a synthesis of our present knowledge of interstellar dust

A S Y N T H E S I S OF O U R P R E S E N T K N O W L E D G E

OF I N T E R S T E L L A R D U S T

P A U L S. WESSON St. John's College, Cambridge

and

Dept. Applied Maths. and Theoretical Physics, Cambridge University, U.K.

(Received 10 September, 1973)

Abstract. The present state of knowledge as regards interstellar dust is reviewed in Section 1 (Intro- duction); Section 2 (Composition of Dust Grains: graphite, silicate, dirty-ice, diamond); Section 3 (Size of Grains: mainly r -~ 10 -6 cm); Section 4 (Charge and Temperature of Grains: charge varies from 1-10 electrons (H I clouds) to 500 electrons (HIr clouds); temperature of grain material is about 10-20K); Section 5 (Distribution and Origin of Grains: confined mainly to discs and arms of spiral galaxies, having had a passive origin by etttux from late-type stars or carbon-stars); Section 6 (Cos- mogonical and Cosmological Aspects of Interstellar Grains : accretion by electrical-image forces of one dust grain onto a similarly-charged grain links up the absence of dust and gas in elliptical galaxies with the absence of a magnetic field of the type found in spirals. The origin of the 3 K background radiation field could be produced by a population of rotating silicate grains ofr ~ 10 -7 cm); Section 7 (Conclusion).

1. Introduction

The idea that large amounts of dust and other matter might be found in the spaces

between stars was discussed as long ago as 1929 by Eddington (1929), who speculated

during a radio broadcast that the accretion of matter by stars and the continuing

contemporaneous loss in the form of radiation might tend to an equilibrium state

in which stars of roughly constant mass would populate nebulae consisting of diffuse

matter of various kinds. This speculation was subsequently developed and studied

until, a decade later, D u n h a m (1939) was able to set limits on the amount of dark

matter in interstellar space by observations o f X 2 0 r i o n i s , including an estimate o f

the interstellar electron density o f 1-10 cm -3. This study, of the mot ion of stars,

restricted the density o f intergalactic matter to 0 < 3 x 10 .24 gm cm -3, though

D u n h a m was led to an unacceptably low density o f dust (10 -26 gm cm -3) by using

an incorrect analogy with meteoritic matter (5 x 10 -25 gm cm-3) . It is now realised

that the amount o f dark material between the stars is large enough to cause reddening

of stellar light sufficient to reduce a star of the brightness of Sirius to that of an 11 th

mag. object: this (most extreme) case is demonstrated in the observations of a severely reddened star in the Cyg OB2 stellar association (Bromage, 197l).

2. Composition of Dust Grains

Observations o f extinction and polarization o f starlight resulted in suggestions that

dust was primarily composed of iron, ice, graphite, d iamonds or silicates; or a

combinat ion o f any or all of these. The attributes and shortcomings o f these various possibilities have been discussed in the excellent book of Wickramasinghe (1967).

Space Science Reviews 15 (1974) 469-482. All Rights Reserved Copyright �9 1974 by D. Reidel Publishing Company, Dordrecht-Holland

470 PAUL s. WESSON

Iron spheres of diameter 0.02/~ can give agreement with extinction curves, but an interpretation of stellar absorption (1 mag kpc -1) on this hypothesis would lead to an unacceptably high abundance of iron in comparison to the accepted solar abundance (op. cit., p. 109), even after new determinations of the latter with modern data on oscillator strengths. Some (anionic) iron does appear to be present in supernovae

explosions, but it is doubtful if this represents a notable rate of injection into the interstellar medium (Manning, 1972).

Carbon, in the form of graphite, is expected to form grains by condensation of vapour in the outer atmospheres of stars, and the initial crystals, which are either pristine or formed around seed nuclei, can grow in an atmosphere of carbon vapour to a radius of order 10 .7 cm or 10 .6 cm for sphere-shaped grains, depending on the nucleation efficiency, with similar sizes for platelets, the timescale for growth being about 107 s. This process occurs in the atmosphere of N stars, which expel the grains and cause a rate of increase of grain density of 5 x 10 -37 gm cm -a yr -1, the graphite grains having a lifetime ~ turnover time of interstellar matter in the neighborhood of the Sun, i.e. 109 yr, < Galaxy's age of 101~ yr (op. cit., p. 89). By this process, the

105 N stars observed in the Galaxy can build up a grain density of 10 .27 gm cm-3

in 109 yr. There is thus a well-defined mechanism for creating graphite grains and injecting them into interstellar space, giving this hypothesis an inherent advantage over the ill-fated iron-spheres concept.

Ice grains can grow quickly in space, reaching a size of ~ 10 .4 cm in about 3 x 109 yr. This large size, relatively speaking, does not give agreement with the desired wavelength -1 tendency of the extinction curve, and so a random destructive process must also be postulated if ice grains are to represent a workable hypothesis. The dirty-ice model which developed to give agreement with extinction, polarization and albedo data, fails to give agreement with observations of stars which illuminate nebulae which are seen by back-scattered light. In addition, efforts to obtain a dust-grain model to explain the entire range of the interstellar extinction curve gradually forced

the abandonment of the ice model in favour of a compound system in which graphite grains have mantles of ice. The latter model can explain all the available data on extinction, polarization, albedo and back-scatter, and so represents a truly acceptable working basis.

