 4.57 2.55 0.06 0.04

0.32 4.80 2.81 0.03 0.06 0.29 4.20 2.58 0.04 0.05

*The values for A•' and A•" represent the maximum discrepancy, at any frequency above 30 Hz, between an experimental and a theoretical value at the same frequency. These maximum dis- crepancies occur mainly in the frequency range of 100 to 1000 Hz.

684O ALVAREZ: LUNAR SIMULATOR AND DIELECTRIC I)ROPERTIES

TABLE 3. Parameters of the Two Cole-Cole Relaxation Mechanisms at 298øK B

Low Frequency* High Frequency*

P• T• T•

g/cm 3 10 -6 sec • K0, K•, A•, t A•" ñ 10 -6 sec • •0' •' AK' ñ A•" ñ

1.8 5.6 0.45 2.91 2.48 0.01 0.01 10.8 0.50 2.53 2.43 0.01 0.00 2.0 8.3 0.51 3.38 2.68 0.04 0.02 4.1 0.44 2.7 3 2.59 0.00 0.01 2.2 12.5 0.49 3.63 2.90 0.01 0.01 8.8 0.48 2.96 2.82 0.01 0.00

*Low frequency consists of the frequency range of 30 to 1000 Hz, and high frequency consists of the frequency range of 10 to 100 kHz.

ñThe values for A•' and A•" represent the maximum discrepancy between an experimental value in the low- or high- frequency ranges and the value obtained from the theoretical distribution at the same frequency.

moisture produce decreasing values in K,,' at 298 ø and 373øK; comparison of K,,' values at 298øK.• and 298øK• (see Tables 2 and 3) and comparison of •o' values at 373øKA and 373øK• for a density of 2.0 g/cm '• (see Table 2, low- frequency relaxation) substantiate this observa- tion. In spite of the change in. •,,' between 373øK.• and 373øK• the relaxation time • and

the parameter /• do not experience significant variations, suggesting that moisture has little effect on them at 373øK. The 30-Hz-frequency points (Figure 5) for 373øKA and densities of 1.8 and 2.0 g/cm 3 have been interpreted as arising from a residual electrode impedanc• on the basis of the data plots at 298øK•, which clearly manifest this effect [Dansas et al., 1967].

The curves corresponding to the three densi- ties at 298øK• are plotted as •" against •' in Figure 6. They could not be fitted to Cole- Davidson distributions; instead two Cole-Cole distributions have been fitted to each set of

data [Cole and Cole, 1941]. The analytical ex- pression for such a distribution is

KO ! • Km! = + (2)

1 q- (jo•') 1-" where the symbols correspond to those of (1); recall the fact that the relaxation time has

different, meanings in a Cole-Cole and in a Cole-Davidson distribution and that • is a

parameter that assumes values between 0 and 1. Equation 2 and the parameters shown in Table 3 fit the data within the error limits' and AK") given in the same table. The relaxa- tion times between the low- and high-frequency relaxations differ by 3 orders of magnitude, strongly suggesting different causative mecha- nisms. The •' values of the high-frequency re- laxation at 298øK• are essentially the same as those obtained for the Cole-Davidson distribu-

tions at 298øK•, as was expected from moisture effects.

DISCUSSION

The above observations establish a general pa•tern for the dielectric behavior of the lunar

0.50 [ 0.40[- T -- 298 øK B -- 0.$0[-- p = •.8 z.o z.z -

0.20[-- .03 -

• .40 •. 6 0 •.80 5.00 520 5.40 5.60

Fig. 6. Plots of •" against •' for the distributions at temperature of 298øK• and densities of 1.8, 2.0, and 2.2 g/cm •. Each one is described by two Cole-Cole distributions represented by dotted curves; the corresponding sets of parameters appear in Table 3. The numbers by the data points of the distribution of density of 2.0 g/cm • represent frequencies in kilohertz.

