Physics of the Earth and Planetary Interiors, 14 (1977) 299—320 299

© Elsevier Scientific Publishing Company, Amsterdam — Printed in The Netherlands

POSSIBLE PHYSICAL PROCESSES CAUSING TRANSIENT LUNAR EVENTS

J.E. GEAKE and A.A. MILLS Department of Pure andApplied Physics, The University of Manchester Institute of Science and Technology (UMIST), Manchester (Great Britain) Departments of Geology and Astronomy, University of Leicester, Leicester (Great Britain)

(Revised and accepted February 10, 1977)

Geake, J.E. and Mifis, A.A., 1977. Possible physical processes causing transient lunar events. Phys. Earth Planet. Inter., 14: 299—320.

Transient lunar events appear to involve two main effects: the obscuration of surface detail, and changes in bright- ness and/or colour which could be caused either by modification of the way in which incident sunlight is scattered, or by the emission of additional light. We find it difficult to explain the obscurations in any other way than to assume that clouds of fine surface dust are raised either by bursts of gas emission from surface fissures, or by impacts; the possible duration and density of such clouds are considered. Modification of the albedo of a dust surface by agitation has been demonstrated in laboratory experiments: un- der certain conditions the albedo may increase, but the change appears to be permanent at atmospheric pressure; it may be reversible under lunar vacuum conditions. The most likely lunar process of this type again seems to be the agitation of surface dust by gas emitted from fissures; also, the scattering of sunlight by dust clouds could, under some conditions, result in weak colour effects. Processes that could result in the emission of light include incandes- cence, luminescence or thermoluminescence, glow discharge in gas clouds (possibly enhanced by the presence of charged dust grains), and lightning-type discharge in dust clouds. We conclude that the lightning-type discharge is the process most likely to be bright enough to be visible from earth, against the sunlit . We therefore conclude that transient lunar events of the different types that have been reported could be explained by various processes that may occur in gas-borne dust clouds.

1. Introduction (usually red or blue), and sometimes more strongly coloured (usually red); a few have been described as There have now been over 1,400 reports of tran- red and sparkling (e.g., Herschel, 1787; Greenacre and sient changes on the moon’s surface, as observed by Barr, 1963). Some star-like points have also been re- telescope from the earth (Cameron, 1976). While it is ported. possible that many of these are due to instrumental, None of these events have left visible permanent atmospheric, or physiological effects, it seems probable char?~es.The lateral scale of typical events appears to that some are real, and it is the purpose of this contribu- be of the order of tens to hundreds of kilometres, al- tion to consider which physical processes offer possi- though ones less than about 10 km across would prob- ble explanations. ably not be noticed (apart from very bright point Two main types of effect have been reported as oc- flashes). Their duration is usually from minutes to an curring either separately or together: hour or two, but they would probably need to last for (1) Obscurations, involving blurring or loss of sur- a fewminutes to be confirmed. Correlations have been face detail, and loss of constrast. found — in time with lunar perigee (Middlehurst, (2) Changes, usually increases, of brightness. These 1968), and in position, especially with the edges of are sometimes colourless, sometimesweakly coloured maria (Middlehurst and Moore, 1967) — although 300

Fig. 1. Photographs of the crater taken by Alter in 1956, showing slight obscuration of surface detail in (B) as com- pared with (A) — particularly of an undulating rille in the upper right-hand part of the crater floor. The two photographs were taken with different filter/plate combinations, and therefore used different wavelength bands, centred in the near IR for (A) and the blue for (B). The blue-light picture is less clear all over, because of light scattering in the earth’s at- mosphere, but Alter considered that features in Aiphonsus were further obscured on the blue-light plates by some additional local cause, as compared with features in nearby craters. The effect was present on four pairs of photographs, taken over 1.25 h under excellent seeing conditions; no blue-light photograph was shown that did not show the obscuration. North is at the top. (From Alter, 1963).

both these correlations are weakened by the possibility (b) Modulation of incident sunlight due to: of selective observation, once they had been announ- (i) Agitation of surface dust by gas emission or ced. by electrostatic effects. Several physical mechanisms for transient events (ii) Scattering by dust clouds. have been proposed, including: One of the main difficulties in trying to assess these (1) As causing obscuration: mechanisms is the shortage of physical measurements (a) Scattering of light by dust clouds, or instrumental records of transient events. Those (b) Gas fluorescence, i.e. gas released from the sur- available include Alter’s 1956 photographs (Alter, 1963) face and excited by solar UV or X-rays to give a veil of which suggest a slight temporary loss of detail on the light, resulting in loss of contrast, floor of the crater Alphonsus (Fig. 1),and some further (2) As causing changes of brightness: photographs listed by Cameron (1976). More specific (a) Emission of light by: evidence is provided by Kozyrev’s 1958 spectrum (i) Luminescence (including thermolumines- which he ascribed to C2 gas emission from the central cence) excited by solar radiation or particle emission. peak of the same crater, as shown in Fig. 2 (Kozyrev, (ii) Incandescence of surface material, heated in 1961); also, Kozyrev (1963) later obtained other some way. emission spectra, including an H2 emission spectrum (iii) Electrical glow or spark discharges in gas- for the crater . Kopal and Rackham (1963) borne dust clouds, photographed an apparent brightness change around 301

g~ 9 *

4000— -

4100

4200— -

t _ 4300~~

-- r— - 4400

4500—

6600 —

4700 —

4800—

4900— ______5000— 5100—

B

4677 4714 4737 Fig. 2. An emission spectrum obtained by Kozyrev in 1958, and ascribed to gas coming from the central peak of the crater ~ Aiphonsus. ~ A. The position of the spectrograph slit. B. Negatives oftwo spectra, of which the right-hand one is 01 normal scattered sunlight, whereas the left-hand one also shows - - - a band of emission ending at about 4,740 A, and corresponding to the point on the slit where it crosses the central peak. 4600 4700 4800 C. The profile of this emission spectrum, which resembles the C © 2 Swan bands. (From Kozyrev, 1961). 302

Kepler, which they ascribed to luminescence, and from the list above, and we shall start by considering Petrova (1967) made photometric measurements of a two surface effects that have been suggested as possible

~ow in the same region. Spectra obtained by Grainger causes of light emission — luminescence and incandes. and Ring using the line-filling method were thought to cence. imply surface luminscence, but have since been as- cribed by them to a surface polarisation effect. Some polarisation anomalieshave been reported by 2. Luminescence Dzhapiashvili and Xanfomaliti (1966). The only Apollo instrumental record of a transient change indicated an Lunar samples have now been shown to luminesce emission of gas, but this was later considered to have in the laboratory under proton or electron excitation, come from the spacecraft. Three Apollo astronauts although UV radiation is ineffective. However, al- while in lunar orbitreported flashes from the surface, though luminescence spectroscopy has been found to but could notbe sure that these were noteye effects due be a useful method of elucidating the solid-state to cosmic ray particles; however, one of the astronauts properties of these materials (e.g. Sippel and Spencer, that saw these flashes (Mattingly) did not, for some 1970; Geake et al., 1976), it is clear that luminescence reason, normally see cosmic ray eye flashes. emission is too faint to make a detectable contribution None of the instrumental evidence is unambiguous to the light from the lunar surface as observed from and undisputed, although Kozyrev’s spectra are prob- the earth. The evidence is summarised in Fig. 3, due to ably the strongest evidence available. For the rest of Nash and Cone! (1971); it is evident that for the known the large number of events reported we must rely on excitation intensity due to protons arriving as solar descriptions of them by many different observers, wind — or even from solar flares — the resulting light mostly amateurs, with widely different telescopes, emission intensity, even if the surface materials were equipment, observing conditions and experience. In of 100% energy conversion efficiency, would still be order to obtain more evidence, networks of observers several orders of magnitude too weak to be detect- have been organised in an attempt to alert well- able from earth against the sunlit lunar surface, by any equipped observatories when an event occurs, so as to of the techniques used. In fact, the highest efficiency obtain photographs and spectra. The Lunar Interna- found for lunar samples (for separated plagioclase) tional Observers Network (LION) was organised by was about 0.1% (Geake et al., 1972). Middlehurst during the , but is now Luminescence emission might just be visible against inoperative, as is Operation Moon Blink (Cameron and the dark side of the moon, but no plausible excitation Gilheany, 1967); W.S. Cameron has also organised a process has been proposed, as the solar particles travel program for the Association of Lunar and Planetary in almost straight lines from the sun; even spiralling Observers, and H. Ford * has organised a network in the of their paths due to the earth’s magnetic field would U.K. A network run by NASA for seven years failed not cause them to fall appreciably beyond the termina- to confirm any of the observations reported to it. tor. Excitation from all other sources, e.g. cosmic rays, However, it is not the purpose of this paper to evalu- is several orders of magnitude less intense than that ate the evidence for lunar transient events, but rather from the sun. to considerhow they might be caused, on the assump- Thermoluminescence,whereby energy is stored in tion that at least some of the events reported really cold material and eventually released as light, has been did occur on the moon. We will now try to assess some suggested as a process relevant to transient light emis- of the proposed mechanisms, on the basis of the cvi- sion from the moon (e.g., Sun and Gonzalez, 1966; dence we now have about conditions on the lunar sur- Sidran, 1968; Blair and Edgington, 1970). However, face, and to decide which are possible and which are this process suffers from the same lack of expected in- not. It is convenient to take them in a different order tensity, and also, the conditions of energy storage and release are hard to envisage. Thermoluminescence emission due to heating at the dawn would * Director of the Bntish AstronomicalAssociation s Lunar Section, and TLP network coordinator. Address: Mills Observa- require energy to be received and stored during the tory, Balgay Park, Dundee, Scotland. lunar night, and it is not clear how it could get there. 303