A graphite particle of initial radius 5 x 10 .6 cm will grow an ice mantle out to 10 .4 cm in 109 yr (Wickramasinghe, 1967, p. 97), provided the grain temperature is less than about 8 K, so ensuring that atoms of O, C, N which strike the grain are able to stick, and freeze on. This process is limited by the disruptive effects of cloud collisions, by encounters of clouds with hot O and B stars, and by the supply of oxygen atoms needed to enable the mantle to grow. Pure ice grains are destroyed in H n regions in a time of <104 yr for a 10 .5 cm particle. Thus, it is expected that the graphite cores, being refractory, can survive the destructive processes just mentioned whereas the ice mantles will melt. Collisions between H I clouds usually take place at velocities below the 20 km s -1 relative speed necessary to cause evaporation of the grains by direct collision, but the transformation of H I to H II clouds, a phenom-

A SYNTHESIS OF OUR PRESENT KNOWLEDGE OF INTERSTELLAR DUST 471

enon occurring on average once per I08 yr per cloud, will lead to removal of the ice mantle. The finite amount of oxygen in the interstellar medium also provide an attained limit to the effectiveness of ice mantles, since a graphite core of 5 • l 0-6 cm radius can only grow to 10 -5 cm before the interstellar oxygen is exhausted. This conclusion still holds on average, even though grains are losing and gaining oxygen atoms continually.

The graphite core + ice mantle theory does not give complete agreement with the wavelength dependence of polarization on the size parameters deduced for extinction data (namely, r .... =0.054/~, dirty-ice mantle of outer radius r=0.16 #). This is not serious, however, because there is no compelling reason for believing that the grain sizes responsible for most of the polarization are the same as those giving the majority of the extinction: a field of 10 -5 G is needed to allign composite particles as just just specified and give agreement with polarization measurements as interpreted on the Davis-Greenstein theory; but an average galactic field of smaller strength will leave those grains unalligned which contribute most to the extinction, while still giving agreement with the polarization data as an effect predominantly of large graphite particles. The latter is possible with a field of only 10 -6 G, so allignment is a very efficient process.

There is little doubt that the preferential tendency of grains in a magnetic field to become alligned, and so polarize light passing through the field, is the mechanism actually operating in interstellar space (Martin, 1971). The only serious alternative to this mechanism of Davis and Greenstein is a proposal by Gold (see Wickramasinghe, 1967, p. 141), who considered dust clouds streaming supersonically through the interstellar gas, leading to a transfer of angular momentum tending to cause a total angular velocity lying roughly at 90 ~ to the relative velocity of streaming. Elongated dust grains would tend to spin with their long axes perpendicular to the angular velocity; polarization observations indicate that this direction of the orientation of the long axes is parallel to the galactic plane, implying that the clouds should move preferentially at right angles to the galactic plane: there is no evidence for a general tendency of this kind to the extent required, so that Gold's mechanism is probably not applicable.

A recent proposal of Harwit (1970) is that photons, possessing intrinsic angular momentum h directed along the line of propagation, interact with grains and set them spinning, in the sense of rotational Brownian motion. The grains will spin with their long axes perpendicular to the line joining the grain and the star which is emitting the photon radiation. Background light is polarized by this mechanism, the overall effect in the Galaxy being polarization parallel to the Galactic plane, since most radiation originates in the Milky Way strip. This is an attractive hypothesis, but its full consequences have not been worked out. The conclusion at the moment must therefore be that the use of the Davis-Greenstein mechanism is valid, and that polarization data imply the existence of graphite grains covered with ice, and by necessity, of some pure graphite particles as well.

The success of graphite in exploring the known properties of interstellar material led to the suggestion that diamonds might also be found in space. Diamond, however, is

472 PAUL S. WESSON

a questionable choice as regards its ability to explain the extinction curve over the whole range of the infrared, visible and UV, and in other of its optical properties (Donn and Krishna Swamy, 1969). Diamonds cannot form in cool stellar atmospheres, like graphite grains, although diamonds can form at 1000K if seed diamonds are present initially: this process is ultimately abrogated by an increasing tendency for graphite formation, unless a high pressure (10-50 atm) of hydrogen is present to remove it by reaction. In any thermodynamically unstable situation, graphite tends to form at the expense of diamond, and not vice versa, so in all probability diamond is not an important constituent of interstellar dust unless it forms in an exotic manner, e.g., on a substrate of SiC. This viewpoint is not taken by Landau (1970) who, by modifying the previously-suggested size distribution law to allow for removal of grains by cosmic ray break-up from

n(r)ocr -3's, to n(r)ocr-3"s(1-e-r/ro) , (1)

with ro=0.03 #m, finds that the extinction curve can be reproduced. On the other hand, Velovykin (1970) has claimed, by analogy with carbonaceous meteorites, that the carbon of interstellar space is in an aromatic form. In the face of these contra- dictory opinions, it is as well to be content with the graphite and ice-mantle picture, and to leave aside the possibility of other forms of carbon until such time as they are shown to be needed.