ALVAREZ: LUNAR SIMULATOR AND DIELECTRIC PROrERTIES 6841

simulator and the effects introduced by w•rying (i.e., Figures 5 and • and Tables 2 and 3); in amounts of moisture. The responses at 100øK the distribution at the latter temperature the show small dispersions, with maximums in tan 8 amount of moisture is larger. As it increases at around 300 Hz, associated with the frozen in magnitude, the low-frequency distribution moisture in the sample [Alvafez, 1973a]; at affects higher frequencies, creating a smoother this temperature and in •he frequency range •ransfiion between the low- and the high- studied, there is no evidence that the basalt frequency relaxations. The •response of the powder itself presents relaxation phenomena. At powder becomes overwhelme(• by the moisture 298øK,(Figure 6) it is observed that two effects and the two distributions coalesce into independent relaxations appear' the high- a single one approximately described by a frequency relaxation is attributed to the dielee- Cole-Davidson distribution. The extrapolated tric properties of the powdered basalt. The •' value remains essentially unchanged during low-frequency relaxation is attributed •o inter- the transition, agreeing with expected moisture actions between the moisture in the sample and effects. the powdered basalt as follows: it has been The Cole-Davidson distributions at 373øK• proposed [Alvafez, 1973b] •hat adsorbed water (Figure 5) present the peculiarity of having molecules increase the surface conductiviW of increased their relaxation times with respect, to resistive rock specimens by creating additional the distributions at 298øK• (see Table 2). It is allowed energy levels in the surface of the well known that, when thermally activated sample; in the present case we propose that mechanisms are present, they produce inereas- the low amounts of moisture involved create ing conductivities and decreasing relaxation a series of isolated adsorption centers in which times with increasing temperatures [Saint- the local conductivity value is raised, by the Areant aad Straagway, 1970]. According to the same mechanism, with respect to the surround- increase in relaxation time at 373øK, one would ings. have to rule out the possibility of such a mecha-

The result is a model equivalent to that of nism operating throughout the temperature conductive grains (e.g., spheres) imbedded in range of 298ø-373øK, and yet one would have a resistive matrix that gives rise to a low- to account for the higher conductivities at frequency relaxation of the Debye type through 373øK, as obtained from a : •"o•o, where a Maxwell-Wagner effect [Koops, 1951; Alva- represents the ohmic plus dielectric conduc- rez, 1973c]. The low-frequency response in the tivities, • is 2= times the frequency, present experiments actually corresponds to a tan 8, and.e,,: 8.85 X 10 -• f/m. Cole-Cole distribution of relaxation times in- One could not attribute the increase in con-

stead of a single relaxation time (i.e., Debye ductivity to water effects; at any rate, one relaxation); such an effect is attributed to a would expect a smaller density of isolated distribution of conductivities in the isolated centers of higher conductivity at the higher centers and to the various shapes they must temperature, since water molecules should be acquire. The plausibility of this explanation is excited to higher energies, reducing the popula- brought about by the statistical nature of the tion of adsorbed molecules. Thus it seems rea- adsorption and desorption processes, coupled to sonable to assume that a change in the main the variations in dielectric response induced by conduction mechanism takes place between different shapes of conductive inclusions [Sillars, 298 ø and 373øK. Up to some temperature 1937]. value, above 298øK and below 373øK, the main

When larger amounts of moisture at the same conduction mechanism would be controlled by temperature are present in the pore system, moisture, and from there on a thermally acti- one would expect an increase in the density of vatcd mechanism would predominate, as has isolated centers of higher conductivity. Accord- been described elsewhere [Saint-Amant and ingly, the magnitude of the Maxwell-.Wagner Strangway, 1970; Straagway et al., 1972; effect would be increased and would produce Chuag et al., 1972]. larger •,,' values as well as larger energy losses. The thermally activated mechanism should This result actually occurs when the distribu- cause the overall conductivity of the sample tions at 298øK• and at 298øKA are compared to be increased; in contrast, the water mecha-

6842 ALVAREZ: LUNAR SIMULATOR AND DIELECTRIC PROPERTIES

nism would modify only isolated centers. The thermally activated mechanism should account for the increased relaxation time of the Cole- Davidson distribution at 373øK and should

mask the contribution to energy losses arising from remaining adsorbed water molecules. One has to recognize that in this type of experi- ment the results will always contain some water effects unless considerably higher temperatures are employed (e.g., 600øK). The present data are not sufficient to prove the change in con- duction mechanism proposed; however, the consistency of the results in three independent experiments at three different densities indeed supports the occurrence of such a change.

DIELECTRIC V^RI^T•O•S I• T•E REGOLIT•

Some inferences can be made regarding the dielectric variations undergone by the lunar regolith as a function of temperature and den- sity. These will be, of course, based on the pres- ent results and should be reviewed whenever

similar studies on actual samples of the lunar regolith are available.

For the purpose of analyzing its dielectric variations the regolith can be divided into two layers: a surface layer of around 5-10 cm in depth in which the dielectric response shall be controlled by surface temperature variations and • second layer, beneath the first one and extending down to the basement, in which the dielectric properties will be controlled by den- sity.

The response of the surface layer can be out- lined in consideration of two temperature re- gions: (1) the region between 100 ø and 300øK, in which •' and tan 8 increase slowly with increasing temperature (i.e., corresponding to the present observations in which •' increases up to 6% and tan • increases up to 0.07 from the values at 100øK); and (2) the region be- tween 300 ø and 373øK, showing rapidly increas- ing values of •' and tan • with increasing tem- perature (i.e., corresponding to increases in •' of around 10% at, 10 '• Hz and up to 65% at 30 Hz and increases in tan 3 from 0.04 to 0.08 with respect to the values at 298øK). The limit- ing value of 300øK is only tentative; we be- lieve that it may go up to 325øK.

