You are on page 1of 29

Planet. Space Sci. 1971, Vol. 19, pp. 1229 to 1257. Pergamon Press.

Printed in Northern Ireland

PHYSICO-CHEMICAL PHENOMENA IN COMETS-III


THE CONTINUUM OF COMET BURNHAM (1960 II)
A. H. DELSEMME and D. C. MILLER
Ritter Astrophysical Research Center, The University of Toledo, Toledo, Ohio 43606, U.S.A.

(Received in final form 30 April 1911)

Abstract-The new model of the cometary head proposed in papers I and II is developed and
applied to comet Burnham. It takes into account the likely existence of a halo of large icy
particles surrounding the nucleus. These particles are steadily stripped from the nucleus by
evaporating gases. Their terminal velocity and their rate of evaporation set the size of the halo.
The existence of the icy halo influences in two ways the photometric characteristics of the
coma. This paper establishes the photometric shape of the continuum as reflected by the icy
grains, and compares it to the observed continuum of comet Burnham. Paper IV will compare
the predictions of the model with the photometric profile of the molecular emission bands of
C2, in the same comet.

1. INTRODUCTION
Two former papers, referred to herafter as paper I (Delsemme and Wenger, 1970) and
paper II (Delsemme and Miller, 1970), have established the terms of reference for a new
model of the cometary nucleus and coma. It is still based on Whipple’s (1950) icy con-
glomerate model, but it takes into account the likely existence of a halo of rather large icy
particles surrounding the nucleus.
In paper I, a laboratory study has shown that, as water is likely to be abundant in the
cometary nucleus, snows of clathrate-hydrates of gases will form with a peculiar structure
showing large grains with a narrow size distribution in the 0.1-l mm range. It has also
shown that, when simulating cometary conditions, these grains are constantly stripped
from the main body of snow by evaporating gases.
Paper II discusses how to handle the simultaneous presence of gas hydrates and of
sorption phenomena in the snows of the cometary nucleus. The regulating factor of gas
production in incoming comets is shown to be the sublimation rate of the icy lattice of the
clathrates, which is practically the same as the sublimation rate of water ice. The other
gases are liberated in proportion as the cavities of the clathrates open by evaporation. This
process works also in the evaporation of the icy grains. Their velocities and their rates of
evaporation set the size of an icy halo surrounding the nucleus. As the icy grains evaporate,
they release molecules or molecular fragments within the icy halo; this extended source
modifies the photometric profiles of the molecular emissions within the inner coma. The
observed lifetimes in the inner coma, which have been attributed so far to hypothetical
precursors, could be influenced or even totally explained by the lifetimes of icy grains.
The general aim of the present paper is to explore quantitatively the implications of
such a model on the observational parameters, specifically, on the photometric profiles
of the continuum and of the molecular bands.
Our first task is therefore to establish the space distribution of the icy grains, at a steady
state, from their equation of motion and their evaporation rate. From this, we will predict
the photometric profile of the continuum by integrating, along the line of sight, the cross-
sectional areas of the grains evaporating in the icy halo. We will compare the predictions
with the continuum observed in comet Burnham. In a forthcoming paper IV, we will
predict the photometric profile of the molecular bands from a similar integration of the
1229
1230 A, H. DELSEMME and D. C. MILLER

distribution of the emitting radicals as they are liberated within the extended source provided
by the icy halo, and we will compare with the photometric profile of Ca in comet Burnham.

2. PRESENCE OF WATER
Since the publication of papers I and II, the arguments presented on the circumstantial
evidence for the presence of the water molecule in the cometary nucleus have been much
reinforced by the discovery of the Lyman a halo of comets 1969 g and i.
Five primary processes are energetically possible for the photo-dissociation of the
water molecule by the solar ultraviolet. They have been listed in Table 1 in decreasing

TABLE1
Photodissociation of water by the solar ultraviolet
A. In the first continuum (1800-1400 A)
(1) H,O + hv --+H(9) + OH(X”H)
(2) H,O + hv --, Hz + O(l@
B. In the second continuum (1400-1150 A)
(3) Ha0 + hv --c H(*S) + OH(&C+)
(4) Ha0 + hv --, 2H(25Z)+ O(‘P)
(5) Ha0 + hv --, 2H(Q’) + O(lB)

order of their wavelength cutoff. An important result (Ung and Back, 1964) definitely rules
out process (2) and therefore confirms process (1) only in the first continuum of water.
Experiments in flash photolysis (Venugopalan and Jones, 1968) suggest that several proces-
ses compete in the second continuum of water. This fact is probably linked with the well-
known existence of a structure of diffuse bands (predissociation phenomena) superimposed
on the second continuum. The relative production estimates of the competing processes
cannot yet be unambiguously established for cometary conditions. However, the energy
that can be absorbed in the solar spectrum by the first continuum of water is about ten
times larger than by the second continuum. Process (1) is therefore responsible for the
photodissociation of more than 90 per cent of the water molecules while less than 10 per
cent could be explained by processes (3), (4) and (5).
Both the hydrogen and hydroxyl halos seem therefore explained mainly by reaction (l),
producing OH and H in their ground states, with subsequent excitation of these molecular
fragments by fluorescence. The further photodissociation of OH could, in this case,
probably explain the intensity as well as the profile of the observed red line of [01], because
this dissociation can energetically lead to H(2S) + O(lD) for all wavelengths shorter than
195OA, as can be verified from the dissociation energy of OH. Some contribution from the
dissociation of CO,, if any, is however not excluded.
The brightnesses of OH, and [01] seem consistent with a production rate of water vapor
of the order of lOa mol/sec (Biermann and Trefftz, 1964). It will be seen hereafter that
this rate corresponds fairly we11 with what can be vaporized by the sublimation of water
or of clathrate snows from an average nucleus.