A new phase of the subject appeared with the discovery in interstellar space of silicates, these being implicated by observations of broad absorption features at 20 and 10/tm in the direction of the Galactic centre (Hackwell et al., 1970). Silicates probably form only about 0.4% by mass of the interstellar medium, and their importance is often overrated: unrealistic interpretations of the data lead to such claims as the possibility of differentiating almandine from andradite as the prevalent silicate (Manning, 1970b) and the identification of phlogopite (Manning, 1970a); but these excessive claims of exactitude were repudiated by Duley (1970).

A mixture of graphite, iron and silicate grains, with sizes respectively of 0.045- 0.07 #m, ~<0.02/~m, 0.15-0.18/~m, can explain the observed extinction curve and low grain-albedo (Wickramasinghe and Nandy, 1970a, b), and combined data on polarization and extinction are important in delineating the best model of the interstellar medium (Nandy and Seddon, 1970b). By working the method in reverse, valuable results on interstellar absorption may be obtained, especially in the UV (Nandy and Seddon, 1970a; Henry and Carruthers, 1970). The philosophy that all types of possible grains ought to be considered as candidates for interstellar dust has also been supported by Gilra (1971), who considers silica, silicon carbide and graphite to be present in the ratios of 5: 4:1, and the whole problem of multicomponent models of graphite and silica in particular has been comprehensively studied by Wickrama- singhe (1969).

3. Size of Grains

The large visual extinction, of 1 mag. kpc -1, which is observed, in conjunction with


the approximate wavelength-1 extinction law, implies the existence of grains of size

10-6-10 .5 cm. The cross-section for the extinction of optical radiation is about

0.3 rcr 2, i.e. a reasonable fraction of the geometrical cross-section, and the observed extinction requires a dust density ofQ ~> 10 - 26 gm cm - 3 (Wickramasinghe, 1967, p. 54). Such a mass density, if the grains are of graphite, can be provided by continuous injection at a rate of d~/dt=5 x 10 .37 gin cm -a yr -1 from the known N stars in the Galaxy, as noted elsewhere. The grain density will build up to 10 -26 gm cm -3 in

about the Galactic lifetime of 101~ yr. The growth of a particular grain can be assumed to progress by capture of O, C and N atoms in an ambient gas having a Maxwellian distribution of atomic velocities. The rate of increase of mass of a grain is then

d r 4~r2 d t x (grain density),

and the grain radius grows linearly with time, from an initial radius r~, according to

r = r i + c o n s t x t .

With suitable numerical constants inserted, a growth rate is found which would give a radius of 3 x 10 -s cm in times of order 10 s yr (op. cit., p. 58). When grain growth has come to equilibrium with forces tending to destroy or erode the grain, the main one of which is sputtering in H II regions, the destruction probability per unit time of a grain of radius r may be written P(r)=Dr s ( / )=constant) ; when destruction and growth reach equilibrium, the number distribution as a fraction of radius then settles down to

n(r) = n(ro) exp dr/dt)-(s+ 1) :+1 , (2)

as shown by Wickramasinghe (1967, p. 72). Due to uncertainties in the determination

of P(r) and dr~dr, the usual practice is to choose an r o to give best agreement with observation while adopting an expression of the form

n ( r ) = n ( t o ) e �9 (3 )

The best value of r o is chosen to fit this exponential law and the observations of

oo

(i) extinction, Q (2) = ( Q (r, 2) ~rZn (r) dr

0

o (2 ) - Q (21 ) (ii) normalised extinction, E (2) = const q

0 (2~) - O (21)'

where 21 and 22 are standard wavelengths. It is found that r o ~-0.075 #. There is, by the exponential law, a great population of small grains compared to large; the latter are an important constituent of the medium; and extinction measures show the in- fluence of those grain having r > r 0 (Spitzer, 1968, p. 80). Overall, the density of grains of the sizes noted ( - 1 . 4 x l 0 -26 gm cm -3) amounts in mass to about 1~ of the

474 P A U L S. WESSON

usually assumed H I gas density, the two populations, gas and dust, occurring together in spiral arms of galaxies, as indicated by observations of coincident extinction (by grains) and Ha emission (neutral gas).