ENERGY EFFICIENCY —‘-~ ~

I ~ *~ 1 PLAGIOCLASE RICH SEPARATE (FROM 10058) / / / ~ ::: ~ ~ 2 MEDIUM-GRAINED CRYSTALLINE (10058) ~/‘ / / .~-. ~ E ~ FINE-GRAINED CRYSTALLINE (10020) —~/ / / ,_—~2’ ~ 10 4 0.15-1.0mm FINES, CRUSHED /~/~/ ~ /

6 MICROBRECCIA~OO59)~/” •~_2~’ //~ // o~ 7 FINE-GRAINED VESICULAR (10057) /ROCK 1206 s/ / / 5. 6/ / / >~ ~ 10_i 4 APO(LO 12 CORE FINES ,~ ~ // ~ ~

9 ~-2 ////~//AIN~NoEs~ l0-~ 1 INCIDENT10 102PROTONio~ INTENSITY,io~ l0~ erg106 cn1210 sec1I0~ io

SOLAR WIND PROTONS 100 500 1500 SOLAR FLARE PROTONS TEMPERATURE OF ROCK POWDER, ~C Fig. 3. A summary (from Nash and Conel, 1971) of the luminescence emission intensities under proton excitation for materials of different energy conversion efficiencies. Laboratory measurements of Apollo-il and -12 lunar samples are plotted; the diagonal dashed lines bound the region occupied by most terrestrial and meteoritic materials. The horizontal dashed lines show the thresholds for detection from the earth of lumines- cence, against the sunlit lunar surface, by different techniques. It is evident that for solar-wind or flare protons, materials ofeven 100% efficiency would be at least three orders of magnitude below detectability; measured efficiencies for lunar samples were in fact found to be between 0.1 and 0.0001%. A proton intensity sufficient to cause detectable luminescence would also cause de- tectable incandescence.

Thermoluminescence emission on the dark side, where 3. Incandescence it might just be visible if it occurred , has been pos- tulated as being caused by some surface disturbance Herschel (1787)made one of the earliest observa- (e.g. a chance impact) exposing sub-surface material tions of a transient lunar event, actually on the dark that had absorbed high-energy solar or cosmic-ray side of the moon, and he described it as red and spark- particles throughout its history. However, on the dark ling, rather like glowing charcoal thinly veiled with side, sudden exposure of sub-surface material would ashes. Wehave therefore calculated how bright red-hot surely cool it rather than heat it, except for the heat lava would appear to be, if there were indeed some on generated by the impact itself; and in that local area the moon. We have considered a 1-km square of hot incandescence would probably far outshine any in- surface, on the grounds that, as 1 km on the lunar duced thermoluminescence, surface is about the smallest resolvable distance, then a It is therefore verydifficult to regard any form of smaller patch would look the same size but fainter, luminescence as contributing detectably to lunar tran- whereas a larger patch would look the same brightness sient events, on either the light or the dark side of the but larger. The apparent brightness of this 1-km square moon. red-hot patch turns out to depend very critically on 304 the temperature chosen, in this region; e.g., taking “red- the minimum thickness of cloud necessary to obscure hot” as 1 ,000 K and “orange-red-hot” as I ,250 K,there them would be of the order of their vertical height or is a difference of about five stellar magnitudes between depth — probably at least tens of metres for the smallest them. Thus at 1,000 K the apparent surface brightness visible features; i.e., a thin surface layer of cloud or of a 1-km square is about 10,5m, whereas at 1,250 K it mobile dust, however opaque, will not obscure fea- has increased to about 55m For comparison, the sun- tures on a kilometre scale because it will tend to fol- lit lunar surface 2hasatanfullaveragemoon,visualdiminishingbrightnessrapidlyof lowbe attheleastsurfacea fewshape.minutesThefordurationit to be oflikelythe tocloudattractmust toaboutabout4,7m km 2 at the quarter phase, due to the attention, and some such events havebeen reported as peculiarities6-7mofkmthe lunar photometric function. Lava persisting for several hours. at 1,250 K would increase the brightness by about 50% Two possible types of cloud are ballistic clouds and (to 2) at , and by a factor of about gas-borne clouds. 4 (to43m km 2) at the quarters. The effect would therefore52mbekmjust about visible on the bright side of the 4.1. Ballistic clouds moon, especially away from full moon; it would, of course, be easily visible on the dark side. Kozyrev As the lunar “atmosphere” is normally negligibly (1961) indeed ascribed a second event, whose spec- tenuous from the dust-bearing point of view, we will trum he also photographed, tohot lava at about I ,200 first consider clouds that do not require gas. If lunar K. However, there is no geological evidence that any surface dust were ejected, for example by an impact, part of the lunar surface has been red-hot in recent then each grain would follow a ballistic trajectory times, or that hot lava has been ejected recently, so it more or less independently, and if its initial velocity seems most unlikely that incandescence could be the were below the escape velocity it would eventually cause of any of the observed transient events. The only land back on the surface. Fig. 4A shows a man-made exception might be random heating due to large im- lunar transient event which may be of this type; it is pacts, which would be very rare and would not show described as the cloud produced by the impact of the any correlations. Furthermore, any areas attaining Russian spacecraft Lunik V. This would be equivalent these temperatures would be conspicuous objects in to a point explosion, and would involve boththe the IR long after they had ceased to be visible, kinetic energy of the spacecraft, and the release of None of the surface processes considered therefore fuel and other gases. While it is not known how big a seem to be likely explanations of observed light emis- contributionwas made by the temporary presence of sion, and they certainly cannot explain obscurations, gas, the dimensions and duration of the cloud are con- so we will now consider processes involving gas and sistent withthe purely ballistic ejection of dust in all dust clouds near the surface. directions, so this may have been the dominant effect, with the gas having little further influence after the

initial explosion. The first photograph, labelled t = 0, 4. Dustclouds on the moon was said to havebeen taken “just after impact”, al- though the actual time interval was not stated. It ap- It is difficult to explain large-scale obscuration of pears to show a pair of white patches, presumably the lunar surface other than by dust clouds, so we opaque clouds of dust of higher reflectivity than the will now considerhow these might be produced and undisturbed surface; these have disappeared 9.5 mm sustained. As regards the scale required: the smallest later, to reveal surface detail. For some reason, only the lateral distance on the lunar surface resolvable from lower elongated patch is ascribed to the impact: it the earth is about 1 km, so the smallest recognisable seems more likely that both patches have the same features are probably at least 10 km across, and the cause, as surface features near the impact could affect smallest clouds likely to attract attention by obscuring the spatial distribution of the ejected cloud material. such features would probably need to be tens to a Also, the t = 0 photograph appears to be of lower hundred kilometres or so across. As most such fea- general quality than the others; however, there is a tures are normally seen most clearly by their shadows, convincing change in the cloud between the photo- 305