3. RANGE OF VAPORXZATION RATES OF VOLATILES


The vaporization rate 2, in mol cm-2 set-l, of cometary snows is a parameter
playing a major role in the model. Its range must therefore be carefully assessed. To
compare different volatiles, we will also define Z. as the vaporization rate induced by an
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1231

energy flux equal to the solar constant FO on a perfectly absorbing snow normal to the
solar flux. 2, acts as an upper limit, as usually the snow albedo is not zero and its surface
is not normal to the solar flux. If the latent heat of vaporization of the snows per molecule
L were temperature independent, 2, would be a constant whatever the temperature of the
snows, at least in the lower temperature range where we can neglect the back radiation from
the snows to space; namely:
Fo = Z,L (1)

and the vaporization rate Z,, would be independent of any steady state equilibrium. If we
introduce now the back radiation by Stefan’s law, Equation (2) defines the temperature of
the steady state:
F. = Z,,L + aT4 (2)
where u is Stefan’s law constant.
Many workers have used Equation (2) with a constant L; in particular Weigert (1958),
Huebner (1965) and Huebner and Weigert (1966). But for most of the volatiles, L varies
with the temperature, as can be seen by a few examples collected in Table 2. A notable
exception is water ice, as well as the clathrate hydrates of gases, whose latent heats hardly
vary between 125” and 273°K; (see Table 2). For most of the other volatiles, Z,, must be

TABLE 2. VARIATION OF THE LATENT HEAT OF VAPORIZATION L,,WITH TEMPERATURE, IN calmole-' (L,,IS
CORRECTED FOR VACUUM)

Helium") Hydrogen(l) Nitrogen@) Methanecal

T”K LO LO T LO T LO
3 22.6 2: 212 40 1610 40 2270
4 20.9 30 142 60 1510 60 2234
5 11.5 32 90 80 1153 80 2170
6 0 33 11 100 886 100 1829
34 0 120 403 191 0

Solid hydrate
Ammonia”’ Formaldehyde(*) Water icf? of methane@’

T LO T LO T LO T
100 7309 181 5930 150 11,700 150 llL&
150 7210 200 5736 200 11,480 200 11:200
195 7119 252 5128 250 11,220 250 9950
240 5580 300 10,980 300 9420

(1) Johnson (1960).


(2) Ziegler et al. (1962) and (1963).
(3) Overstreet et al. (1937).
(4) Walker (1964).
(5) Washburn (1928).
(6) Computed from dissociation of hydrate and vaporization water ice, per mole gaseous mixture.

computed from Equation (2) by introducting L = L(T) into it, and assessing first the
steady state temperature of vaporization. In the same way, to compute Z, the temperature
of the steady state must first be found by numerically solving the energy balance equation
per unit surface area:

Fdl - A,)rH-2 cos 8 = ~(1 - A,)T4 + Z(T)L(T) (3)


1232 A. H. DELSEMME and D. C. MILLER

where A, is the nuclear albedo for the solar flux, A, is the nuclear albedo for the back
radiation to space (range of wavelength from 15 to 30 ,u), rH is the heliocentric distance, 8
is the angle between the direction of the solar radiation and the normal to the nuclear
surface, Z(T) is the flux in number of vaporizing molecules cm-2 set-l. But the flux of
vaporizing molecules Z/cm2 set can be computed from the saturated pressure of the gas
that would be at equilibrium with the vaporizing snows:
p = nkT (4)

where n is the number density of molecules per cm 3. The kinetic theory model of condensa-
tion implies that all the molecules Z- that collide with the snow surface will indeed condense,
while the steady state implies that the condensing flux at equilibrium Z- (equil.) equals the
vaporizing flux Z+ (equil.). Now, if we substitute vacuum to the gas pressure, the condens-
ing flux drops to zero, but the vaporizing flux does not change. This kinetic theory model
was experimentally verified by Knudsen (1910) and Langmuir (19 13, 1917). Z in vacuum
can therefore be predicted from Z- (equil.) that is known in terms of collision number of
gas molecules :
Z (vacuum) = Z- (equilibrium) = &z! (5)

where n is defined by (4) at equilibrium and r?is the mean speed of the molecules. In par-
ticular, for a Maxwellian distribution at temperature T, one has:

- . (6)

Using (4) for n and (6) for 8, the mass flow rate mZ can now be expressed by (5) as a
function of p:
m l/2
mZ=p (7)
( 27TkT1 *