4. Charge and Temperature of Grains

Most grains carry negative charges due to capture of electrons from the surrounding plasma. By considering the trajectory of an ion or electron deflected by some means to a grazing incidence with a grain of radius r, and assuming equipartition of energy between grains and ions, Spitzer (1941) was able to derive collision cross-sections for electrons and (singly charged) ions with a dust grain of radius r and charge Zo (units of e) given by

{ 2Zge2~" { 2Z~ ae = ~r 2 1 ~v~rJ ' 6i = ~zr2 1 + mlv2~r). (4)

When integrated over a Maxwellian velocity distribution, assuming equilibrium to be set up between gains and losses of charges,

oo cx)

f ne~VJ(Ve)dV~=f n~aiVif(Vi)dVi, 0 0

it transpires that the average grain potential turns out to be negative, V ~ - -2.5 kT (Wickramasinghe, 1967, p. 95). Thus, for H I regions where T~- 100K, the average charge on a grain is about 1 electron, while in H II regions, where T~ 104 K, the average charge is about 100 electrons. These values are uncertain by a factor of about four, and Spitzer (1968, p. 146) gives the average charge in H II regions as 450 electrons. Not all grains will be charged, anyway, and the spread in number of charges per grain will no doubt be considerable. Changes in charge by + 1 electron are expected to occur about once every 105 s (Spitzer, 1968, p. 208). Obviously, some grains with a positive charge will be found, owing to the photo-electric emission of electrons from grain sur- faces. This will be a major effect near bright, early-type stars (Spitzer, 1968, p. 145), and plays a significant role in Wickramasinghe's theory of retention of grains near galactic nuclei (Wickramasinghe, 1970a) and in the radio emission from spinning dust grains in H II regions (Hoyle and Wickramasinghe, 1970b).

Apart from the importance of the temperature of the medium in which the grains are immersed, the temperature of the grains themselves is of crucial importance both for the nature of their optical properties and for their very existence. Metallic grains have large Q, (2), the efficiency factor for absorption, in the visible, but a small value in the infrared, leading to temperatures of 30-100K as estimated by Spitzer (1968, p. 144) and 56-118K as estimated by Van de Hulst for 0.1 # metallic spheres (see Wickramasinghe, 1967, p. 93). The temperatures for other materials are somewhat lower, such as 30K for 10 -5 cm graphite grains and 15K for ice grains of 0.2 grad. These temperatures become lower for larger-sized particles because Q, (it) varies with


the particle radius: ice particles with radii > 1 # show a tendency in which the grain temperature drops by a half as the radius increases from 1 to 10/~, and a similar effect holds for graphite (Spitzer, 1968, p. 145). Dielectric materials probably have strong absorption bands in the infrared, leading to expected grain temperatures of 10-20K in H I regions (see Wickramasinghe, 1970c). Such estimates of the temperature are not very certain, however, since lattice defects are known to seriously affect the temperature of crystal lattices: weakly-bound impurity atoms cause electric dipole transitions in the far infrared (2-~0.1 cm), and lead to broad emission bands tending to result in a temperature of about 4K (Wickramasinghe, 1967, p. 93) for particles of 10 -5 cm radius.

5. Distribution and Origin of Grains

The dust lanes seen in some galaxies, and the discs of obscuring matter seen in objects like NGC 4594 (the 'Sombrero Hat'), leave no doubt that dust is concentrated in the discs of spiral galaxies. Carbon can condense in the outer atmospheres of cool stars, and direct evidence for the close association of dust with stars comes from observations of Bok globules (Reddish, 1971) and of the intrinsic polarization of Mira variables, which latter stars seem to be losing mass by efttux of carbon platelets, at about 10 .6 M o yr -1 (Donn et aI., 1966). A mechanism for mass ejection from red giants has been proposed by Wickramasinghe, Donn and Stecher (1966), and by coupling of the dust grain motion to the surrounding gas provides an explanation of the observed steady gaseous outflow at 10 km s -1 from stars such as e Orionis. It has also been suggested that supernovae produce dust grains composed of various elements by condensation during the 30-1000 yr period during which the opacity of the material surrounding the site of the explosion remains high (Hoyle and Wickramasinghe, 1970a). The predicted rate of increase of grain density is d~/dt-~ 7 x 10- 36 gm cm- 3 yr- ~ surrounding the site of the explosion remains high (Hoyle and Wickramasinghe, 1970a). Collissions of grains, perhaps escaping from cool stars (Gilman, 1973), would result (Bandermann, 1972) in a bimodal size distribution for velocities -~ 10 km s -1.

In connection with the supernovae theory, dust surrounding pulsars would be expected to smear out pulsed radiation by scattering, causing the time coherence of the pulse to be largely lost, and increasing the angular size of the source to that observed: Slysh (1969) claimed that this process would smear 90% of the radiation from NP 0532, assumed to be all of pulsed origin, leaving only 10% still detectable as an actual pulse (see Naranan and Shah, 1970), but an analysis using a model of grain size given by

n(r) dr = const exp {-5(r /ro) 3} dr (5)

shows that in fact only a 20% smearing effect is possible. The same conclusion is reached by Bowyer et al. (1970), who find that a 90% smearing for the Crab Nebula pulsar gives parameters which, applied similarly to the source Sco XR-1, result in disagreement with observation.

