Low pressure radon diffusion: a laboratory study and its implications for lunar venting

Geochlmica et Cosmochlmica Acta, 1976, Vol. 40. pp.375 lo 380. Pergamon Press. Prmted m Great Bntam

Low pressure radon diffusion : a laboratory study and its implications for lunar venting LARRY JAY FRIESEN

Department

of Geology, University of Georgia, Athens, Georgia 30602. U.S.A. and JOHN A. S. ADS

Department

of Geology, Rice University, Houston, Texas 77001, U.S.A.

(Received

17 April 1975; accepted in reuised form 25 August 1975)

Abstract-Results of a study of radon migration through columns of fine particulate materials, at total pressures of 2-20 x lo-* torr, are reported. Materials studied were: NBS Glass Spheres (SRM 1003) Emerson & Cuming Eccospheres (IG-lOl), activated coconut charcoal, Lipari obsidian, and W-l Standard Diabase. Rates of diffusion were used to derive heats of adsorption for radon on the materials tested. The most reliable values found clustered around 8-9 kcal/mole. These high heats of adsorption, if typical for most materials, combined with low percentages of radon emanation by lunar soils found by other researchers, imply that random walk diffusion will not be an important mechanism for redistributing the radon and the radon daughters produced in the lunar regolith. In particular, since random walk migration is not a sufficient mechanism to account for localized high concentrations of radon-222 and its daughter polonium-210 observed by the Apollo 15 and 16 command modules, an alternative mechanism is proposed, in which radon would be swept to the surface by other gases during intermittent venting events.

INTRODUCTION

where R is the universal gas constant, Tis the Kelvin temperature, and M is the gram-molecular weight of radon; (b) the mean free path 1, which will be some mean void dimension (for spherical particles

THE PURPOSE of this paper is to report

results of a laboratory study of radon diffusion through soil moderate vacuum conditions columns under (FRIESEN, 1974), and to point out implications these results have for radon migration on the Moon and

&X 1 -P’

(2)

for lunar gas venting. where ? is the mean radius of the spheres and P is The initial motivation of this study was to test a the porosity); and (c) the mean ‘residence time’ that theoretical model which had been proposed for the a radon atom remains adsorbed each time it diffusion of radon through the lunar regolith by encounters a solid surface FRIESEN and HEYMANN(1972) and FRIE~EN(1972). In r = raeQ/Rr particular, values were sought for the heat of adsorp(3) tion of radon on a wider variety of materials than where Q is the heat of adsorption and r,, is a constant had been previously tested (G~~BELIand STAMMBACH, dependent on the particular solid under consideration 1951; GYJBELIand ST~~RI,1954), and to obtain those (DE BOER, 1953). [TO lies within the range of values under vacuum conditions. 10-14-10-‘2 set for most solids. As it is most freThe model being tested is one in which radon quently close to lo-l3 set (DE BOER, 1953), this value atoms are viewed as undergoing a random walk has been assumed here for all materials studied for through a baffle system, where the baffles are the surwhich no precise values could be calculated or found faces of particles. Under conditions prevailing within in the literature.] The full expression for the diffusion the soil near the lunar surface, the mean free path coefficient in the model is for encountering other gas molecules is many orders of magnitude longer than the mean free path for 1 12 D=-(4) encountering soil particle surfaces. This is not a ‘bil3 7 + ,ljo liard ball’ model. Each time a radon atom encounters (FRIESEN, 1972). a soil particle surface, it is adsorbed, and reemitted at some later time in a random direction. The diffusion coefficient in this situation the mean thermal speed

will depend

on (a)

APPARATUS, PROCEDURE, AND ANALYSIS In order to test this model, it was necessary to study the diffusion of .radon through soil or other particulate material, under circumstances where the mean free path

(1) 375

L. J. FRIESENand J. A. S. ADAMS

376

t

To

atmo.sDhere

Rubber hose

Countmg bottle

L

Power supply AEC-320-3

/

Fig. 1. Schematic diagram of diffusion line and counting system. BV I. 2. and 3: stainless steel ball valves. SwT’s: Swagelok T-connections. modified by addition of pipethread connections to accept gauge tubes TGTl and 2 of thermo~upIe pressure gauges PGI and 2. UK’s: Ultra-torr Unions. RnSC: radon source container. SV: solenoid valve. ST: sample tube. containing particulate sample column. fgt: short glass tube with one end flared to fit counting bottle. PM: photomultiplier. SCA: single channel analyzer.