In cometary theory, this kinetic model was first discussed and used by Delsemme and
Swings (1952). The same model was used later by Weigert (1959), Watson et al. (1961),
Huebner (1965) and Delsemme (1966). It is important to notice that in this model, the sur-
face of the snows is defined as the macroscopic surface. All the microscopic irregularities
that are smaller than a mean free path must be discounted because they reabsorb their own
vaporizing molecules. This remark is very important for comets because the mean free
path in the gaseous phase may be very large. In the energy balance Equation (3), Z can now
be expressed by (7) in terms of the pressure p at equilibrium. Some tables give p as well as
the latent heat L, as a function of the temperature (see International Critical Tables,
National Bureau of Standards Tables, or the references given in Table 2). The latent heat
values L have usually to be corrected for vacuum, by subtracting the gas expansion work
RT from the equilibrium value of L.
The numerical solution of (3) gives the steady-state-temperature as well as Z(T). The
results for Z,, are given in Table 3 for a few typical snows. If one does not consider materials
less volatile than water, the global range of Z,, goes from 101*to 1021mol cm-2 see-l. On
the other hand, if one limits the model to nuclear temperatures higher than the critical
points of helium and hydrogen (6°K and 33°K respectively) the typical range for Z, goes
from 1018(water snows and clathrate hydrates of gases) to 10ls (free methane snows, and
nitrogen). Figure 1 gives a few numerical solutions of (3) as a function of the heliocentric
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1233

TABLE 3

2, in 10ls
mol cm-2 se& $!) (2)

Helium ,lOOO <2 t2


Hydrogen 106 7,5 6.5
Nitrogen 14.3 40 35
Methane 10.6 55 50
Formaldehyde 5.0 * *
Ammonia 3.1 * *
Carbon dioxide 3.5 121 107
Gases from hydrous ammonia 2.7 213 193
Gases from clathrate hydrate of methane 1.9 214 194
Water 1.7 215 195

2, is the vaporization rate reached at the subsolar point of a perfectly absorbing


nucleus at 1 au. from the Sun.
To is the steady state temperature ofZ, (subsolar point, non rotating nucleus).
TI is the steady state temperature of (ZJ4) (average temperature of a rotating
nucleus).
* Temperatures not computed because L(T) not accurately known.

I I ! I I I I\, I
-06 -04 -02 0 02 04 06 06 IO

log heliocentric distance, A.U.

FIG. 1. VAPORIZATIONRATEZ, mol cm-2 se+, FOR VARIOUS SNOWS AS A FUNCTION OF


HELIOCENTRIC DISTANCE, IN a.u., COMPUTED FOR THE STEADY STATE TEMPERATURE OF A ROTATING
COMETARY NUCLEUS WITH AN ALBEDO A = 0.1.

distance. From 3 to O-1au., 2 varies from IO” to 1020for a rotating methane or nitrogen
sphere, and from 1014to 10ls for a rotating water ice sphere. We will use these two examples
hereafter as two likely limits for snows.
4. ROLE OF GRAVITY IN LIMITING THE SIZE OF GRAINS LEAVING THE NUCLEUS
The importance of this particular problem is linked with the large grains that are to be
considered in our model. Weigert (1959) has already discussed it for smaller dust grains.
3
1234 A. H. DELSEMME and D. C. MILLER

However, his momentum transfer model assimilates the grain to a flat disk. Appendix 1
describes a more realistic model for the momentum transfer, that is going to be used here-
after. For a grain just leaving the nucleus, the gravitational force F, is:
FQ = GAcM~R-~ (8)
with G gravitational constant, M mass of the nucleus, Mo mass of the grain, R radius of the
nucleus.
For a spherical grain resecting the gas molecules diffusely in all directions (Ap~ndix 1)
the molecular drag force F8 is
F, = mv,Z?~a,~ (9)
where m is the molecular mass, v, the mean efflux velocity of the gas molecules, Z the
vaporization rate of the nucleus and a, the initial radius of the spherical icy grain.
The ratio K of the two forces (8) and (9) is:

(10)

where d and da are the densities of the nucleus and of the grain respectively.
The limiting size a, below which a grain can be accelerated away from the nucleus is
set by K = 1, that is:
9 mV,Z
a,=--. (11)
16rr GRdd,
For any other grain size a,, the ratio of the two forces is:

~=a,. (12)
a,
For a numerical estimate, let us take both densities d = d, = 1 g cm-3, an efflux velocity
trO= O-3km/set (Section 6), and the molecular mass of water for the vaporizing gases.
One has:

a, w 2.4 x lo-l2 % . (13)

For a 2 km nuclear radius, the results can be summarized in Table 4. One sees that large
chunks of snows could be taken away by the largest vaporization rates. Of course, this
does not imply that there is a physical mechanism producing chunks that large. At this
point it is therefore tempting to try to visualize whether such a fragmen~tion process can
actually exist. For instance, the vaporization of glaciers on Earth leaves pointed masses
and pinnacles of ice standing between the crevasses. These seracs break down steadily into
very large chunks of ice, by a mechanism that is well understood: crevasses are first opened
by stresses accompanying the flow of the glacier over an uneven underlying surface, after
which evaporation is driven mainly by air convection. The same mechanism cannot act in a
cometary environment. With no sizeable gravity and no sizeable atmosphere, a possible
cause for major fragmentation is meteoritic impact. Another possible cause for minor
surface fragmentation is thermal stress. Whatever their origin, crevasses would usually be
in darkness. Colder than the surface, their walls would condense surrounding vapors as
frost. They would tend to fill up, forming a vein of unusually high albedo. On the other
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1235