Dust may also originate in galactic as opposed to stellar explosions, as again

476 PAUL S. WESSON

suggested by Hoyle and Wickramasinghe (1968). Expanding gases, before their 10S-yr dispersal time is reached, can form condensing phases of C, Si and Fe, the exploding gaseous material being thrown out from a postulated 105 M o object, which becomes catastrophically unstable in the galactic centre concerned. If dust grains are widely distributed by such a galactic explosion or by injection into the interstellar medium from numerous discrete sources, the grains may well play a vital role in theories of star formation and cosmology as discussed by Hoyle et al. (1968), and examined in Section 6 below.

Whatever the main source of dust grains, their presence in the disc of our own Galaxy and in other galaxies leads to notable consequences: in particular, dust may scatter X-rays in the disc, so explaining the Galactic component of the diffuse X-ray background. Wickramasinghe originally suggested that this enhancement was caused by scattering of the isotropic radiation background (Wickramasinghe, 1970b) but as was quickly pointed out by Webster (1970) and Mack (1970), there would be as many photons scattered out of the line of sight as into it, so no enhancement can result from such scattering. Wickramasinghe (1970e) pointed out in reply that scattering of X-rays by grains could still produce a diffuse galactic component if the radiation has its origin in discrete sources in the plane of the Milky Way; also, soft galactic X-rays are expected to have a noticeable effect on the allignment of grains (Wickramasinghe, 1970d). In the direction of the galactic centre, thermal emission from dust may be responsible for the infrared excess observed there. This hypothesis (Wickramasinghe, 1971) was proposed in analogy with the large infrared luminosities of Seyfert galaxies: the spectrum of the radiation from the centre of our own Galaxy is in some ways similar to that of Seyfert galaxies like NGC 4151, and the infrared radiation could have its origin in a model common to most galactic centres, in which a dust disc sur- rounds a central source of UV radiation, low-energy cosmic rays (.,, 10 MeV) or X-rays (30 keV). Time variability of Seyfert luminosities effectively restricts the exciting source to one or other of the last two mentioned.

Wickramasinghe has developed this theory, finding that the dust grains (graphite) must be embedded in a region of ionised hydrogen in order that they be retained against the effect of radiation pressure (Spitzer, 1968, p. 207) near the galactic centre: the grains are positively charged by the photo-electric effect, and a decoupling time of 10s-101~ yr of the dust with respect to the plasma can be obtained, leading to a mass loss of 103 M o over 101~ yr by outflow of grains from the centre (Wickrama- singhe, 1970a; Okuda and Wickramasinghe, 1970). By calculating the ratio of radiation pressure to gravitational attraction, Chiao and Wickramasinghe (1972) find that uncharged grains are driven out of spiral galaxies at 105-106 cm s-1 into intergalactic space, the time for a grain to drift a distance of one scale height, normal to the plane of gas and dust forming the galactic disc, being 3 x 10 9 yr, the mass loss rate from a Galaxy like our own being 10-1-10 -2 M o yr -1. This efltux will build up a dust density in the Universe of ~ 10 -34 gm cm -3. Since there is no reason to expect the etttux to be inhibited, this may represent a likely figure for the density of the intergalactic medium (Schmidt, 1971). Charged grains, with a charge Zoe given by


the relation of Spitzer (see Chiao and Wickramasinghe, 1972, p. 365) as

Zge2/rKT ~- 2.51

would probably still escape at this rate along open-ended field lines in the Galaxy. The injection of grains into interstellar clouds, as assumed in a lot of work on

radiation processes, has been investigated by Wickramasinghe (1972a), who finds that the grains are stopped if the gas number-density is of order 10-30 cm -3. The inter- action of rotating interstellar grains with cosmic low-frequency radiation has an especially noticeable effect on small iron grains (Martin, 1972a), and, at higher frequencies, light impinging on grains modifies the tendency to allignment under most mechanisms, notably magnetic fields (Martin, 1972b). Observations of circular polarization in light which has traversed a dust field can give information on the refractive index of the grains and so enable a decision to be reached as to whether the grains are dielectric or metallic (Martin, 1972c). Circular polarization of interstellar origin has indeed been observed in the direction of the Crab nebula, and indicates that the majority of the grains have a dielectric composition (Martin et al., 1972).

Two dust grains in a radiation field experience an attractive force due to radiation pressure because each grain reduces the radiation density at the other by an amount

nrZUQ, 4/ra 2 '

where U= energy density, Q~ = efficiency factor for absorption (there is a correspond- ing factor Op for radiation scattering pressure), and a = distance between the (identical) grains of radius r (Spitzer, 1968, p. 211 ; Rees, 1971). This modification of the radiation field produces a force of attraction between the grains of

~zQ~Qpr~U Fr - 4a 2 (6)

provided the absorbed radiation is emitted at a lower frequency for which Qp is negli- gibly small. For dielectric particles in an H II region, the grains and gas are effectively coupled together by collisional friction, but in an H I region the gas and grains may move separately. Charged grains, of course, tend to gyrate about field lines if there is a magnetic field present, but the time for which a grain can travel smoothly trans- verse to the field is limited by the average time between impacts with electrons, which is ~ 105 s for a gas density of about one atom cm-a. This is of considerable importance in H I clouds, where the average grain charge is 1-10 electrons, but not in H ti regions where it is of order 100 electrons. Taken in conjunction with the short spiralling times of grains around lines of the magnetic field, it must be concluded that the motion of grains relative to the ambient medium is, on average, limited to a direction parallel to B, as enunciated by Spitzer (1968, p. 208).