t~O O~5m 45m 95m

80km

IO’/amin 10 9 7’/t

-~ 160km 320km

Fig. 4. A. A man-made transient lunar event, consisting of the dust cloud produced by the impact of the Russian spacecraft Lunik V on May 12, 1965, in the SW corner of Mare Nubium. Photographs by Prof E. Penzel, from the tracking station at Rodewisch, Democratic Republic of Germany. (From Penzel, 1965.) B. Dimensions of trajectories for dust particles projected in all directions from the same point on the surface, with the same initial 1. The extreme trajectories only are shown — the highest (for vertical projection) and the widest (for projec- velocity of 500 m s tion at 45°);the dashed curve is the envelope of all the intermediate trajectories. Some times of flight to points on the surface are given; for the intermediate trajectories there are two times to each point, and only the longer is given. Surface curvature has been neglected, and does not make a significant difference. The dimensions and duration of the resulting cloud of dust grains in flight would be roughly consistent with the cloud shown in (A). graphs at 0.5 and 4.5 min, which are of comparable For comparison, Fig. 4B shows the distances and quality. The maximum lateral dimension of the cloud flight times for dust particles ejected in all directions is given as about 225 km, with a height of 80—95 km. withthe same initial velocity (actually 500 m s’). The 306 observed cloud is roughly consistent with this, in that 4.2. Charge-levitated dust clouds its height, width and duration are related in about the same way. Another suggested dust-raising process is levitation A puzzling feature of the photographs is that the due to surface grains acquiring charges of the same cloud is shown at its largest at t = 0, whereas, however sign, for example by the impact of charged solar wind the cloud was formed, it would surely increase from particles such as protons, or by the photo-electric ef- nothing to a maximum size, and then decay as the dust fect due to solar radiation. Gold has demonstratedgrain- settled. It therefore seems more likely that the first hopping caused by protons in the laboratory, and we photograph was actually taken a few minutes after im- have found the same effect in our proton excitation pact, when the cloud had become conspicuous. The system. However, in neither case were the conditions explanation offered in the report quoted is that the those of the lunar ultra-high vacuum; Salisbury and cloud was of dust thrown up by the retro-rockets as Glaser (1964)have shown that when the ambient pres- the spacecraft approached the surface, presumably at sure is low enough for the dust grains tobe thoroughly a low angle. This could indeed explain an elongated outgassed,then even grains carrying the same sign of cloud of maximum length at the moment of impact, charge stifi adhere to each other by Van der Waals’ but it ignores the second part of the cloud, shown in forces, which are much stronger than those due to the the photographs; also, it seems unlikely that the retro- charges they acquired in their apparatus, It is therefore rockets could throw the dust anything like far enough difficult to see how the repulsive forces due to charge sideways to explain the width of the cloud. Judging effects could detach lunar surface grains, but if they by the Apollo landings, dust is thrown hundreds of were detached, by some other means, it is possible metres (e.g., at Surveyor II by Apollo 12), whereas in that charge effects might help to keep them levitated. this case it would need to be thrown tens of kilo- Two possible situations are: metres. (1) If onlypart of the surface layer of charged grains We have discussed these photographs is some detail is detached, by some means, then the detached grains because they appear to be the only ones of a transient see the remainingsurface as an extended sheet of event on the moon whose cause is known for certain; charge, and experience a uniform force upwards, in ad- we conclude that the most likely explanation of the dition to random inverse-square repulsions from resulting cloud is that it is purely ballistic, following neighbouring detached grains. the ejection of dust by the impact explosion. If this is (2) If all the charged layer is stripped off, then the so, it follows that dust clouds of sufficient size and detached grains are only repelled from each other; the duration to cause noticeable obscuration can occur nett effectis toforce the upper grains upwards and the without requiring gas emission from the surface to lower grains downwards, and the situation reverts to keep them up: they could be caused by meteorites of (1). comparable kinetic energy to Lunik V. However, these The thickness of the resulting charge-levitated dust would be extremely rare, and would not correlate with cloud probably depends mainly on instabilities due to anything, in position or time. They would not produce non-uniform charging of the surface, but it seems un- light emission, except perhaps for an initial flash, al- likelythat such a cloud could be either thick enough though it is evident from Fig. 4A that the cloud looks or dense enough to cause obscuration; in any case, temporarily brighter than the undisturbed surface, by charge effects can only assist some other means of de- scattered sunlight. Furthermore, an obscuration lasting taching surface grains, such as gas release or thermal more than a few minutes could only be caused by such shock. Furthermore, no such moving dust has been a large meteorite that it would probably leave a new seen by astronauts on the lunar surface. Evidence, visible crater, and none have yet been seen. We there- from flow patterns, for there having been some dust fore conclude that, although meteorite impacts are a movement probably refers to such movements earlier possible cause of observable transient events, very few, in the moon’s history, when conditions, especially as if any, of those observed have in fact been caused that regards gas emission, were very different. The only way. It is nevertheless possible that some clouds caused evidence for any such dust movement at the present by gas release are in effect baffistic, with the dust grains time is that Apollo detectors on the in free flight after initial projection by escaping gas. lunar surface have occasionally detected grains moving 307 horizontally at around dawn; these may possibly have detached, then the gas molecules wifi be unimpeded been detached by thermal shock, and then levitated once they have emerged; effectively, they will stream by charge. The only chance of seeing this effect from out radially in all directions with only minor encounters the earth might be if enough dust were thus detached with each other, and having a range of velocities around to produce a temporary change in the albedo; indeed, the RMS value, which is proportional to (TIM)’12 some events involving transient brightening have been (where T is the absolute temperature and M is the mo- reported around the dawn terminator, although they lecular weight), but is independent of pressure. The have usually been ascribed to thermoluminescence, effective temperature is uncertain. The lunar daytime The rest of this paper will now be concerned with temperature is over 100°Con the surface, but it drops processes which depend on gas being released from to an almost constant value of about 0°Ca metre or so the lunar surface. down; also, gas expanding into a near vacuum is strong- ly cooled. However, a temperature range of 0 ±100°C 4.3. Gas-borne dust clouds only makes a difference of about ±20%to the RMS velocity, so it will not significantly affect our argu- The most obvious way in which lunar surface dust ment if we use the value for 0°Cthroughout. might be raised to form a cloud is by means of gas We will start by considering 112, as this is likely to escaping from the surface. There is no doubt that be by far the most abundant gas; it is also the lightest, gases are present inside the moon: these will include and therefore has the highest RMS velocity (1.7 km s1 gaseous constituents of high cosmic abundance, mainly at 0°C).This happens to be almost exactly the ye- H 2 and He, and small amounts of compounds such as locity of a body in a circular orbit round the moon, H2S, H2O, CO, C02, ~°2 and CH4, plus their dissocia- near the surface (1.68 km s’), The lunar escape veloc- tion products, including possibly C2 as found by ity is ~ times this (2.38 km s’), and from the Kozyrev (1961); Kozyrev (1963) also identified 112 Maxwell distribution about 15% of the molecules will from a further spectrum. The solar wind consists most- have escape velocity or higher, and will be lost imme- ly of H~and He~ions, with some heavier ions such diately. Molecules emerging with the RMS velocity will as Ar~and Net. The lunar surface provides electrons follow ballistic trajectories, and may in factland any- to neutralise these ions, which are then adsorbed as where on the entire surface of the moon. For example, gas molecules by surface grains. These grains may one leaving vertically will ascend to a height of 1 lunar eventually become buried, and may subsequently re- radius (1 ,738 km) and land back in the same place 154 lease some of their adsorbed gases. Finally, the decay mm later; one leaving horizontally would, theoretically, of the initial40Ar.radioactiveGases fromconstituentsall these sourceswill providecould inHe, justand wouldskim thearrivesurfacebackofattheitsmoon,startingifpointit were108smooth,mm principleRn and be trapped in cavities in the regolith, and later. Those leaving at intermediate angles would fol- might then be released from time to time due to tidal low elliptical orbits of the same total energy,and or thermal stresses, which might trigger the fracture of therefore with the same major axis (equal to the diam- a storage cavity in which a high pressure had built up. eter of the circular orbit, i.e. the lunar diameter), The catchment volume might be verylarge, especially and with a common focus at the lunar centre. These at the edges of maria if deep cracking and internal dis- orbits all cross the lunar surface somewhere, and the