hand, major meteoritic impacts could shatter the nucleus into several icebergs, and could
explain the relatively large frequency of comets breaking into two or more constituents.
Of course, if these icebergs do not separate soon enough with sizeable velocities for tra-
jectories to follow distinguishable orbits, they will again collapse against each other until
vacuum welding fuses the reconstituted nucleus into one piece with deep crevasses. This
trend of using ices as a glue to improve the secular structural solidity, depicts a major
property of the icy conglomerate model against fragmentation.
Our present discussion does not lead unambiguously to any physical m~hanism leading
to a fragmentation into large chunks of snow in the range of the meter size. If such a
mechanism does exist, a nuclear region may develop, one or two orders of magnitude
larger than the assumed nucleus, where many very large separated chunks of snow are
alternately shaken away by evaporation outbursts but fall back and are kept together by
gravity until they fragment below the maximum permissible size. This mechanism would
imply that the fragments that leave the chaotic nuclear region are all just below the maxi-
mum permissible size a,. If they were much smaller, they would have indeed left the
nuclear region much before. This process leading to a sharp distribution of sizes near a,,
has already been emphasized in another context (very small grains) by Huebner and Weigert
(1966). Now, if ana is not very much larger than the size distribution observed in paper I,
the following sequence of events is unique: icy grains of the proper size escape into the
coma. However, if ana is much larger than the observed size distribution, it would be
simple-minded to conclude that they are going to fragment again and again very quickly,
because of our laboratory results. It must be kept in mind that grains several orders of
magnitude larger could not have been observed in our experiments, for the simple reason
that the main body of snow was limited to a size of a few centimeters only, in a one or two
liter vacuum chamber (see pictures in paper I). During the short duration of the accelera-
tion period (see hereafter), two possible processes are therefore possible and should be
distinguished. Either the big chunks of snows will fragment again and again, or they will
not. In the first hypothesis, the process is likely to stop only when the size distribution
observed in paper I is reached. In the second hypothesis, the big chunks may loose small
grains only, stripped by their own evaporation, as they act in turn as the main body of snow
as in our experiments. In this last instance, part of the grains of the halo will reach the
size distribution described in paper I, whiie the other part will remain in chunks of the same
order of magnitude as the maximum size a,,,. This process of fragmentation from a few large
chunks to many smaller grains makes it unlikely that a model of optically deep uniform
layers could easily describe the nuclear region. Rather, the shadow of big chunks would
slow down momentarily the vaporization at places. The model of a small spherical nucleus
will still be used hereafter, understanding that it only defines the cross sectional area for
vaporization or photometry.
The last columns of Table 4 describe the importance of gravity for a particle of the
millimeter size. Two cases are neatly separated: For Z/u0 larger than lOl*, the drag forces
are overwhelming and gravity can be neglected. For Z/a0 smaller than lOI’, gravity is an
important factor that will control the size of the particles in the icy halo. Figure 1 shows
already that for water, gravity will control the appearance and the size of the halo near 3
or 4 a.u. Another conclusion is of importance: when any evaporating material is present,
the relative speeds of particles invoked by Lyttleton (1968) to explain the coma and tail
phenomena are entirely negligible in comparison to the magnitude of the momentum
transferred by the gas drag forces at short heliocentric distances.
1236 A. H. DELSEMME and D. C. MILLER

TABLE 4. MAXIMUM RADIUS a, OF THE ICY GRAINS DRAGGED AWAY BY THE VAPORIZATION
(NUCLEAR RADIUS 2km)

ra11) value of K for different a,


(a.u.) 1ogZ’Q alll 10 /” lmm 10m

3.46 14 12P 0.8 >l >1


2.16 17 1.2 cm 8 x lo-’ 8 x 1OV >l
0.14 20 12m 8 x lo-’ 8 x 1O-6 0.8

(1) Typical heliocentric distance with K = 5 (see Section 6).


(2) Rotating nucleus with A = 0.1 at distance r,.

5. NUMBER AND MASS OF ICY GRAINS IN THE HALO


Two ways are open to assess the possible number of grains in the halo: the laboratory
experiments (paper I) and the cometary observations.
In the laboratory, the production rate observed was only assessed visually. This gives
an order of magnitude of one grain cm-2 sec- r. The average grain’s diameter is 0.6 mm
(Fig. 2A, paper I). With an apparent density of 0.44 g cm-3, its average mass is about
5 x 1O-5g, giving a mass production rate of grains of 5 x lop5 g cm-2 se&, as compared
with a mass production rate of water vapor + methane of 2.65 x 101s mol cm-2 se+
= 8 x 10e5 g cm-2 sec- l. The ratio of the grain and gas production rates is therefore
m = 0.63, whose order of magnitude only should be meaningful.
The assessment from cometary observations relies on the comparative brightness of the
central condensation to the global brightness of the continuum. In many comets showing
a continuum, and in particular in new quasi-parabolic comets which usually show a strong
continuum, the light distribution is often stronger in the continuum than in the molecular
emissions. On the other hand, E. Roemer’s determinations (1969) show that in most cases,
the brightness of the head is, on the average, at least five magnitudes, that is 100 times
brighter, than the central condensation. Assuming that the albedo of the grains is the same
as the albedo of the nucleus, the total cross sectional area of the visible grains is at least 100
times larger than the cross sectional area of the central condensation, which might still be
slightly brighter than the ‘true nucleus’. This conclusion is independent of any phase
function, as it would be the same for the nucleus and the grains. If we assume a spherical
nucleus of 2 km radius, its cross sectional area is 10tl’l cm2. This means a cross sectional
area of at least lOI3 r cm2 for the visible grains. The initial mean diameter of the grain being
0.6 mm, as the evaporation proceeds linearly, the average cross-section per grain is 10-2’g
cm2 and there are 1016gr visible in the coma. This result implies that the number density
of the grains is small enough to see almost all of them. Their lifetime being 104.5set (Section
9, formula (25), with 2 = 1017.2),there is a grain production of 1O*1’5 gr/sec-l, or a pro-
duction rate of 0.67 gr/cm-2 sec- I. Their initial mass being low4 3 g, the mass flow rate of
the grains is about 1O7’2g set-I, or a mass flux of ti, = 10-4.5 g cm-2 set-l. Therefore
m = 0.43.
This could be a lower limit, as it is based on the assumption that the central condensation
brightness is the brightness of the true nucleus. It could also be an upper limit as the fine
dust component might also contribute a sizeable fraction of the continuum light even if its
mass component remains negligible, because of the small size of a large number of grains,
therefore diminishing the contribution of the large icy grains. The fact that our two assess-
ments point to the same order of magnitude shows that there is at least no inconsistency.
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1237