Wesson (1973) has made use of this notable streaming tendency of grains to show how grains in H II regions can attract one another, coalesce, and form, eventually, very large grains. Wesson uses the theory of electrical images to derive the force of

478 PAUL S. WESSON

attraction between two, similarly-charged grains due to the induced charge (of opposite sign) on one sphere owing to the proximity of the other sphere. This polarization attraction force is ~ 10 iv times stronger than the gravitational force between two grains, and 1015-1016 times stronger than the radiation force (6) above. Eventually, given a cloud which is initially composed of grains of all possible sizes, those grains of size r >10 .5 cm are removed by accretion. After 109-10 l~ yr, the initial grain distribution, the motions of whose particles are governed by Spitzer's streaming mechanism, is split up into (a) grit-sized grains of mass 10-9-10 -8 gm (r ~ 0.1 ram), which are undetectable by ordinary means; and (b) grains of 10-6 cm size, which are presumably those being observed in polarization and extinction measurements. The argument which leads to this result implies that the grains had a passive origin, and not an explosive one: origin in late-type stars is thus supported, while origin in supernovae or galactic explosions is ruled out. The reasoning of Wesson (1973) has the advantage over similar hypotheses of explaining why grains of 10-6 cm size are observed, since, in all feasible theories of grain origin, there is no check on the production of grains of all sizes over a wide spectrum in r, the grain radius. The argument leads one to expect a distribution of the form

n(r)ocr -~, 3<~/~<4 (7)

6. Cosmogonical and Cosmological Aspects of Interstellar Grains

Cosmogonically, the account of accretion given by Wesson (1973) could explain the absence of dust in elliptical galaxies: if H I or H II clouds collide (at velocities of 10- 20 km s-I) , accretion of grains, and simultaneous fragmentation (into r ~ 10 -7 cm grains, if the relative velocity of collision is high), will occur. The presence of dust is thus synonymous with an absence of cloud-cloud collisions, which in turn implies the necessary presence of a (collision-inhibiting) magnetic field. This in turn suggests some connection with spiral structure, so that an argument could be constructed for elliptical galaxies which would work in the following sense: (i) E-type galaxies have no spiral arms, ~(ii) there is no large-scale ordered B-field in E-types, ~(iii) clouds containing dust collided uninhibitively during an earlier epoch, ~(iv) the dust was accreted into large bodies (r ~ 1 mm) or fragmented into small bodies (r ~ 10 -7 cm), both of which are undetectable with present techniques. Also, (v) the gas component of the clouds was dispersed by the repeated effects of cloud-cloud collisions.

The last paragraph is something of a speculation. In another direction, interstellar dust has been put to cosmological use, as it were, by attempts to explain galactic radiation of various types as being a consequence of a certain type of behaviour of dust grains.

The existence of dust in the vicinity of the galactic centre is connected with the possibility of explaining the non-thermal emission from that region on a dust-model basis. It is usually assumed that such radiation has a synchrotron origin, but Hoyle and Wickramasinghe (1970b) have proposed that spinning dust grains, acting as


rotating dipoles, are a plausible source for the radiation. Stars embedded in H II regions emit photons which remove electrons from the grains present, there being asymmetries in the charges carried by opposite hemispheres of the (dielectric) grains of about _ 10 electrons with a 200 electrons total charge (see elsewhere); this asym- metry gives rise to a dipole moment

p = e r Z ~ / 2 . (8)

These dipoles can be set spinning by photo-emission of electrons, electron capture, encounters with protons, collisions with neutral atoms, scattering and absorption of photons, and re-emission of photons. The last two processes in particular cause a high frequency of grain rotation v, leading to a rate of emission of radiation

16~Z4v4p 2

R - 3c 2 (9)

Grains of radius r <~ l0 -6 cm can cover the whole wave-band observed for the non- thermal emission from H II regions. The inner parts of the Sag A radio source can be explained by the presence of 3 x 10 a M o of grains, an intense high-energy source of 107 L o being indicated somewhere in the centre of the Galaxy. This example of the use of interstellar grains in cosmogony was taken further, however, into the cosmological realm, by the question of the origin and nature of the 'cosmological' 3 K background radiation. (Applications of (9) are given by Hoyle and Wickramasinghe, 1969.)