ruption occurs there (e.g., Runcorn, 1976) — and these impact point travels right round the moon as the mi- are just the places where many transient events have tial direction goes from horizontal to vertical. Mole- been seen. Whether such gases, leaking from the sur- cules of higher velocity, but below escape velocity, can face, would disturb the dust would depend on the pres- similarly land anywhere on the lunar surface, but they sure and on the mass flow rate, and therefore on the go further out and take longer; those leaving at less nature of the gas, the storage capacity of the gas-traps, than 1.68 kin ~ cannot get so far round the moon, and the pressure they could withstand before bursting. but can still cover a wide area. It follows that gas mol- What happens to gas released from a trap in the ecules emitted unimpeded from a single leak can be lunar surface wifi depend very much on the details of distributed over the entire lunar surface after an hour its encounters with surface dust grains. Ifno dust is or so, assuming that they do not hit anything on the 308 way. As the “normal” pressure of the lunar atmo- depend entirely on chance local details of dust-grain sphere that they will encounter is < 10_12 torr, at configuration, and on whether the gas is trapped at a which the mean free path is about i0~km (ten times high enough pressure to be able to detach dust grains the lunar circumference), molecules are in fact un- to form a cloud near the emission hole. If any dust is likely to collide. However, although the gas is soon raised, some molecules will have been deflected from distributed very widely, for this very reason the pres. their radial streaming in the process, and following gas sure will be extremelylow, and the pressure near the molecules will encounter the dust cloud and some of emission point will in fact drop by a factor of about them may be scattered. The degree of randomness of 1016 in the first second. There seems to be no pros- motion thus acquired will determine how longthe gas pect of such gas showing itself as a transient event — molecules stay around; however, it seems probable except that it might be just detectable by instrument~ that for gas escaping from a single high-pressure leak, elsewhere on the surface, such as the Apollo cold- the situation will usually be nearer to the streaming cathode gauge. situation than to the random one, and that the local Although 112, the lightest, fastest and most abun- pressure will therefore drop very rapidly. It may be dant gas, has been takenas an example, the same pro- that dust grains are in effect projected by gas, but are cess will occur with other gases. The only difference soon on their own in ballistic trajectories, with the for heavier gases is that a smaller proportion of the ambient gas already at too low a pressure to have any molecules will have escape velocity initially, so more further influence on them. will be retained; also, a smaller proportion will emerge We must now consider in more detail just how dust at more than 1.68 km s~,which is the condition for grains may be detached by a flow of gas, so as to form being able to spread ballistically to all parts of the a cloud. For the extreme situation, of an area contain- moon. Both these effects result in a higher proportion ing a multitude of small leaks, we can start by looking landing near the emission point. When gas molecules at the behaviour of an ordinary fluidised bed at at- do hit the surface, they will either bounce off, or be mospheric pressure (see, e.g., Rowe, 1965). In that adsorbed and perhaps released later; in either case they case, gas entering the porous base of a deep dust layer thereafter contribute to the equilibrium of gases near just levitates the grains when the pressure drop through the surface. the layer equals the weight of the layer, per unit area. The other extreme situation is the one in which gas The just.levitated layerhas only expanded in depth by molecules completely loose their preferred radial direc- a few percent, and behaves like a liquid when disturbed. tions of motion by many encounters with dust grains Increasing the gas flow causes only slight further ex- soon after release; they wifi then soon acquire com- pansion, but introduces instabilities in the form of pletely random directions of motion, while retaining gas bubbles rising through the dust; a further flow in- their Maxwell distribution of velocities. They then con- crease causes violent bursting of these bubbles, eject- stitute a local temporary atmosphere, and the rate ing dust grains from the surface to form a cloud which both of spreading and of loss will decrease rapidly with remains suspended by the gas flow. increasing molecular‘weight. The gas cloud will spread If the pressure above the bed is now reduced below by thermal diffusion, and molecules may be lost when atmospheric, this begins to have a serious effect when they happen to acquire escape velocity outwards by the mean free path of the gas molecules becomes com- collision. The half-life of such an atmosphere ranges parable with the grain size, which happens, for exam- from a few minutes for H 9 years for Ar; pie, at a pressure of a few torr for 1O-btm grains. Ex- however, for these heavier2 moleculesto over i0the dominant periments by Miller and King (1966) have shown that loss process is eventual ionisation by solar UV, follow- the effect is that while the deeper layers of the bed are ing which it only takes a few minutes for them to be unaffected, the top few centimetres become totally accelerated away by the magnetic field associated unstable and blow off to form a suspended cloud. The with the solar wind, properties of gas-fluidised ash flows, under lunar con- What actually happens to the released gas, between ditions, have been analysed theoretically by Pai et a!. these verywide extremes of ballistic streaming on the (1972); they conclude that this may have been a pos- one hand, and thermal diffusion on the other, must sible way of transporting material for crater filling, 309 and that it offers an explanation of features such as of just obscuring the surface. Furthermore, the fluid- sinuous rules and “ marks”. However, the condi- ised bed itself would not cause obscuration as it would tions they consider involve high temperatures and follow the surface, and its attendent surface cloud copious supplies of gas, which may have been so when would probably notbe deep enough either; probably the major features of the moon were being formed a only the initial ballistic cloud would be effective, billion or more years ago. It is important to realise that Nevertheless, even the surface cloud would probably this is not relevant to present-day transient lunar events, cause a temporary change of albedo, and might be as conditions are now very different. seen as a transient event involving a brightness change In a very high vacuum (.~~10_l2torr) such as now without obscuration; but for any brightness change normally prevails at the lunar surface, the dust grains ascribed to a high-albedo dust cloud (e.g. Fig. 4A) to will be almost completely out-gassed, and this makes leave no permanent change (and none has ever been a major difference. Salisbury et al. (1964) have shown seen), the dust must settle back afterwards to re-form that below a pressure of about i0~9torr, dust grains, the initial surface complexity characteristic of dust in stripped of any adsorbed gas layer, strongly adhere to a very high vacuum. This will be further discussed later. each other by inter-molecular (Van der Waals’) forces; As the flow of gas through dust, under lunar condi- their labo~atoryexperiments showed that the adhesive tions, appears to be the key process in tryingto under- force was between 12 and 200 times the particle weight stand transient lunar events, we will now consider this (depending on particle size), suggesting that under in more detail. Three different flow regimes may be lunar surface gravitational conditions grains would involved in different regions: stick together with a force of several hundred times (1) The stable fluidised bed, described by Ergun’s their weight. Furthermore, when in this state, mere equation (see, e.g., Davidson and Harrison, 1963); in exposure to gas, even at atmospheric pressure, does not this situation levitation,with slight expansion, just penetrate and remove these bonds. It therefore seems occurs when the pressure drop through the layer, h.P, likely that under lunar vacuum conditions the gas flow equals the weight per unit area. Thus, z~.P= gpt 0.4 required to levitate such a layer of out-gassed grains, torr per centimetre thickness t (for dust of density by tearing them apart to disrupt these forces, would p = 3,000 kg m3, and with g for the moon = 1.62 m be so large that, once separated, the dust layer would be ~_2). However, this ignores the molecular adhesion of far beyond the stable fluidised bed state, and the grains the initially out-gassed grains, which may be several would be forcibly ejected. In other words, it would hundred times their weight, suggesting that a pressure seem that a stable fluidised bed, producing flowing of hundreds of torr per centimetre depth would be dust, may not now be possible on the moon, and if the needed to start the process. gas flow can detach the grains at all they will burst out (2) For grains blown up to form a cloud, Stokes’ to form a ballistic cloud. Assuming that it is ejected law will hold, as long as the mean free path for the at less than escape velocity, each grain will fall back gas molecules is much less than the grain size — say under gravity and either hit the surface or be levitated 1/10; e.g. for H 2 at a pressure of 1 torr this would ap- by continuing gas emission. When the gas flow ceases, ply to grains of more than about 10 ~zmdiameter. in the grains wifi fall under gravity and would re-out-gas the Stokes region, the drag force tending to carry the within a few minutes of the pressure falling below grains upwards with the gas stream is proportional to iO torr, or so. ~aV, where i~is the gas viscosity (which is independent A stable fluidised bed would therefore only be pos- of pressure), a is the particle diameter, and Vis the sible if an initial disruptive burst of gas, sufficient to gas velocity. detach the grains, were followed by a long period of (3) For smaller grains, or lower pressures, such that reduced flow just sufficient to support the grains of the the mean free path is>> a/l0, streaming flow occurs, bed and of the surface cloud that would form over it and the drag force on a grain is now given by the rate with an external near-vacuum. However, the amount of loss of momentum by the impinging gas molecules of gas required to maintain this fluidised bed would that actually2, wherehittheit; gasthisdensityforce isPGnowdoesproportionaldepend ontothe muchprobablylessfargasexceedmerelypresent-dayto sustainasupplies.dust cloudIt requirescapable pGaVpressure. The result is that grains thrown up by the gas 310