Therefore we will accept that the mass flow rate of icy grains is of the same order of magni-
tude, although perhaps somewhat smaller or larger than the mass flow rate of the gas.

6. DYNAMICS OF ICY GRAINS


Each grain stripped from the nucleus by the evaporating gases is assumed to start at
rest and to be accelerated by the gas drag forces to a terminal velocity vT reached in a time
t,. The nucleus shall be considered as an isotropic source of gas, streaming radially with the
velocity v,, defined by the mean radial efflux velocity. The effect of icy grain evaporation
can be neglected in determining its acceleration, since the gas-grain interaction is completed
before the grain’s mass is much reduced by evaporation. The addition of energy by the
solar flux as well as the asymmetry introduced by the solar wind are also neglected in this
first approximation.
Two approaches can be followed to compute the momentum transfer, Weigert’s (1959)
and Markovich’s (1963). Weigert pioneered a simple model for the momentum transfer
from the gas to the grains. In particular, Weigert used implicitly a flat disk model for the
grain, and assumed v,, = $0, d being the mean speed as defined in kinetic theory. Huebner
and Weigert (1966) improved the preliminary model by using half its momentum transfer.
It is proved in Appendix 1 that this assumption is justified for a spherical grain. They also
assume without justification that v,, = 45. Weigert’s value of the initial mean radial velocity,
namely $6, is obtained when one considers the efflusive flow of gas into vacuum from a
small hole in a pressurized vessel. In this case, the flux polar diagram would follow Lam-
bert’s law. However, for vaporization from a solid, the escaping molecules must overcome
the surface force field. Their mean velocity is correspondingly reduced and the flux polar
diagram becomes an oblate spheroid with its minor axis normal to the surface (Kennard,
1938). In this case, the mean radial efflux velocity is $ti, that is Huebner and Weigert’s
value. This is true only for a sublimating solid whose surface is a perfect plane with no
pores down to the molecular level. Surface irregularities larger than molecular size random-
ize again the average polar diagram of the vaporization velocities into Lambert’s law
(Lyubitov, 1967). All that can then be said about the mean radial efflux velocity is that,
depending on the surface porosity, it lies between a maximum and a minimum value:

The minimum value corresponds to the unrealistic case of a solid whose surface is a perfect
plane without any pores down to the molecular level; the maximum value corresponds to
very deep and narrow pores oriented at random, that simulate effusion through holes
without any force field. The value u,, = 0.66 seems a sensible compromise for actual snows.
This discussion was needed because the initial mean efllux velocity v0 plays an important
role in the gravity cutoff (Section 4) as well as in the grain’s terminal velocity (Section 7).
It is worthwhile mentioning that the flux remains anfi whatever the efflux velocity vO.
The foregoing considerations do not take into account the results of any molecule-
molecule interaction farther away from the nucleus. Huebner and Weigert’s approach is
however satisfactory for their particular problem, namely the study of the halos of comet
Schwassmann-Wachmann I (1925 II), because at 6 a.u. from the Sun there is pure effusion
in vacuum. On the other hand, at shorter heliocentric distances, the molecular collisions
are frequent enough to transmit momentum and energy. As the gas expands towards
vacuum, molecular collisions quickly transform part of the random kinetic energy of the
molecules into a radial stream. The speed of sound could therefore represent already a
1238 A. H. DELSEMME and D. C. MILLER

better approximation than a/2, in the vicinity of the surface; it is

= 0.71 rs

for water vapor. But the gas goes on expanding adiabatically (if we neglect the solar
energy absorbed) and the work of expansion is converted into more kinetic energy of the
radial stream. A better approach is therefore to consider this isentropic expansion as a
fluid dynamics problem. For an expansion to infinite volume, the limiting gas velocity D,
is given by the enthalpy content of the gas:
!r
$mvo32= C, dT
s0
T being the surface temperature of the nucleus. The true terminal velocity of the gas is of
course lower than this, but it can be computed easily as a fraction of v,. This terminal
velocity is reached when molecular interactions become negligible and when pure effusion in
vacuum starts. This happens when the mean free path I of the gas molecules is of the order
of the distance r to the nucleus, which occurs for water vapor at:
rl = 4.5 x 10-r’ ZR2. 04)
For a nucleus of R = 2 km, pure effusion is reached at IO5km from the nucleus at 1 a.u.,
but at 10 km only at 3 a.u., while for Schwassman-Wachmann I (6 a.u.) it is reached
already at the surface of the nucleus even if it is very large. As the terminal velocity of the
grain is usually reached at some distance from the nucleus (lo2 km) it is obvious that the
velocity increases of the gas cannot be neglected. The only valid approach for most of
the comets is therefore the fluid dynamics approach. Such an approach has been used
first by Markovich (1963) and Schulman (1967) in the USSR, and more recently by Prob-
stein in the U.S.A. (Probstein, 1968; Finson and Probstein, 1968) for a rather similar
problem dealing with the acceleration of small dust particles by those drag forces due to the
expanding gas in the cometary head, and the isophotes of dust tails were successfully
predicted. However, if the simple effusion model gives too small gas velocities, the fluid
dynamics model is likely to give too large gas velocities. Indeed, the enthafpy, which gives
the total amount of work available for the adiabatic expansion is obtained by integrating the
heat capacity at constant pressure, over the temperature. The heat capacity is the sum of 3
terms :
C, = +R + C (rotation) + C (vibration).