was closed, and the sample tube side was pumped down by the same procedure to a pressure matched as closely as possible to that on the RnSC side (which had often increased in the meantime due to system outgassing). Initial pumping of each side was through the needle bypass valve to prevent sudden pressure changes from blowing material out of the uranium ore or the sample column. As soon as pumpdown was complete, BV2 and BV3 were both closed, the pump was shut off, and a background count was taken with solenoid valve closed. To saveguard against overloading the exposed photomultiplier and minimize spurious counts. all runs were carried out in a low-level counting lab entered through an airlock. The double doors of the airlock act as a light seal as well as a pressure seal. Counts were made with laboratory lights switched out and all light sources on the equipment masked. Experimental runs were conducted with the counter operating on the Recycle mode. in which it would initiate a new count 1 see after each previous count had been completed. During some runs, the counter was switched on and the solenoid valve switched open simultaneously, and the solenoid valve left open during the entire run. During others. the solenoid valve was opened for 60sec. closed again, and the counter switched on immediately thereafter. The value of vZ used in the analysis depends on which of these procedures was used. Although pressures obtained during these experiments were far above the low lunar vacuum, they did meet the criterion requiring the mean free path for gas molecule collisions to exceed that for soil particle surface encounters. The highest pressure during any experimental run was _ 0.2 torr. and most runs were done at pressures less than 0~1 torr. At 0.2 torr. the mean free path for radon atoms against co&ions with other gas molecules is in excess of 15Ojtm. The only case where the mean free path against encounter with particle surfaces might approach this is the W-l sample described below. In no other instance are the two mean free paths even within an order of magnitude of each other. The runs reported here were performed at room temperature. Runs with the sample tube cooled by dry ice were attempted. but the results were confused and inconsistent, and have not been included here. In analyzing the data, the radon source container plus the tubing leading to the sample column are viewed as a reservoir of volume V,, having a concentration of radon cl. The flared glass tube and counting bottle make up a reservoir of volume V, and radon concentration cL. The two reservoirs are connected by the sample tube (whose axis is defined as the .x-direction) of length (or column height) L and internal radius r. having concentration of radon c(.x,ti. The diffusion in this situation reduces to the one-dimensional diffusion in the .x-direction. The flux of radon at any point in the column and any time t (I = 0 being defined as the instant at which the solenoid valve is opened), J(x,t), can then be written

for particle surface encounters was shorter than the mean free path for gas molecule collisions. To accomplish this, a simple vacuum diffusion line was constructed (see Fig. I), Used with this diffusion line was an alpha-particle counting system, also shown in Fig. 1. Explanation of system component labels is included in the caption to Fig. 1. Tubing in the line is stainless steel, : in. outside diameter, except for the bypass tubing to the needle valve. which is gin. outside diameter. The sample tube (ST) is a length of tubing with a disk of wire mesh at one end to support the particulate sample column. Interior diameter = I.080 cm. interior depth = 19.6 f @03 cm. The radon source used throughout this study was 48 g of the following uranium ore standard: New Brunswick Laboratory, Atomic Energy Commission, New Brunswick, N.J., Analyzed Sample No. 73. Uranium Ore, 1.0’; U. Experimental operation of the equipment was as follows: samples tested were cleaned prior to loading by repeated washings with acetone and/or methyJ alcohol to remove any contaminants, especially water (the charcoal was not washed, however, to avoid any chance of deactivating it). Porosity of the sample was determined by comparing column volume with volume of an equal weight of ‘solid’ ,ln our analysis, we have approximated this differential equation by the difference equation sample. (The liquid displacement density of the IG-101 hoilow spheres, described below, was determined by the manufacturer. We determined the liquid displacement density for a sample of the charcoal in our 1aboratory.t The sample tube was then connected into the system and evacuation begun. With solenoid valve and BV3 closed, BV2 was opened and the radon source side of the As the run durations were many times shorter than the system pumped down. Pumpdown normally continued unhalf-life of radon-222 (for practical purposes, the only isotil pressure on the RnSC side read = lO--’ torr. Then BV? tope we are dealing with: see explanation below), the

Low pressure number of radon atoms in the system for the duration of any run is considered to be a constant, I. No = c,(t)r/,

C&)V2+

+

I0

&Pc(x,

t) dx.