will tend to be carried withit, until the pressure falls duced by astronauts and their vehicles suggests that in to a critical value which depends on the grain size: for fact the dust usually comes off as a fine spray. It there- example, by Stokes’ law, the rate of fall of a grain in fore seems that the formation of a dust cloud at all H2, and therefore its rate of slip with respect to a ver- requires pressure rather than quantity of gas, but that tical gas stream, would be 1 m in about 5 mm for a for the resulting dust cloud to last longer than the few 10-pm grain, 9 h for a 1-pm grain, or 1 for a minutes required for the dust to settle ballistically re- 0.1-pm grain. This is independent of pressure, but as quires either a continuing supply of gas or a sequence the pressure is reduced, it only holds until the mean of separate puffs, perhaps from different traps in the free path has increased to about 1/10 of the grain size; same area, triggered by the same overall stress. This for H2 this occurs at a pressure of about 100 torr for pop-gun effect, whereby grain adhesion allows poten- 10-pm grains, 1,000 torr for 1-pm grains, or 10,000 tial energy to build up as gas pressure, may be an essen- torr for 0.1-pm grains. Below these pressures, the drag tial mechanism for transient events to be visible. force is much smaller, and also decreases linearly with pressure; by these pressures, the fall rate is already 500—50,000 times its Stokes’ flow value, for 10—0.1- 5. Obscuration by dust clouds pm grains, and thereafter it increases as the square root of the pressure decrease. As the pressure near the The formation process for dust clouds on the moon surface is likely to decrease rapidly withheight, and involves too many uncertainties for it to be practi- with time after the gas emission ceases, it seems prob- cable tocalculate the energy required, on the basis of able that, as suggested earlier, the driving force on a the mechanics of formation. We can, however, do some grain becomes negligible rather early in its trajectory very simple calculations, based on the energy required and that it is thereafter on its own in purely ballistic to lift enough dust to produce observable obscuration, flight, apart from rare collisions with gas molecules and which should at least indicate the order of magnitude. with other grains, and that gas levitation plays no fur- The simplest starting point is to say that a cloud of ther part. At high pressures, small grains are carried dust will cause complete obscuration if there is instan- withthe gas more effectively than large ones, but for taneously just one grain everywhere, in line of sight, small grains the pressure at which Stokes’ law breaks at some height. The contrast at which surface details down, causing the grains tobe dragged along much are seen will fall off continuously as the amount of less effectively, is reached sooner, so small grains may dust increases, and wifi certainly have reached zero by actually fare worse when the pressure is falling rapidly. this stage. In practice, the situation is much more com- We saw earlier that a considerable force is probably plicated than this very simple approach: the resulting required to detach the initially out-gassed surface contrast reduction will depend on the grain size in rela- grains, which are adhering to each other by Van der tion to the wavelength, and the contrast reduction Waals’ forces, and that the mere presence of gas does that amounts to obscuration will depend on the light- not penetrate these bonds;a gentle seepage of gas ing conditions and on the actual contrast of the unob- from the lunar surface, which may well occur, would scured surface details. The scattering of light by dust not therefore produce a dust cloud at all. For this to will be considered in more detail in Section 7.2, but happen it is probably necessary for the gas to be we will assume for now that about 1/10 of the above trapped securely enough for it to build up the pressure amount of dust would cause significant obscuration required to disrupt the adhesive dust layer. When the of lunar surface detail, as seen from the earth. This pressure of the gas has built up to near the limit that will probably notbe out by more than an order of mag- the surface can hold, its release may be triggered by nitude, which is near enough to be useful. 2 grainssome tidalwouldorbethermalprojectedstress,as anddiscussedthe detachedabove. Itdustis also of 10-pmOne graingrains,everywhereof densityrequires3,000 kgaboutm32,000(or 200kg kgkm~ possible that clods of dust might be thrown off, which km~2of 1-pm grains, or 20 kg lrm2 of 0.1-pm grains), would not get very far; these would merely reduce the SO we are assuming that 1/10 of this is required, i.e. effectiveness of the process, as they would not produce 2—200 kg km2, depending on the grain size. As the a dust cloud, but the appearance of disturbances pro- smaller grains are more effective scatterers of light, we 311

2would be enough about 16 times as much energy as the multiple-puff will assume that about 10 kg km~ to cause reportable obscuration. If we take the smallest ballistic cloud covering the same total area for the cloud likely tobe noticed as being about 10 X 10 km, same time. It is worse for larger clouds, as the energy then the minimumamount of material required is of per unit area is proportional to the diameter, D, where- the order of i03 kg. The energy required to lift this as the time taken is proportional to\/D. to a height of 10 m (about the minimum that would obscure anything) is about i04 J. For comparison, the energy associated with a large lunar seismic event 6. Light emission from gas-borne dust clouds is in the range 102_lOs J (G.V. Latham et al., 1972); it follows from this that one would expect It is well known that dust grains tend to charge up some correlation between observed obscurations and electrostatically on collision with other grains or with seismic events. Even if the actual seismic event were surfaces: large potential differences have beenmea- only acting as the trigger, the gas release itself ought sured in desert sandstorms, and lightning discharges to produce a detectable seismic signal. There have have frequently been seen in clouds of dust and ash been some tentative reports of such correlation, but ejected from volcanoes, as ifiustrated in Figs. 5 and 6. as in all aspects of this investigation, it would be of We have already discussed the probability that ob- great interest tohave more evidence, served obscurations on the lunar surface are caused by The obscuration, thus started, would probably gas-borne dust clouds; we will now consider whether need to last for a fewminutes to be noticed; this electrical discharge in these clouds is a possible cause could happen either if a continuing gas stream levitated of the light emission sometimes observed in transient the grains, or if they simply fell under gravity and were lunar events. replaced by dust raised by a further puffof gas. Tak- Light emission from gas, without dust, has also been ingthe latter case as the simplest, dust projected to a suggested, and the spectrum obtained by Kozyrev (Fig. height of 10 m would take about 7 s to rise and fall, 2) was ascribed by him to the solar-excited emission so a continuing obscuration would require about eight of the C 2 Swan bands in gas released from the central such puffs per minute, or a total energy of about iO~‘~. peak of the crater Alphonsus;he also reported that A 5-mm obscuration, the shortest likely to5beJ.no-This thanduringusual,this timeand thatthe areathis extraappearedbrightnessbrightersuddenlyand whiter assumesticed,wouldthe worstthereforesituation,involveofaboutthe gas5 that- i0 raised the ended, whereupon another spectral exposure was start- dust not staying long enough to support it; if the gas ed which did not show the emission bands. While it molecules acquired sufficiently random directions of seems convincing that this event was real, and that it motion by colliding with dust grains to form a diffus- was bright enough to be visible, it is difficult to under- ing cloud, as discussed earlier, then the amount of stand how solar-excited gas fluorescence could produce energy required to maintain the dust cloud might be the intensity required to make a significant difference at least an order of magnitude less. against the brightness of the sunlit lunar surface: the These calculations have been for the smallest cloud efficiency of the process is low, and the sun’s output likely to be noticed. A larger cloud may be produced is relatively low at wavelengths short enough to be either by a large number of gas sources distributed over absorbed by the gas. It may be that the visible bright- the area, in which case the energy required simply goes ening was actually due to sunlight being scattered from up pro rata;or the same visible effect could be produced dust in the cloud. The presence of charged dust par- by a single point emission of high-pressure gas forming in the gas cloud might also contribute to the exci- a single ballistic cloud of the required width, However, tation of the gas emission shown in the spectrum; in- as a ballistic cloud necessarily has a height 1/4 of its deed, Kozyrev himself originally attributed a fall-off width more energy is needed, as the dust is raised to a in the violet part of the spectrum to the presence of greaterheight: e.g., a 10-km cloud involves giving ‘the gas-borne dust, but he laterwithdrew this explana- dust grains enough energy to raise them to a height of tion (see footnote in Kozyrev, 1961, p. 367). It has

2.Skm — but they take nearly 2 mlii to go up and been demonstrated in the laboratory (e.g. by Mifis. come down again, so the nett result is that it requires 1970; and Fig. 7) that charged dust in h 2 at low pres- 312

-Th~

~1~

Fig. 5. A long-exposure photograph showing a large number of lightning discharges in the cloud of dust and ash over the Vesuvius, during eruption. (From Colemen, 1949.) 313