In any event, equal temperature increments will produce different dielectric variations in the two temperature regions. When these

observations are coupled to the temperature variations in the lunar surface during a lunarion [Robie and Hemingway, 1971], a clearly asym- metric response results for the dielectric prop- erties of the surface layer: it will be fairly constant and will be approximately represented by the response at 100øK during roughly seven tenths of a lunation, and it will undergo rela- tively strong variations in the remaining three tenths of the lunation.

The material below the surface layer (i.e., the second layer) has been found to vary rapidly in density with depth [Mitchell et al., 1972], whereas the surface temperature varia- tions are strongly attenuated [Robie and Hem- ingway, 1971]. Such • situation results in being mostly dependent on density in the lower layer. The relative increments in •', produced by temperature and density increments in the present experiments, also establish the pre- ponderance of density over temperature as the controlling parameter: Figure 7 shows that only below 1 kHz an increment of approximately 300øK in temperature and an increment of 0.4 g/cm '• in density (i.e., in the density range of 1.8-2.2 g/cm 3) produce similar variations in tan • seems to be rather independent of density in the density range analyzed.

We found that the data at room temperature (i.e., 298øK, and 298øK½) do not follow Ray- leigh's mixing formula [Campbell and Ulrichs, 1969] in the frequency range of 30-10 • Hz. The difficulties involved in relating the density of the powdered material to its dielectric re- sponse, when temperature variations are in- volved at frequencies below 10 •, are obvious in Figure 7.

COnCLUSiOnS

A powdered basalt simulating material of the lunar regolith was subjected to temperature variations similar to those found in the lunar surface during a lunarion. To obtain the dielec- tric response under simulated lunar conditions, all measurements were made in vacuums of

2.2 X 10 -• to 4.0 X 10 -• torr, which, in addi- tion, allowed for an evaluation of moisture effects in the powdered sample.

The dielectric responses at 298 ø and 373øK were identified as Cole-Cole or Cole-Davidson

distributions, and their parameter values were determined. Two Cole-Cole distributions were

ALVAREZ' LUNAR SIMULATOR AND •)IELECTRIC PROPERTIES 6843

I I I I I I I 5.00 ,,, •98 •,•

I00 K /•,• ioo e K A , ,. • 298•KB

•.40 L I

_ 373 3.00- I0 K Hz ••8*KA -

2.60-

Z40 I I I I. I I I

I I ! I I I 373

I KHz 3.20 - •98"KA _

// _.-"

2.60 - . - ' ' -

i.60 i.80 2.00 2.20

DENSITY P (•/c• =)

•i• Wm•e•Atu•e A• ffe•ue•e• • •A•Amete•. •eme• • •e•W, •om ].• •o •.• •/•m •, •o-

by temperature •fi•fio• from 1• • to •3•. O•]• •o• ffe•ue•e• belo• 1 • •o •e •em•e•- •u•e e•e• become •om•ble to •e•W e•e•t•

•e •ge• a•ly•e•.

found for each set of data at 298øKB, one corresponding to a high-frequency relaxation (with • in the range of 10 -• sec) and another to a low-frequency relaxation (with • in the range of 10 -3 sec). The former was attributed to the powdered basalt; it was proposed that the latter is caused by a Maxwell-Wagner effect that originates in isolated water adsorption centers; in such centers, adsorbed water mole- cules would raise the conductivity with respect to the uncontaminated surroundings. The data at 298 ø and 373øK suggested • change in the main conduction mechanism taking place be- tween these two temperatures: near 298øK, conduction seems to be controlled by moisture, whereas near 373øK it appears to be controlled by thermally excited carriers.

The lunar regolith was divided into two layers: • surface layer of 5-10 cm deep in

which the dielectric response is thought to be controlled by temperature variations and a subjacent layer in which the variations in •' are thought to be controlled by density; in this layer the tan 8 values would probably be slowly varying functions of density. An asymmetry in the dielectric properties of the surface layer during a lunation was suggested: during ap- proximately seven tenths of a lunarion this layer would show a slowly varying behavior in its dielectric properties, undergoing rapid varia- tions in the remaining three tenths of the luna- rion.

Acknowledgments. The author acknowledges A. Dey, G. R. Olhoeft, and H. F. Morrison for comments on the manuscript; D. W. Strangway for providing the lunar simulator; and B. Jain for help in the laboratory. This work was carried out under NASA grant NGR 05-003-447.

P•EFERENCES

Alvarez, R., Lunar permafrost' Dielectric identi- fication, Science, 179, 1122, 1973a.

Alvarez, R., Effects of atmospheric moisture on rock resistivity, J. Geophys. Res., 78, 1769, 1973b.