Hoyle e t al . (1968), have formulated a hypothesis in which graphite grains, coated with a layer of frozen hydrogen, condense and begin forming star-like bodies until the temperature has risen enough to evaporate the hydrogen: condensation then stops, and the temperature drops until the hydrogen can re-condense. In this way, the existence (but not, presumably, the isotropy) of the 3 K background is explained, since the critical condensation temperature is approximately 3K. This theory also explains the coincidence of the energy density of the microwave background (6 x l0 -la erg cm -a) with the approximately equal energy density produced by the hydrogen-burning reactions in stars. (This equivalence was unwittingly worked out by Eddington (1930, p. 371), before the microwave background was discovered). The discovery of silicates in space, subsequent to this theory, did not invalidate it, since the critical feedback mechanism can now operate by graphite re-radiating energy in the infrared to the silicate grains, which then re-radiate at longer wavelengths into space, thus providing the connection with the 3 K background; the star formation process, however, is now dependent on the silicate grains, which are expected to be cooler than the graphite grains.

The hypothesis of Hoyle e t al . (1968) has some rather serious objections to contend with, it should be noted. In particular, extinction observations indicate that most particles have come to equilibrium at about 17K, between absorption of energy from the interstellar radiation field and re-emission in the infrared. Such temperatures would cause catastrophic evaporation of solid hydrogen mantles, if such ever existed, since hydrogen gas evaporates off a l0 -5 cm grain in only a few seconds, the maxi-

480 PAUL S. WESSON

mum lifetime being 107s for grain temperatures of 3K (Krishna et al., 1969). Impurities of large cross-section tend to reduce the temperature, but even so it does not seem likely that the solid hydrogen-coated grains of Hoyle et al. (1968) can survive for times of the order of 106 yr, since they would be reduced to graphite particles in an astrophysically short time. A similar conclusion was reached by Greenberg and de Jong (1969), who asked the question: can solid hydrogen condense on interstellar grains? Their answer to this question was, to all intents, a negative one.

The origin of the 3 K background radiation, on the conventional interpretation, is beset with paradoxes, as outlined by Davies (1972). Wesson therefore set about the task of finding if it was possible to explain the 3 K microwave field as dipole emission from rotating, charged dust grains confined to the Galaxy and its halo (Wesson, 1973). The hypothesis runs in this way: grains of r ~ 10 -v cm may well be profuse in the Galaxy due to attrition of larger grains by cosmic rays. Such a population of small grains, immersed in a plasma of temperature 10*-106 K, and radiating at the rate (9) of Hoyle and Wickramasinghe (1970b), can explain the energy density of the 3 K background. The interstellar plasma, which has an electron density in the range 0.06-0.2 cm-3 at a temperature of ~ 104K (Mills, 1969; Bridle and Venugopal, 1969), is assumed to be energised by starlight, so that the coincidence in energy densities noted above is now explained in an indirect way, using silicate grains as intermediate agents. By comparing the expected grain dipole-radiation spectrum with the black-body curve, Wesson deduced that the grains must have a size distribution of

?.-4-

n (r) oc e~/r,/2_ 1" (10)

This distribution, which to a first approximation (~--const, suitably chosen) is n(r)oc r-*, agrees roughly with the distribution (7) and with the distribution (1) of Landau (1970), which latter two distributions are: n(?.)oc r -s or r -4, and n(r)oc oc r - s.s, approximately.

The cosmological and cosmogonical aspect of interstellar dust, it is now evident, have many similarities which suggests the real possibility of uniting several astronomi- cal disciplines via the use of the dust grain medium.

7. Conclusion

To conclude the above-given review of the study of interstellar dust, it is heartening to find that several aspects of dust theory, and of astronomy in general, can be synthesized into an acceptable overall picture which nowhere strains credulity beyond the point of reasonable acceptance. Grains of r ~ 1 0 -6 c m, composed of silicates and graphite are virtually certain to be present in our Galaxy, and presumably give rise to the extinction and polarization which is observed in the light from distant stars. The grains are probably alligned by magnetic fields, or alternatively by the effects of a Galactic photon flux, and have charges of 1-10 electrons (H I regions), up to 100-500 electrons (H II regions). These grains are probably the only surviving fraction


of a wide mass -p roduc t ion spectrum, the larger part icles having been accreted in the

way envisaged by Wesson (1973) or else f ragmented by impacts . The accret ion

a rgument has notable impl ica t ions for galact ic theory (Section 5). The presence of

a large number of smaller grains ( r < 1 0 -7 cm) cannot be observa t iona l ly proven

at present, bu t such a popula t ion , if it exists, could explain the existence of the 3 K

microwave background . One may thus summarise the present s i tuat ion opt imis t ica l ly

in that var ious branches of a s t ronomy can be shown to be inter- l inked via the med ium

of interstel lar dust ; and more such links m a y be expected to be fo r thcoming as the

result of future research in the field.