_ Area covered •in frame of time- lapse ‘Th ‘~Z.I _

~ picture

:1234~ ____

SCALE AT SURTSEV I

Fig. 6.A. A drawing, from photographs, of the cloud plume over the volcanic island Surtsey, during eruption. B. An enlarged drawing of part of (A), showing the positions ofa large number of lightning discharges. (From Anderson et al., 1965.) sure (about 0.5 torr) can produce a glow discharge the spectrum because it is concentrated into a narrow showing the characteristic H emission spectrum; how- spectral band. It seems more likely that the visible ef- ever, Mills finds that for this glow tobe’ visible to the fect reported by Kozyrev was produced by the scatter- eye, a darkened room is needed. It seems unlikely that ing of sunlight by the same dust. such a glow discharge would be visible by eye against All the light-emission processes so far considered the sunlit lunar surface, although it might show up in have been of inherently low brightness;however, an-

Fig. 7. Apparatus to demonstrate the glow discharge excited by moving dust. Dry sand in a glass globe containing H 2 at a pressure of 0.1—0.5 torr is disturbed as the globe is rotated, and a glow discharge is seen. 314 other process relevant to gas-borne dust clouds is the that lightning discharges are commonly associated with possibility of lightning-type discharges. These would the cloud-plumes of ash ejected by volcanoes; Figs. 5 occur between clouds with a nett excess of oppositely and 6 show examples, and the process has been dis- charged grains, somehow separated, rather than be- cussed, for example, by Hatakeyama and Uchikawa tween pairs of single grains. The important difference (1951) and by Anderson et al. (1965). is that the light emission intensity of a flash is now on- In the lunar surface layer there are grains both of ly limited by the amount of energy stored before dis- different materials and of all sizes down to the sub- charge occurs, sowe can now get away from the low micron range. There is no obvious reason why the efficiency inherent in continuous processes such as charging process proposed should be affected by a low luminescence and glow discharge. Also, the emission ambient pressure, although little work has been done of the light as a number of discrete but unresolved on this. It therefore seems reasonable that agitation flashes is consistent with the description of the light of the dust, during cloud formation by a release of gas, from some transient lunar events as “sparkling” (e.g., should result in charging; certainly the dust thrown up Herschel, 1787;Greenacre and Barr, 1963); the red by the astronauts and their vehicles behaves as if colour these observers all reported is consistent with charged. It is also true that some of the grains will al- the discharges having occurred in H2, which is likely ready have been charged photoelectrically while on the to be the most abundant gas. Middlehurst (1968) also surface, but as this can only affect the thin layer of records that both Halley and de Louville saw what grains actually exposed on the surface it is unlikely to they described as “lightning” on the face of the moon, make a significant contribution to the charge in the during , cloud. For this explanation to be plausible, there must be a charging process, a charged-cloud separation process, 6.2. The charged-cloud separationprocess an energy storage process, and the necessary conditions for electrical breakdown of the gas. For a lightning-type discharge to take place, it is necessary to generate a large potential difference. It is

6.1. The charging process not enough to have individual charged grains — it is necessary to have a large accumulation of charged Frictional electrostatic charging of dissimilar dielec- grains, in two separated clouds or zones, one with a tric solids is well known. However, measurements in large excess of positive charge and the other of nega- sandstorms, and laboratory experiments with sand tive charge. Ifthe dust grains are being borne aloft by grains, have shown that when dielectric grains even of a gas flow, then their terminal velocity with respect to the same material, but of different sizes, collide, then the gas flow depends on their size, whatever the flow they tend to acquire opposite charges, the larger ones regime, i.e. the smaller grains are carried up faster, and becoming positively charged. This seems to be because therefore further, than the larger ones; as they also ion mobility increases with temperature, so any tem- tend to be oppositely charged, it is possible for the re- perature gradient tends to produce a charge gradient. quired separation of cloud zones of opposite nett In a collision between two grains of different sizes, charge tobe produced. This explanation is consistent the heat generated causes a bigger temperature rise for with the observation that the negative charge is usually the smaller grain, because of its smaller thermal capac- uppermost in volcanic clouds. ity; the resulting temperaturedifference between the grains therefore causes a charge transfer during their 6.3. The energy storage process brief contact. This chargingmechanism is discussed by J. Latham (1964) in relation to sandstorms, with many The energy of charge separation will accumulate, references to observations and to laboratory experi- providingthe rate of charge separation exceeds that of ments. Fig. 7 shows apparatus inwhich agitated sand recombination and other losses. It will do so until the produces a glow discharge in low-pressure gas; radio potential difference reaches the minimum for a dis- interference from sandstorms also suggests that elec- charge to occur, whereupon all the stored energy is trical discharges are taking place. There is no doubt released. The energy is stored initially as potential ener- 315

gy — as the pressure that builds up in the gas before its about 300 V; both of these values depend somewhat release from the lunar surface; the amount of energy on the gas, as indicated. If PS is further reduced, the

that can be stored in this way — and ultimately the in- required potential difference increases sharply, and a

tensity of the light flash — is presumably increased by discharge soon becomes virtually impossible. The cx- the strong forces of molecular adhesion between the planation for this is that the electrode separation has initially out-gassed grains, which would greatly increase become less than the mean free path of electrons in the pressure required to break through. Observers have the gas, and as electrons will now on average cross the reported areas of sparkling light, lasting a few minutes; gap without collision,the avalanche required for a dis- this may be because many such pockets of gas are re- charge cannot build up. leased at different places and times, but triggered by In the lunar case, the supposition is that the parts the same overall stress and giving one flash each during of the cloud containing larger and smaller grains re- the event. spectively are increasing their nett charges of opposite sign, and also getting further apart vertically, at the 6.4. Conditionsfor discharge same time as the ambient pressure is dropping. If, as seems likely, the pressure is falling faster than the The necessary conditions for discharge are given by separation is increasing, then PS decreases with time Paschen’s law, as illustrated in Fig. 8. The graph relates (and the potential difference required for breakdown

the minimum potential difference required to cause a therefore also decreases) — at the same time as the ac- breakdown, to the product of the ambient gas pres- tual potential difference is building up, so the condi- sure P and the electrode separation S. As PS is reduced, tions for a discharge may be attained. As the optimum there comes a value, of about 1 torr cm, at which the value of PS is about 1 torr cm,if the pressure is to be required potential difference reaches a minimum, of sufficient to levitate the grains it seems probable that this discharge process would take place rather near to

— —• the surface — say within a few metres.

‘j’- 30kv The emission spectrum of a lightning discharge is partly that of the ambient gas and partly that of vapour- Breakdown ised “electrode” material, with the gas dominant above Voltage the Paschen minimum and the electrode materialbe- low it. The red colour associated with the “sparkling” transient events is consistent with 112 or He, which are in any case the gases most likely to be present at suffi- cient pressure. J.A. O’Keefe (1970, and personal corn- munication, 1970) has pointed out that all the visible

He 250 : emission lines ofboth atomic and molecular hydrogenin- N 2 270 3~v - volve levels that are at least 10 V above the ground state. This virtually excludes solar excitation, at least as far 260 ______as the most abundant gas is concerned, as the solar Press. x Sepn. 760 output at or above this energy is much too small to

~:~° torr~cm produce the necessary intensity; on the other hand, N2 06 electrostatic discharges evidently involve at least 300 V. 0204 We therefore conclude that the occurrence of light-

XeO-5 fling discharges in gas-borne dust clouds is a possible Fig. 8. A specimen curve (not to scale) illustrating Paschen’s mechanism of light emission, consistent with the re- law — the relationship between the minimum breakdown volt- ported appearerice of some transient lunar events. As age required for discharge, and the product of ambient gas the intensity produced depends on a storage process, pressure and electrode separation. The numerical values are , . . different for different gases, and some examples of those for the mechanism cannotbe dismissed on efficiency the mininium point are given. For more complete data, see grounds as producing light too faint to be seen from Gänger (1953). ‘ the earth, as can all the other luminescence and gas- 316

emission processes. The primary energy storage process viewing angle, but could be as much as 100%. No appears to be that of attaining the gas pressure required colour changes were observed, and the settled powder toburst the surface, and this is followed by its conver- retained its increased reflectivity when the vibrator sion into electrical energy by charge separation. Unfor- was turned off. Unfortunately, therefore, it is not tunately, this proposed mechanism does not lend it- clear how a transient brightening on the moon could self to laboratory demonstration, because the absolute be achieved: any area thus brightened, possibly by

scale is crucial — just as one cannot simulate an ordi- seismic agitation, might be expected to remain bright nary thunderstorm in the laboratory. One can only do afterwards, unless the effect of the ultra-high vacuum experiments on details of the process, and observe it environment was to re-form the “fairy castle” struc- on a large scale in volcanic clouds, albeit at a very dif- ture very quickly. ferent ambient pressure. The effect of a percolating flow of gas, rather than mechanical agitation, on the reflectivity of a finely-di- vided dark-coloured powder has been investigated by 7. Modulation of sunlight Purser (1973). His general findings are summarised in Fig. 9, which shows four stages of the process: For transient events involving brightness changes, (1) The initial state, with a static bed of dust, and occurring on the bright side of the moon, we can roughened and ridged with a spatula as in Garlicket consider ways in which reflectedand scattered sun- al.’s experiments. light might be modified in intensity. This gets away (2) Fluidisation by a gentle flow of gas through the from the low efficiencies inherent in surface emission bed, whereupon the reflectivity rises as the surface be- processes such as luminescence, although it is of no comes smoother. help in explaining dark-side events. It is convenient to (3) More vigorous disturbance by a stronger flow subdivide the possible mechanisms into two types, INCIDENT which may in practice occur either separately or to- LIGHT gether: (1) those involving agitation of fine surface dust; and (2)~thoseoccurring in dust clouds.