[P = porosity, defined as in equation (2).] Since all the radon is initially in the RnSC system, one has the initial condition

(6)

side of the

c,(O)1/, = N”,C,(O) = c(x,O) = 0. For continuity,

one requires

cl(t) = Fc(O,t) and c&) = Fc(L,t), where F is the fraction of radon atoms column which are in the gas phase, rather at any given time. F=

average average n/G

adsorption

in the sample than adsorbed,

flight time

time + average

flight time

30

? + n/o

IV

With these conditions

and definitions,

radon

377

diffusion

for t, for an empty sample tube at a corresponding pressure (this is described in more detail further on). When one has a value for K, one obtains the diffusion coefficient D by using equation (7a). Then z is calculated using equation (4). One can then use a known or assumed value for r0 to calculate Q, the heat of adsorption, from equation (3). One benefit of this technique of monitoring the diffusion by radioactive counting is that one need not worry about interference caused by inadvertently measuring other gases than radon. The only substance in this system which is both gaseous and alpha-radioactive is radon. Furthermore, we are dealing almost exclusively with radon-222. The 4-set half-life of radon-219, daughter of uranium-235. is so short compared with t, for even an empty-tube run that it can be neglected. The amount of thorium-232, which produces radon-220 (half-life 51.5 set), in the uranium ore standard is negligible. Another benefit of this method is that it permits a study of the diffusion of extremely small quantities of radon. Useful results were obtained with plateau count rates as low as 1 count/set. This corresponds to a total number of radon-222 atoms in the counting bottle on the order of 5 x 105, or only -2 x lo6 radon atoms in the entire system.

we can write SAMPLES

1

’

z

s0

c’#) = v

J(L,@&P

de

s

= nr2f’DG’) *e_KBde LVZ =

nr’PDc,(O) LV,

0

(1 - e-K’) _’

Finally, we expect that as t + tends to become uniform throughout i.e.

cl(~)

= cz(co) = Fc(x,w)

= L

(7)

K cc, the concentration the system, J(L,t)srzP

dt

No = V, + V, + rrr=LP/F Evaluating K =

equation

(7) at t = co,

F?(F)

+(~w~P)~(&).

(7a)

For runs where the solenoid valve was left open for the entire run, V, is 250cm3. For cases where it was opened for 6Osec only, V, is 70cm3 (except for the charcoal sample, when it was 88 cm3). V, in all cases is 100 cm3, and r is 0540 cm. Values for L and P depend on the individual sample and will be noted when the samples are described. In order to find K for a given experimental run, one determines the length of time required for the count rate to reach the midway point between the background level and the plateau rate it was approaching. Defining this ‘rise time’ as t, and assuming that the approach to a plateau is exponential, K = 0.693/t,.

(8)

So far this approximation had ignored the delay time caused by the system rather than the sample, i.e. the delay between the instant the solenoid valve is opened and the time that the radon is equilibrated throughout the volume above the top of the sample column. In practice, t, is corrected for a given run by subtracting from it the value

TESTED

The particulate samples investigated in this study were: A. NBS Standard Reference Material 1003, Calibrated Glass Spheres (%30pm). SRM 1003 and the Eccospheres (below) were chosen in part because lunar material contains a great deal of glass, and in part because the sphericity of the particles defines the geometry very well. The mean free path of the random walk can be calibrated precisely, using equation (2), and thus we have a ‘calibration point’ of sorts. Specific-surface mean diameter for SRM 1003 = 14.5 f 04pm. r. for this material (assumed) = lo-l3 sec. L = 18.35 k 0.07 cm. P = 0.469. L = 6.39 pm. B. Emerson & Cuming Industrial Grade Eccospheres, IG-101. These are hollow glass microspheres. Composition: sodium borosilicate glass. Mean particle radius, by number : particle - 32 pm. L = 18.63 k O@tcm. P = 0.395. A = 21 pm. z. (assumed) = lo-l3 sec. C. Activated coconut charcoal. This was selected because estimates for Q used in the theoretical model (FRIESEN, 1972) had been based on an extrapolation from the heats of adsorption of other noble gases on graphite, which has the same to (DE BOER, 1953). Both the charcoal and the Lipari obsidian (below) were ground and sieved through wire mesh filters. Particles selected for the samples had diameters between 32 and 69 pm. The arithmetic average of these two extremes was used as the ‘mean diameter’ in computing I for the charcoal and the obsidian. Although these particles are not exact spheres, it was assumed that equation (2) was a satisfactory approximation for all materials tested. If the charcoal particles possess pores with average dimensions much smaller than the average interparticle dimension, 5 z, to, and Q for charcoal are all overestimated. L = 0.33 k O@lcm. P = 0.12. i, = 3.5 pm. r. = 5 x 1Ol4 set (DE BOER, 1953). D. Lipari obsidian. This is a volcanic glass found on the island of Lipari. It possesses an above average content of potassium, uranium, and thorium, but exhibits little emanation of radon (the latter property it shares with lunar materials). As a natural glass, it provides a checkpoint for comparison with the artificial glasses above. L = 17.7 k O+lcm. P = 0.458. 1 = 21 pm. ~~ (assumed) = lo-l3 sec. E. Standard Diabase. This was selected because lunar returned samples contain a great deal of basaltic material. In this instance, only an upper limit for mean particle size is available, and thus only an upper limit for mean free