Alvarez, R., Complex dielectric permittivity in rocks: A method for its measurement and

analysis, Geophysics, 38, in press, 1973c. Campbell, M. J., and J. Ulrichs, Electrical prop-

erties of rocks and their significance for lunar radar observations, J. Geophys. Res., 74, 5867, 1969.

Chung, D. H., W. B. Westphal, and G. Simmons, Dielectric properties of Apollo 11 lunar samples and their comparison with earth materials, J. Geophys. Res., 75, 6524, 1970.

Chung, D. H., W. B. Westphal, and G. Simmons, Dieletric behavior of lunar samples: Electro- magnetic probing of the lunar interior, in Pro- ceedings o• the Second Lunar Science Confer- ence, edited by A. A. Levinson, p. 2381, MIT Press, Cambridge, Mass., 1971.

Chung, D. H., W. B. Westphal, and G. R. Olhoeft, Dieletric properties of Apollo 14 lunar samples, in Proceedings o• the Third Lunar Science Conference, edited by D. R. Criswell, p. 3161, MIT Press, Cambridge, Mass., 1972.

Cole, K. S., and R. H. Cole, Dispersion and ab- sorption in dielectrics, J. Chem. Phys., 9, 341, 1941.

Cremers, C. J., Thermal conductivity of Apollo 14 fines, in Proceedings o• the Third Lunar Science Conference, edited by D. R. Criswell, p. 2611, MIT Press, Cambridge, Mass., 1972.

Dansas, P., P. Sixou, and R. Arnoult, Utilisation de l'effet Maxwell-Wagner pour l'(•tude de di•lectriques pr•sentent des pertes par relaxa- tion dipolaire et une forte conductivitY, in XIV

6844 ALVAREZ: LUNAR SIMULATOR AND Dm•,r•CTR•C PRofracTures

Colloque Amp&'e, edited by R. Blinc, p. 647, North-Holland, Amsterdam, 1967.

Davidson, D. W., and R. H. Cole, Dielectric re- laxation in glycerol, propylene glycol, and n- propanol, J. Chem. Phys., 19, 1484, 1951.

Gold, T., M. J. Campbell, and B. T. O'Leary, Optical and high-frequency electrical properties of the lunar sample, Science, 167, 707, 1970.

Gold, T., B. T. O'Leary, and M. Campbell, Some physical properties of Apollo 12 lunar samples, in Proceedings o• the Second Lunar Science Con]erence, edited by A. A. Levinson, p. 2173, MIT Press, Cambridge, Mass., 1971.

Howell, B. F., Jr., and P. H. Licastro, Dielectric behavior of rocks and minerals, Amer. Mineral., 46, 269, 1961.

Hoyt, H. P., Jr., M. Miyajima, R. M. Walker, D. W. Zimmerman, J. Zimmerman, D. Britton, and J. L. Kardos, Radiation dose rates and thermal gradients ia the lunar regolith: Ther- moluminescence and DTA of Apollo 12 samples, in Proceedings o• the Secon•d Lunar Science Conference, edited by A. A. Levinson, p. 2245, MIT Press, Cambridge, Mass., 1971.

Katsube, T. J., and L. S. Collett, Electrical properties of Apollo 11 and Apollo 12 lunar samples, in Proceedings o[ the Second Lunar Science Con]erence, edited by A. A. Levinson, p. 2367, MIT Press, Cambridge, Mass., 1971.

Koops, C. G., On the dispersion of resistivity and dielectric constant of some semiconductors at audio frequencies, Phys. Rev., 83, 121, 1951.

Mitchell, J. K., W. N. Houston, R. F. Scott, N. C. Costes, W. D. Carrier III, and L. G. Brom- well, Mechanical properties of lunar soil; density, porosity, cohesion, and angle of internal friction, in Proceedings o[ the Third Lunar Science Con]erence, edited by D. R. Criswel!, p. 3235, M!T Press, Cambridge, Mass., 1972.

Robie, R. A., and B. S. Hemingway, Specific heats of the lunar breccia (10021) and o!ivine dolerite (12018) between 90 ø and 350øK, in Proceedings o! the Second Lunar Science Con]erence, edited by A. A. Levinson, p. 2361, MIT Press, Cam- bridge, Mass., 1971.

Saint-Amant, M., and D. W. Strangway, Di- electric properties of dry, geologic materials, Geophysics, 35, 624, 1970.

Sillars, R. W., The properties of a dielectric containing semiconducting particles of various shapes, J. In•t. Elec. Eng., 80, 378, 1937.

Strangway, D. W., Moon: Electrical properties of the uppermost layers, Science, 165, 1012, 1969.

SI.rangway, D. W., G. R. Olhoeft, W. B. Chap- •nan, and J. Carnes, Electrical properties of lunar soil-dependence on frequency, tempera- ture and moisture, Earth Planet. Sci. Lett., 16, 275, 1072.

(Received February 28, 1973; revised July 12, 1973.)