—~--___ 15° 7.1. Agitation offine surface dust 12 I ~ •44, I /~35~G~( Garlick et al. (1972) were the first to point out that \\ o~. all that was required in order to generate a compara- tively bright patch on the moon was to increase the 8 \ ~. / albedo of that locality, perhaps by physical distur- z bance. The normal lunar regolith owes its remarkably 6 low albedo partly to the “fairy castle” structure as- 65° sumed by fine dust settling under ultra-high vacuum

conditions, and if this could be broken down, and the - ) 75° dust smoothed out, its reflectivity would tend to in- 2 2

crease. - ,/ 1 85° Garlick et al. tested their hypothesis by first mea- - I I I suring the reflectance of Apollo-14 lunar dust in a o 2 4 6 8 io 12

static condition and “roughened” with a spatula — al- REFLECTION 0/~ though the experiment was conducted at atmospheric Fig. 9. Variation of reflectance with angle for a dark powder pressure, so true “fairy castles” could not be built up. (densified lampblack), static and agitated: (1) Initial condition: roughened and static. The powder was then vibrated mechanically, where- (2) Fluidised by a gentle flow of gas. upon it became “fluidised” and the smoother surface (3) Vigorously agitated by a stronger flow of gas, and covered showed the expected rise in reflectance. The magni- by a thick cloud of the disperse phase. tude of the increase was strongly dependent on the (4) Final condition: smooth and static. (After Purser, 1973.) 317 of gas, sufficient to produce a “boiling” bed covered mathematical analysis of the general case of scattering by a dark cloud of the disperse phase; this results in a by irregular and mhomogeneous particles has not yet nett diminution of the scattered light at all angles of been achieved; it has been carried out for homogeneous view, systems of isotropic dielectric spheres, and even in this (4) The final state, after stopping the gas flow had restricted case the behaviour is complicated (Mie, 1908; allowed the bed to settle down into a smooth, static van de Hulst, 1957). Sinclair and La Mer (1949)have layer. This shows the highest reflectivity of all, of up used Mie theory tocalculate the total scatteringcoeffi- to 80% above that of the initial roughened condition. cient for particles of various refractive indices,and of radii These experimental results are in general agreement up to 1 pm. Some of their results are reproduced inFig. 10, with those of Garlick et al., in showing that gentle agi- where the ordinate is the scatteringcoefficient K, defined tation produces an increase in reflectivity; however, as the ratio of the effective tothe actual cross-sectional they also show that more violent agitation actually re- area of the particle; the abscissa is fundamentally in ducesthe reflectivity again. They confirm that the units of a, the circumference of the particle in wave- final result is a permanent increase of reflectivity, thus lengths, i.e. a = 2irr/X (where r is the particle radius also failing to reproduce a transient event. However, and X is the wavelength). Scales are also given of r for this does not necessarily eliminate this process as a pos X = 0.524 pm, and of X for r = 0.3 pm. These curves sible explanation of some lunar TLE’s: it is possible - are evidently complex, withseveral peaks, and K may that if these experiments were repeated under ultra- high vacuum conditions, then the complex grain struc- 6 tures required to give a low surface albedo might be re- formed after the disturbance. ~ ‘I

7.2. Scattering of light by particulate clouds — ~ II Ill \

ii UI V jt.’1’65 particlesThe wayis aincomplexwhich lightphenomenonis scatteredproducingby a cloudtwoofob- 4 — A’ ~ 1:I .\ servable effects — colouration and polarisation. The re- ‘~ V - sult depends strongly on the particle size. K

For particles of molecular size the scattered light is — I characterised by strong colouration. Rayleigh (1899) I / ~ ~ showed that, for particles that were considerably / smaller than the wavelength of the incident light (say / \ ,

proportional to l/Xx, where x decreases progressively - - • - Fig. 10. Theoretical values of the total scattering coetticlent from its Rayleigh value of 4. The colour of the scat- K for spherical particles of three different refractive indices. tered light therefore becomes less saturated. Complete (After Sinclair and La Mer, 1949.) 318

be as high as 5 for certain values of a — i.e, the par- dent beam if the particles are small compared withthe tides may act as having five times their actual areas, wavelength. Iflarger particles are present, the degree If the particle size range is small, then K oscillates with of polarisation diminishes, and for anisotropic particles wavelength, but if the size range is comparable to the the direction of the maximum may become oblique to wavelength range of visible light (—~2: 1), the colour ef- the incident beam; “neutral points” of zero polarisa- fects tend to be washed out, and the scattered light tion may appear, on opposite sides of which the light tends towards white. is polarised in planes at right angles to each other Some specific features and consequences of these (Tyndall, 1869; Green and Lane, 1957). Eventually, actual curves are: when the particles are so large that the scattering be-