L. J. FRIESEN and J. A. S. ADAMS Inltlal pressure=167 x lO-3torr

x x

Fig. 2. Sample

RI%? t,me=70

-

room

temperature tube.

set

Ihe time - 2100 set

run with empty

sample Background

path is i. I 127pm.

known. L = 11.1 f 0.14cm. P = 0.666. T,, = I.5 x lo-‘” set (FRIFSEN.1974). DATA

The data for sample experimental runs are shown on Figs. 2. 4 and 5. In these graphs. the number of counts recorded in each time interval is plotted vs elapsed time since the counter was switched on. Horizontal bars represent experimental values. Values for an exponential approach to a plateau calculated in an attempt to match the experimental count rate curve of a given graph are plotted on that graph as Y’S, The calculated rise time t, is noted on each graph. ‘Room temperatures’ were those in the lab at the times of the runs. Figure 3 is a plot of the rise times for empty-tube runs against the initial pressures. Room temperature runs are plotted here as x’s; dry ice runs are square boxes. With one exception. rise times for room temperature runs appear to be roughly proportional to the initial pressure: f, (in seconds) = 0.48 x initial pressure (in units of lo-’ torr). Values calculated from this proportionality are used in the analysis of the data to correct for the system delay. Although system pressures did rise slightly during some experimental runs, the changes were never large, and it is the initial pressure. during the first few seconds of the run. that must be known to compute to correction for system delay. The analysis for dry ice temperatures is analogous. but will not be presented here.

CONCLUSIONS

AND IMPLICATIONS

Values found for room temperature diffusion coefficients and heats or adsorption for the samples tested are summarized in Tables 1 and 2. The values derived

= 178 counts

0

1400

set 10.000

5000 Time,

SBC

Fig. 4. Sample room temperature run with SRM 1003. Calibrated Glass Spheres. Background count rate is normalized to the counting interval used during that run.

for heats of adsorption at room temperature are probably more accurate than the data on which they are based. Two separate room temperature runs with charcoal resulted in values for D that differed by a factor of 3, and yet in values for Q differing by no more than lo?,,. This is a result, of course, of the fact that Q depends on the logarithm of t. It will be noted that heats of adsorption for the materials tested cluster around 8 or 9 kcal/mole. This is much higher than a value extrapolated from the heats of adsorption of argon. krypton, and xenon on graphite of 52 kcal/mole by FRIESEN(1972). It is also higher than values found by GCJBELIand STAMMBACH (1951) for the heats of adsorption of radon on silica gel and on activated charcoal, and is in the range of the very highest values found by G~BELI and ST~RI (1954) for the heat of adsorption of radon on activated charcoal (the values found by Giibeli and St&i ranged from 2510 to 9.455 kcal/mole, depending on the carrier gas and temperature). High heats of adsorption imply that redistribution of radon and its daughters by random walk is a much less important process than the theoretical model had originally led us to believe. In one calculation. changing the heat of adsorption assumed from 5.2 to 9 kcali mole reduces the probability that a radon atom Inmal ‘loco

pressure s 59 x 10.”

torr

1

*.

*

(

x_

-

-;-

--_-

r---

0

60

180

120

Imttalpressure (xIO-~ torr) OS read

on PGI

Fig. 3. Plot of rise times vs initial pressure for empty sample tubesat room temperature and at dry ice temperature. x ‘s are rise times vs initial pressure for room temperature runs with an empty sample tube. Square boxes are for dry ice runs.