These three terms are actualfy available at the onset of the expansion process. But as the
expansion proceeds, the gas cools down and the rotational and vibrational energies have to
be transferred by equipartition into the translational energy, to be useful for the expansion
work. It is very difficult to deal correctly with this transition from isentropic expansion to
effusion in vacuum. A semiempirical way to treat the problem is to substitute for v,,
corresponding to the total enthalpy of the gas, a smaller v, corresponding to that part of
enthalpy that will be available for collisions. From the natural lifetimes for vibrational
states (Goody, 1964) and the mean number of collisions before vibrational deactivation
occurs (Massey and Burhop, 1952), it is clear that only the pure rotational energy of the
molecule will be transferred into translational energy. Electronic states and vibrational
states will be essentially adiabatic as far as collisional transfer of energy is concerned. For
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1239

water vapor for instance, C, = 8.67 Cal/mole = 8.00 Cal/mole from translations and
rotations +0.67 Cal/mole from vibrations. This leaves only some 8.00 Cal/mole available
for expansion work. The velocity corresponding to the full enthalpy content of water is
0, = 1.848. The velocity corresponding to 8.00 Cal/mole is v, = l-772? and the corre-
sponding y = 1.33. These latter values will be used hereafter for the limiting velocity of
the water molecule expanding into an infinite volume. This result does not practically
change if polyatomic molecules like methane or ammonia are mixed with the water vapor.

7. FLUID DYNAMICS APPROACH TO THE TERMINAL VELOCITY


For smaller grains, Probstein (1968) shows that the terminal velocity is reached in
times of the order of lo2 set, and he understandably neglects gravity. We have checked
(Fig. 2) by a few numerical integrations that the terminal velocity is reached for times of

log time set

FIG.~. ICY GRAIN VELOCITIESASAFUNCTIONOFTHETIMEELAPSEDSINCETHEY WERESTRIPPED


FROM THE NUCLEUS.
A simple momentum exchangemodel and R = 1 km was used with a view of assessing the
order of magnitude of the time needed to reach the terminal velocity.

the order of lo3 to lo4 set for very large grains (and for distances of the order of ten nuclear
radii). The speed and distance of free fall of small grains in the gravity field of the comet
for durations of this order is still negligible and hence gravity will not be considered within
this section.
In Probstein’s approach, the terminal velocity V, is expressed as a function of two
parameters p and m. The larger p, the longer the grain takes to respond to the gas drag.
In our notations, p is defined as (Probstein, 1968, p. 576, formula 4.6).

where da is the grain’s density, and where Probstein’s gas flow rate has been substituted by
1240 A. H. DELSEMME and D. C. MKLER

4=R2mZ. The parameter ttt is defined by the ratio of the mass flow rates of the grains and
of the gas:

?tt=:.
%I?
nz
(17)

Even though we consider evaporating grains, this ratio does not vary very much during the
gas-graininteraction. Figure 3 expresses the terminal velocity nT of the grains as a function

-0

FIG. 3. ICY GRAIN TERMINAL VELOCITY Y,/Va AS A FUNCTIONOF ,!?FOR FOUR VALUES OF m.
V, is the limiting velocity derived in Section 6 from the available part of the enthalpy of the
gas.

of /?, for various values of m. The terminal velocity is introduced here in the form of a
dimensionless ratio V&J,, where v, corresponds to the available enthalpy of the gas. For
water vapor, y = l-30 which gives v, = 1.84~ where fi is the mean speed; however we
have established in the previous section that v, = 1.77~ and y = I-33 was a better assess-
ment because the vibrational energy is not used for expansion work. For Ca, = 1 g cm-8,
R = 2 x lo5 cm, m = 3 x 1O-23g and Z”, = 200”K, ,8 becomes:

/CI= 2,7 x lo=2 . w


To determine a terminal velocity, a value of m is also needed. Figure 3 shows that large
variations in m do not change the terminal velocity very much. m was established tentatively
in Section 5. (m = 0.63 from experiments, m = O-43 from observations). To be specific,
the curve m = 0.4 will be used. For the vaporization rates discussed before, the terminal
velocities are given in Table 5, and are illustrated in Fig. 4 as a function of a,JZ. Notice in
Fig. 4 the sharp gravitational cutoff introduced by a,. The discontinuity in the derivative of
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1241

log Z/a, in cm-3sec-1

FIG. 4. TERMINAL VIXOCI~ V, AS A FUNCTION OF Z/a,.


The solid lines neglect gravitation down to the gravitational cutoff for R = 2 and 5 km. The
dashed lines V,, take gravitation into account. The small dotted lines numbered 1 and 10
refer to the position of the gravitational cutoff for a nuclear radius of 1 and 10 km respectively.
m = 0.4.