(1) For values of a to the left of a maximum, the comes specular, all polarisation vanishes — as for scattering coefficient decreases as the wavelength in- white cumulus clouds. creases, i.e. the transparency increases with increasing Some lunar polarisation anomalies have been re- wavelength; the transmitted light will therefore be ported (Dzhapiashvili and Xanfomaliti, 1966); tran- reddened, and the scattered light will be bluer. sient polarisation effects probably imply temporary Rayleigh scattering is reached in the limit, dust clouds, and may be a sensitive method of detect- (2) For particles large enough to put a to the right ing them. It would therefore be useful to include of a maximum, the converse applies: thus a cloud of - polarimetry in any observationalprogram for the fine volcanic dust in the atmosphere can make the sun investigation of transient lunar events. or the moon appear greenish or blueish. (3) If the range of radii is narrow, and such that a corresponds to a peak for green light, then the trans- 8. Conc1~ions mitted light may appear magenta. Experimental results with smokes and colloidal sus- We have considered a number of physical processes pensions are in fairly good agreement with the theoret- that could in principle cause transient changes on the ical curves. Particularly strong coloration is observed lunar surface. However, some of them do not seem when the size range is narrow; the different colours likely to produce effects that would be visible from present are at their brightest at specific angles, which the earth. amounts to angular dispersion as for spectra. Johnson The most likely explanation of events involving ob- and La Mer (1947) called this effect “higher-order scuration of surface detail is that gas emitted from the Tyndall spectra”. All these colour effects are discussed surface raises clouds of surface dust. The dust grains in detail inbooks by Middleton (1952) and by Green may be levitated by continuing gas emission, or they and Lane (1957), where it is further emphasised that may simply be in ballistic flight after initial projection the saturation and brightness of the colours both in- by the release of trapped gas. crease with uniformity of particle size. For events involving an increase of brightness, The main restraint on the extension of the theoret- luminescence, including thermoluminescence, seems ical analysis by Mie and van de Hulst to more realistic tobe far too faint to be seen from the earth; incandes- situations was the formidable amount of computation cence could be seen, but is very unlikely to occur. The involved; however, considerable progress has been only emission process that cannot easily be dismissed made in the last few years, now that large computers on efficiency grounds is that of lightning-type dis- are generally available. The results reinforce the con- charge, and we conclude that the conditions required clusion that scattering effects of this type, occurring for this to take place could just about be met in lunar in temporary gas-borne clouds of fine lunar dust of a dust clouds raised by gas emission. Such a process may rather narrow size range selected by gas levitationmight be consistent with reports of sparkling red light, and be responsible for some of the weakly coloured tran- of flashes. sient events that have been reported (Rawlings, 1967). The other type of effect that could produce visible Scattered light also tends to be polarised, in agree- brightness changes involves modulation of incident sun- ment with classical theory, the degree of polarisation light, rather than emission; this could occur if the reaching 100% in a direction perpendicular to the mci- albedo were changed by disturbing surface dust, for 319 example by gas emission. For the change to be tran- Dzhapiashvii, V.P. and Xanfomaliti, L.V., 1966, Electronic sient, the dust would have to settle back to its previous polarimetric images ofthe moon, In: W.N. Hess, D.H.M complex structure; it might well do this under high- Menzel and J.A. O’Keefe (•Editors), The Nature of the Lunar Surface, Johns Hopkins Press, Baltimore, Md., pp. vacuum conditions, although laboratory tests have so 275 —277. far only been conducted at atmospheric pressure, and Gänger, B., 1953. Der elektrische Durchschlag von Gasen, these did not show a return to the original low albedo. Springer, Berlin, 581 pp. (see especially Ch. 10). The theory of the scattering of light, especially by Garlick, G.F.J., Steigmann, G.A. and Lamb, W.E., 1972. Ex- particles of rather uniform size, provides a possible ex- samplesplanationstudies.of transientNaturelunar(London),phenomena235: 39—40.based on lunar planation of some colouredbrightness increases as be- Geake, J.E., Walker, G., Mills, A.A. and Garlick, G.F.J., 1972. ing caused by sunlight scattered by dust clouds. Luminescence of lunar material excited by electrons. Proc. To sum up, all the types of transient event reported, 3rd Lunar Sci. Conf., Geochim. Cosmochim. Acta, Suppl. if indeed they occur on the moon at all, seem to be Geake,3, 3:J.E.,2971—2979.Walker, G. and Telfer, D.J., 1977. The cause and best explained in terms of various physical processes significance- of luminescence in lunar plagioclase. Proc. R. that may be associated with dust raised by gas released Soc. London, Symp. The Moon, June 1975. Phios. from the lunar surface. Trans. R. Soc. London, Ser. A, 285: 403 —408. Green, H.L. and Lane, W.R., 1957. Particulate Clouds, Dusts, Smokes and Mists. Spon, London, 452 pp. Greenacre, J.A. and Barr, E., 1963. A recent observation of Acknowledgements lunar color phenomena, Sky Telesc., 26: 316. Hatakeyama, H. and Uchikawa, K., 1951. On the disturbance We are grateful for helpful discussions with of the atmospheric potential gradient caused by the erup- A. Dollfus ofParis Observatory,Meudon,and with col- tion-smoke of the volcano Aso. Pap. Metereol. Geophys., leagues at UMIST including J.R. Himsworth on fluidised 2: 85—89. Herschel, W., 1787. An account ofthree in the moon. beds, M.G. Rusbridge on flow processes in gases, ‘~ Philos. Trans. R. Soc. London, 77: 229—23 1. Latham on electrostatic charging processes, K. Phillips Johnson, I. and La Mer, V.K., 1947. The determination of the on electrical discharge in gases and P.F. Browne Ofl particle size of monodispersed systems by the scattering of scattering processes. We are grateful to The Royal light. J. Am. Chem. Soc., 69: 1184—1192. Society, London, for travel grants. Kopal, Z. and Rackham, T.W., 1963. Excitation of lunar luminescence by solar activity. Icarus, 2: 48 1—500. Kozyrev, N.A., 1961. Physical observations of the lunar sur- face. In: Z. Kopal (Editor), Physics and Astronomy ofthe References Moon, Academic Press, London, Ch. 9, pp. 361—383. Kozyrev, N.A., 1963. Volcanic phenomena on the moon. Na- Alter, D., 1963. Pictorial Guide to the Moon. Barker, London, ture (London), 198: 979—980. 183 pp. (see especially Ch. 16) Latham, G.V., Ewing, E., Press, F., Sutton, G., Dorman, J., Anderson, R., Björnsson, S., Blanchard, D.C., Gathman, S., Nakamba, Y., Toksöz, N., Lammlein, D. and Duennebier, Hughes, J., Jónasson, S., Moore, C.B., Survilas, H.J. and F., 1972. Passive seismic experiment. Prelim. Vonnegut, B., 1965. Electricity in volcanic clouds. Science, Soc. Rep., NASA SP-289, Washington, D.C., Ch. 8. 148: 1179—1189. Latham, J., 1964. The electrification of snowstorms and sand- Blair, LM. and Edgington, J.A., 1970. Luminescence and storms. Q.J.R. Meteorol, Soc., 90: 9 1—95. thermoluminescence under 159 MeV proton bombardment Middlehurst, B.M., 1968. Events on the moon. In: P. Moore of the lunar material returned by Apollo 11. Proc. Apollo (Editor), Yearbook of Astronomy 1968. Eyre and Spot- 11 Lunar Sci. Conf., Geochim. Cosmochim. Acta, Suppl. 1, tiswoode, London, 221 pp. 3 : 2001—2012. Middlehurst, B.M. and Moore, P.A., 1967. Lunar transient Cameron, W.S., 1976. Proc. Spec. Symp., 7th Lunar ScL Conf., phenomena: topographical distribution. Science, 155: NASA Johnson Space Center, Houston, Texas, March 1976. 449—45 1. Cameron, W.S. and Gilheany, J.J., 1967. Operation moon Middleton, W.E.K., 1952. Vision through the Atmosphere.

blink and report of observations of lunar transient phe- ~- University of Toronto Press, Toronto, Ont,, 340 pp. nomena. Icarus, 7 : 29—41. Mie, G., 1908. Beiträge zur Optik trüber Medien, speziell Coleman, S.N., 1949. Volcanoes, New and Old. Museum Press, kolloidaler Metallosungen. Ann. Phys., 25: 377—445. London, 212 pp. (see especially p. 180). Miller, G.H. and King, I.R., 1966. An experimental study of Davidson, J.F. and Harrison, D., 1963. Fluidized Particles. fluidization processes under lunar conditions. NASA Con- Cambridge University Press, London, 156 pp. tract. Rep. CR-627. 320

Mills, A.A., 1970. Transient lunar phenomena and electrostatic Runcorn, S.K., 1976. Proc. Spec. Symp., 7th Lunar Sci Conf., glow discharges. Nature (London), 225: 929—930. NASA Johnson Space Center, Houston, Texas, March Nash, D.B. and Conel, i.E., 1971. Luminescence and reflec- 1976. tance of Apollo 12 samples. Geochim. Cosmochim. Acta, Salisbury, J.W. and Glaser, P.E. (Editors), 1964. Studies of Suppi. 2, 3: 2235—2244. the characteristics of probable lunar surface materials. U.S. O’Keefe, J.A., 1970. Tektite glass in Apollo 12 sample. Air Force Rep. AFCRL-64-970. Science, 168: 1209—1210 (ref. 16), Sidran, M., 1968. Thermoluminescence of the Moon. J. Geo- Pal, S.!., Hsieh, T. and O’Keefe, J.A., 1972. Lunar ash flows: phys. Res., 73: 5 195-5206. isothermal approximation. J. Geophys. Res., 77: 3631— Sinclair, D. and La Mer, V.K., 1949. Light scattering as a mea- 3649. sure of particle size in aerosols. Chem. Rev., 44: 245—267. Penzel, E., 1965. Reported in New Sci., 26: 842. Sippel, R.F. and Spencer, A.B., 1970. Luminescence petro- Petrova, N., 1967. Reported in New Sci., 28: 74—75. graphy and properties of lunar crystalline rocks and brec- Purser, A.G., 1973. The reflectance of static and moving cias. Proc. Apollo 11 Lunar Sci. Conf,, Geochim. Cosmo- dusts. Dissertation, University of Leicester, Leicester. chim. Acta, Suppl. 1, 3: 2413—2426. Rawlings, G.C.I., 1967. Lunar transient phenomena: quasi- Sun, K.H. and Gonzalez, J.L., 1966. Thermoluminescence of Alpenglilhn ortwilight-colour theory. J. Br. Astron. Assoc., the Moon. Nature (London), 212: 23—25. 77: 309—318. Tyndall, J., 1869. On the blue colour of the sky, the polariza- Rayleigh, J.W.S., 1899. On the transmission of light through tion of skylight, and on the polarization of light by cloudy an atmosphere containing small particles in suspension, matter generally. Phios. Mag., 37: 384—394. and on the origin of the blue of the sky. Philos. Mag., 47: Van de Hulst, H.C., 1957. Light Scattering by Small Particles, 375—384. Wiley, New York, N.Y., 470 pp. Rowe, P.N., 1965. Fluidized beds. Sci. J., 1: 59—65.