0

1 5000

I 10.000 Tlrrm.

Fig. 5. Sample room temperature

I 15,ooo

set

run with Lipari obsidian.

Low pressure radon diffusion

319

Table 1. Summary of room temperature values for radon diffusion coefficient. Runs noted with * for charcoal, Lipari obsidian, and W-l diabase are those in which the solenoid valve was opened for 6Osec only. The solenoid valve was open for the entire duration of the run in all other cases. For SRM 1003 and IG-101 Eccospheres, values are averages for all room temperature runs

Material

SRM 1003

IG-101

Charcoal

Lipari obsidian

299

296

291 461

296 1970

Temperature (“R) Corrected rise time, r, (set) K (set-‘)

2140

968

3.24 x 1O-4

I.16 x 1O-4

D (cm’jsec)

0.91

2.4

released at a depth of 4m in the lunar regolith will reach the lunar atmosphere before decaying from 30 to 6.6%. The unimportance of random walk as a redistribution mechanism is further indicated by emanation experiments, which indicate that very little of the radon (typically a few per cent or a few tenths of a per cent) produced within a sample of lunar soil is ever released into the soil voids to begin a random walk (STOENNERet al., 1971, 1972; YANIV and HEYMANN, 1972). BARRETTO(1972) has found that the reasons for this low emanation ratio are apparently that lunar materials are not subject to chemical weathering, and that their level of radiation damage (especially alpha and nuclear recoil tracks) is low compared with most terrestrial materials. Thus the mobility of radon on the Moon is doubly impeded: few radon atoms can find their ways from the mineral grains in which they are formed to the interparticle void spaces, and those that reach the voids seldom migrate far because of the high heats of adsorption. The alpha spectrometers carried on board the Appollo 15 and 16 command modules observed alpha particles from lunar orbit, produced by the disintegration of radon-222 and its daughter isotope polonium-210 on the lunar surface and in the lunar atmosphere (BJORKHOLM et al., 1973a,b; GORENSTEIN and BJORKHOLM, 1973; BJORKHOLM and GORENSTEIN, 1974). Strong local enhancements of these two isotopes were observed, well above the lunar average. The locations of these enhancements are not correlated with gamma-ray activity indicating the distribution of lunar uranium thorium, and potassium (Proceedings of the Fourth Lunar Science

Conference,

871* 1.5 x 1o-3 8.0 x 10-4* 0.32 0.11*

1130* 3.5 x 1o-4 6.1 x 10-4* 0.81 0.54*

W-l diabase 296 1230 1180* 5.6 x lo-“ 5.9 x 10-4* 0.72 0.42*

Plates II and V). Furthermore, the count rates of radon-222 and polonium-210 are not in equilibrium with each other at the enhancement sites, and the count rate ratio of these two isotopes varies from one ‘hot spot’ to another. These facts, coupled with the low emanation rates for radon one expects from a random walk, seem to demand some sort of ‘sweeping mechanism’, where other gases are vented to the surface from great depth, picking up radon as they go. The disequilibria between radon-222 and polonium-210, which are separated on the decay chain by the isotope lead-210, with a twenty-plus year half-life, implies that the rate of emission of radon at the enhancement sites has changed on a time scale comparable with the lead-210 half-life, or perhaps even more rapidly. A nonsteady rate of emission further argues against random walk as being a sufficient mechanism to account for the ‘hot spots’. It is interesting to note that the locations of many of these enhancements, the craters Aristarchus and Grimaldi and the edges of ring-shaped maria, correspond to locations most often reported by astronomers as sites of unusual optical activity (peculiar brightenings and obscurations) on the Moon (MIDDLEHURST, 1967). Perhaps in these optical phenomena we are seeing evidence of the gas venting which is sweeping radon (among other volatiles) to the surface. Since one finds that such peculiar optical events have most often been reported when Earth-generated tidal forces are maximum and minimum (MIDDLEHURST, 1967), a periodicity shared by deep lunar quakes (LATHAMet al., 1972), we suggest that either

Table 2. Summary of room temperature values for radon heat of adsorption Material Temperature (“R) r at this temperature (set) Q (kcal/mole)