TABLE 5. TERMINALVELOCITIESOFICY GRAINSIN m


set-l (R = 2 km, Vm = 860 m set-‘. THE GRAWY
CORRECTION HAS BEEN APPLIED FOR log.?+, < 18

log a, cm
1ogz -3 -2 -1 0 1
14 0 0 0 0 0
15 10 0 0 0 0
16 41 10 0 0 0
17 127 41 10 0 0
18 332 127 41 10 0
610 332 127 41 10
736 610 332 127 41

vT comes from the neglect of gravitation in this section. The two solid lines in Fig. 4
correspond to 2 and 5 km for the radius of the nucleus.

8. TERMINAL VELOCITY WITH GRAVITY CORRECTION


Gravity can be neglected for those particles whose size a, is much smaller than a,
(Section 4). For the other grains, that is near the gravitational cutoff of Fig. 4, a correcting
term can be computed from the equation of motion, rather than to introduce gravity for a
full numerical integration of the fluid dynamics equations.
The equation of motion of the grain is:
MQi - P + GMGMre2 = 0 (19)
where Ma is the mass of the grain, M the mass of the comet, GM,Mr-2 the cometary
1242 A. H. DELSEMME and D. C. MILLER

gravity and P the momentum rate transferred by the gas drag. This equation represents
Huebner and Weigert’s (1966) approach if we use a constant gas velocity v,, in which case P
is given by:
z R2
P = m(v, - i)2 mzo2- - ,
i)
v. r
By introducing a, from (1 I), the equation of motion can now be written:

(21)

But v. is not a constant because of the adiabatic expansion of the gas. Let us substitute
for v. another constant v1 defined as that mean speed of the gas needed to give the correct
terminal velocity of Section 7 to those particles for which gravity can be neglected. When
the terminal velocity i = vTB is reached, i = 0 and vFci becomes

vTn=vl[l
- (y2]. G-9

First, this defines vr, because when gravity can be neglected, a, < a, and vr equals that ttT
which is given by fluid dynamics. Finally the brackets in (22) give a convenient correcting
term for those particles whose size is near a,, as well as the gravitational cutoff required by
Section 4. It must be noticed that, v. is substituted by v1 in the equation of motion, but has
to be kept in the definition of a,, as it is the momentum transferred by the initial gas velocity
near the nucleus, which has to baIance gravity to lift the grains from the nucleus. This can
be explicitIy stated in (22) by substituting a, by its value (11):

(23)

or, with the numerical value already used:

i )i .
a,R m
vTB = vT 1 - 6.44 x lo5 z (24)

This gravity correction has been introduced with R = 2 km and 5 km by the dashed lines
on Fig. 4, which smoothes out the sharp gravity cutoff introduced by considering Section 4
only.
9. LIFETXME OF THE ICY GRAXNS
The grain’s radius CIshrinks by vaporizing at a rate 4 = ~2~~~~. The lifetime of a
grain is therefore:
da ao
q-=:---_. (25)
m 2~3
The vaporization rate of the grain 2, must now be assessed to use (25). In general, Zo is
smaller than A. Three factors make this so:
(a) The vaporizing process is a distillation which implies a fractionation. The vapor-
ization rate of the nucleus 2 is practically set by the most volatile material in the nuclear
mixture. The vaporization rate Z, of the grain is, or becomes very quickly, that of the
residual after the most volatile material has left. For instance, TabIe 3 shows that near 1 a.u.
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1243

for a mixture of methane snows with either water snow or clathrate hydrates, the ratio
of the rates K = Z/Z, is either 6.2 or 5.6, respectively.
(b) The albedo of at least some of the icy grains A, is likely to be larger than the nuclear
albedo A,. On the nucleus, the snow vaporizations must leave behind an irregular ‘crust’
of dark meteoretic material, as in the moraines of our terrestrial glaciers. A low albedo of
the order of O-1 as in Whipple (1950) is therefore not too unlikely. On the contrary, it is
reasonable to expect a whole range of albedos for the departing grains, ranging from say
0.1 to the albedo of the clathrate grains. The albedo A, of the clathrate grains has been
roughly measured by Wenger (1968). It lies between O-85 and O-95. (They looked more like
white snow pellets than transparent hail grains.) The subsequent discussion shows clearly
that the size of the halo will be controlled by the grains of the highest albedos.
(c) At a given heliocentric distance, different regimes of thermal equilibrium may
operate for the nucleus and for the grains. For instance, at 3 au., a water grain is stil1 in
radiative equilibrium while the nuclear mixture would still mainly be in vaporization
equilibrium. This explains why the gap between the two Z curves for H,O and CH, becomes
very large at large heliocentric distances in Fig. 1; at 3 a.u., Z - 1000 Z,. In this case,
the phenomenon is of an entirely different nature. The water or clathrate grains, which are
much smaller (in the range of l-10 D) have lifetimes which become so large that the differ-
ences in the heliocentric motions of the launched grains and of the nucleus become the
factor that control the size of the halo. The implications of this interesting case will be
pursued in a later paper.
In the present discussion, we will consider only heliocentric distances smaller than
2 a.u., where pure vaporization predominates and where item (c) may be neglected in a
first approximation.
Taking factors (a) and (b) into account we can write