References

Bandermann, L. W.: 1972, Monthly Notices Roy. Astron. Soe. 160, 321. Bridle, A. H. and Venugopal, V. R. : 1969, Nature 224, 545. Bowyer, C. G., Mack, J., and Campton, M.: 1970, Nature 225, 125. Bromage, G. E.: 1971, Nature 230, 172. Chiao, R. Y. and Wickramasinghe, N. C. : 1972, Monthly Notices Roy. Astron. Soe. 159, 361. Davies, P. C.: 1972, Nature Phys. Sci. 240, 3. Donn, B., Stecher, T., and Wickramasinghe, N. C. : 1966, Astrophys. d. 145, 949. Donn, B. and Krishna Swamy: 1969, Nature 224, 570. Duley, W. W.: 1970, Nature 227, 1277. Dunham, T. : 1939, Proc. Am. Phil. Soc. 81, 277. Eddington, Sir A. E. : 1929, Matter in Interstellar Space, BBC Publications Ltd., London. Eddington, Sir A. E. : 1930, The Internal Constitution o f the Stars, Dover Publications Ltd., 1939,

New York. Gilman, R. C. : 1973, Monthly Notices Roy. Astron. Soc. 161, short communication, (3 p.). Gilra, D. P. : 1971, Nature 229, 237. Greenburg, J. M. and de Jong, T. : Nature 224, 251. Hackwell, J. A., Gehrz, R. D., and Woolf, N. J.: 1970, Nature 227, 822. Harwit, M. : 1970, Nature 226, 61. Henry, R. C. and Carruthers, G. R. : 1970, Science 170, 527. Hoyle, F. and Wickramasinghe, N. C. : 1968, Nature 218, 1126. Hoyle, F., Wickramasinghe, N. C., and Reddish, V. C. : 1968, Nature 218, 1124. Hoyle, F. and Wickramasinghe, N. C. : 1969, Nature 223, 459. Hoyle, F. and Wickramasinghe, N. C. : 1970a, Nature 226, 62. Hoyle, F. and Wickramasinghe, N. C.: 1970b, Nature 227, 473. Krishna Swamy, K. S. and Donn, B. : 1969, Nature 224, 788. Landau, R. : 1970, Nature 226, 1040. Mack, J. E. : 1970, Nature 228, 543. Manning, P. G. : 1970a, Nature 226, 829. Manning, P. G. : 1970b, Nature 227, 1121. Manning, P. G.: 1972, Nature 240, 547. Martin, P. G. : 1971, Monthly Notices Roy. Astron. Sac. 153, 279. Martin, P. G. : 1972a, Monthly Notices Roy. Astron. Soe. 155, 283. Martin, P. G. : 1972b, Monthly Notices Roy. Astron. Soe. 158, 63. Martin, P. G. : 1972c, Monthly Notices Roy. Astron. Soe. 159, 179. Martin, P. G., Illing, R., and Angel, J. R. P. : 1972, Monthly Notices Astron. Roy. Soe. 159, 191. Mills, B. Y. : 1969, Nature 224, 504. Nandy, K. and Seddon, H. : 1970a, Nature 226, 63. Nandy, K. and Seddon, H. : 1970b, Nature 227, 264. Naranan, S. and Shah, G. A. : 1970, Nature 225, 834. Okuda, H. and Wickramasinghe, N. C. : 1970, Nature 226, 134. Reddish, V. C. : 1971, Nature 232, 40.

482 PAUL S. WESSON

Rees, M. : 1971, in R. K. Sachs (ed.), General Relativity and Cosmology, 1969 Varenna Lectures, Academic Press, New York and London, p. 326.

Schmidt, M. : 1971, Observatory 91, No. 985, 209. Slysh, V. L : 1969, Nature 224, 159. Spitzer, L. : 1941, Astrophys. J. 93, 369. Spitzer, L. : 1968, Diffuse Matter in Space, J. Wiley Interscience Inc., New York. Velovykin, G. P. : 1970, Nature 225, 254. Wickramasinghe, N. C., Donn, B. D., and Stecher, T. P. : 1966, Astrophys. J. 146, 590. Wickramasinghe, N. C. : 1967, Interstellar Grains, Chapman and Hall Ltd., London. Wickramasinghe, N. C. : 1969, Nature 224, 656. Wickramasinghe, N. C. : 1970a, Nature 225, 145. Wickramasinghe, N. C. : 1970b, Nature 227, 265. Wickramasinghe, N. C. : 1970c, Nature 227, 587. Wickramasinghe, N. C. : 1970d, Nature 228, 540. Wickramasinghe, N. C. : 1970e, Nature 228, 544. Wickramasinghe, N. C. : 1971, New Scientist 50, 694. Wickramasinghe, N. C. and Nandy, K. : 1970a, Monthly Notices Roy. Astron. Soc. 153, 205. Wickramasinghe, N. C. and Nandy, K. : 1970b, Nature 227, 51. Wickramasinghe, N. C.: 1972, Monthly Notices Roy. Astron. Soe. 159, 269. Webster, A. S. : 1970, Nature 228, p. 44. Wesson, P. S. : 1973, Astrophys. Space Sei. 23, 227.

a synthesis of our present knowledge of interstellar dust

Documents