SRM 1003

IG-101

Charcoal

Lipari

W-l

299

296

291 1.1 x lo-’ to 3.5 x lo-’ 869.3

296

296

2.0 x 1o-6

59.4 x 10-a

9.9

$14.6

1.1 x lo-’ 8.2

4.9 x lo-’ 9.1

380

L. J. FRIESEN and J. A. S. ALIAMS

the quakes themselves, or the Earth’s tidal stress (which is the most probable trigger for the quakes), may be the mechanism which triggers the release of episodic bursts of gas to the surface from deep within the Moon. Acknowledgements-We wish to thank ELZE HEMMEN for constructing the diffusion line used throughout these cxperiments, and Dr. DAVID REASONER and Dr. FREDERICK ROSSINI for discussions held with them regarding the experiments and their analysis. We wish to thank also RODERICK DUNN, Dr. B. FRANK JONE$ and Dr. DAVID RECTOR for helping us gain insight into the behavior of the diffusion equation under conditions such as those in the experiments. This research has been supported by NASA grant NCR-44-006-142 and Welch grant C-009.

REFERENCES BARRETTO P. M. C. (1972) Emanation characteristics of terrestrial and lunar materials and the *“Rn loss effect on the I_-Pb system discordance. Ph.D. Thesis, Rice University. BJORKHOLM P., GOLUB L. and GORENSTEIN P. (1973a) Detection of radon emanation from the lunar regolith during Apollo 15 and 16. In Luptar Science IV. (editors J. W. Chamberlain and C. Watkins). p. 78. The Lunar Science Institute, Houston. BJORKHOLM P., GOLUB L. and GORENSTEIN P. (1973b) Detection of a nonuniform distribution of polonium-210 on the Moon with the Apollo 16 alpha particle spectrometer. Sciencr 180,957-959. BJORKHOLM P. J. and GORENSTEIN P. (1974) Variation of “‘Rn to “‘PO activity ratio on the lunar surface as observed by the alpha particle spectrometer. In ~u,iur Science V. Vol. I, p. 283. The Lunar Science Institute, Houston.

I)E BOER J. H. (1953) Thr Dvnumicul Churucter of Adsorotion. pp. 30-36. Oxford University Press. FRIESEN L. J. 11972) A theoretical studv of radon diffusion in the lunar regoiith. M.S. Thesis. Rice University. FRIESEN L. J. (1974) Radon diffusion and migration at low pressures. in the laboratory and on the Moon. Ph.D. Thesis. Rice University. FRIESEN L. J. and HEYMANN D. (1972) Model for radon diffusion through the lunar regolith. The Moon 3, 461~~471. GORENSTEIN P. and BJORKHOLM P. (1973) Detection of radon emanation from the crater Aristarchus by the Apollo I5 alpha particle spectrometer. Scirnw 179, 792 -794. CUBELI, 0. and STAMMBA(.HK. (1951) Zur Adsorption von Radon an Aktivkohle und Silicagel. Hclr. Chitn. .-lcrrr 34. 1257

1263.

G~~HLLI0. and STiiRi M. I 1954) Zur Mischadsorption von Radon an Aktivkohle mit verschiedenen TrHgergasen. He/r. Chirn. .4ctu 37, 22242230. LATHAM G.. EWING M.. DORMAN J., LAMMLEIN D.. PRESS F.. Tolisoz N., SUTTON G.. DUBNNEBIER F. and NAEA?+XRAY. (I 972) Moonquakes and lunar tectonism results from the Apollo passive seismic experiment. Pruc. 3rd Lww SCi. Cw$, Geochim. Co,srnochim. Actu Suppl. 3, pp. 2519 2526. M.I.T. Press. MIDDLEHURSTB. (1967) An analysis of lunar events. Rer. Geophys. 5, 173 189. Prrx. 4th L.iriltrr Sci. Con/.. Grochim. Cosmochim. .3ctu Supp/. 4 ( 1974) Plates II and V. Pergamon Press. STOENNER R. W.. LYMAN W. and DAVIS R., JR. (1971) Radioactive rare gases and tritium in lunar rocks and in the sample return container. Proc. 2nd Lunur Sci. Conf.. G~wchirn. Cosntochirw .4cto Suppl. 2, pp. 18 13-I 823. M.I.T. Press. YANIV A. and HEYMANN D. (1972) Radon emanation from Apollo 1I. I?. and 14 fines. 3rd Lunur Sci. Conf .4hstrucrs. pp. 718.. 720. The Lunar Science Institute, Houston.