T=&ffK1 --A,. (27)


mi! l-AA,

As the darker grains evaporate sooner, they disappear closer to the nucleus. On the other
hand, they reflect less light. The size of the halo is therefore set by the largest A,. From
Wenger’s (1968) results, we accept A, = 0.9. To be conservative, we also accept K = 5,
corresponding to a moderate fractionation due to the distillation effect. Thus:

7 = I.5 x 10z4Q_o
set (28)
z *

The lifetime of the largest grain is now obtained by introducing in (27), a0 = a, given by
(11):
1 T
7, = 3.8 x 1OL2- - set (29)
R J- 200 ’

The temperature under the radical sign is the steady state temperature of the outside surface
of the nucleus and comes from v,. The lifetime r, = 2 x 10’ set for R = 2 km and T =
200°K. This means that the lifetime of an icy grain of that maximum permissible size which
can be stripped from the nucleus and sent into the icy halo is of the order of several months,
whatever the heliocentric distance. At smaller heliocentric distances, the larger size of the
chunks that can be stripped away by high vaporization rates is compensated by their own
1244 A. H. DELSEMME and D. C. MILLER

larger vaporization. It is clear that our approximation is no longer satisfactory for short
heliocentric distances. The linear expansion of the halo cannot describe any more the
difference in the heliocentric motions of the particles and the nucleus during several months.
On the other hand, as the theoretical limit a, goes up to enormous sizes with the vaporiza-
tion range, it makes sense to think that a, will be increasingly shifted outside of the natural
size distribution of the grains. From this time on, the maximum lifetime of the grains will
not be determined any more by Q~, but by the largest size a’ < a, that is still present in a
fair amount in the size distribution. From this moment, the halo will shrink with the
shortening of the lifetimes of the grains a’.

10. EXTENT OF THE ICY HALO


At large distances from the nucleus, gravity can be neglected and the grain moves
radially with a constant velocity +o. The extent of the icy halo is then, from (22):

R,=o,[l- &“]7. (30)

Let us now use x = Z/a0 as the independent variable. From formula (13):
112
= &p/2 &+/2
with cm+ se@ (31)

the temperature factor added to b takes account of the nuclear temperature when it is no
longer 200°K. From formula (28):
7 = cx-r with c = le.5 x 1O24cmV3. (32)
The terminal velocity must also be expressed in terms of x. As small values of x will play
the important role, Fig. 3 shows that VTvaries linearly with /3-li2 which is itself proportional
to so/Z from (18). Therefore, V~ can be written:

with cm2 se@.

The range of validity of this relationship is illustrated by the straight part of the slope on
Fig. 4. From (31), (32) and (33), R, can now be expressed in terms of x:
R, = cdR%-1J2 - bdcRx-l (341
R, can be maximized by Druids = 0, which transforms into:
x mao = 4b2R. (3.5)
Substituting x = x,,, into (31) gives the grain size a0 = a, which maximizes the halo:

The icy grains that will reach the largest range before being vaporized are therefore those
whose size is one fourth the limiting size a, of the grains. Because of the sharp distribution
of sizes just under a, that has already been discussed in Section 4, these grains are likely
to be very abundant until a’ < a, is reached. The extent of the halo will be given by
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1245

&f = &ma* b Y introducing (32), (33) and (36) into (30):

T
R, = $ = 1.7 x lo9 (37)
J ZPrn.

If this formula were used for a nucleus containing only a mixture of water ice with methane,
the methane would lower the sublimation temperature of the mixture to a value which

1
T92OO’K

-0 5 05
log hellocenttk distcince(l A U.

FIG. 5. EXTENT IN km OF A FEW ICY HALOS BUILT UP WITH GRAINS OF A SINGLE SIZE, AS A
F~N~~N~F~HE~~cE~RIcDI~AN~E,~.~.

The parameters correspond to grains of clathrate hydrates plus some free methane (K = 5,
A, = 0.9) stripped from a nucleus whose radius is 2 km. 87 per cent of the distribution
observed in the laboratory lies in between the two solid lines (a, = 0.1 and 1 mm); the radius
of grams is given for other hypothetical halos in dotted and dashed lines. The envelope of all
halos for different particle sizes is also shown by a thin solid line near log RE = 4.2 for a nuclear
temperature of 200°K, and log Ra = 3-P for a nuclear temperature of 50°K. Observe the
sharp gravitational cutoff near 3.5 a.u.

depends on the abundance ratio, to be lower than 100°K. For 80°K the radius
but is likely
of the halo given by (37) is 10,700 km. However, the nuclear mixture is still hypothetical.
If its sublimation temperature is much larger than that of methane, the radius of the halo
is accordingly larger. Any heating in the nuclear region, for instance by the chemical
reactions initiated by collisions of highly unstable molecules or radicals, can bring the
same result. A better guess for the nuclear temperature may be 200”K, in which case the
halo envelope reaches 17,000 km.
The halo could also be smaller if our guess is wrong for the nuclear albedo. It must also
be remembered that formula (37) expresses the extent of the envelope of the different halos
containing all the possible grain sizes. This envelope is sketched in Fig. 5. Figure 5 gives
also, as a function of the heliocentric distance, the extent of different individual halos
built by grains of uniform initial size. The two solid lines correspond to particles whose
radius is 0.1 mm and 1 mm respectively. In between these two lines lie 87 per cent of the
particle distribution observed in paper I. The two dashed lines correspond to the particles
that can be used as an upper and lower limit of the likely distribution, for all practical
1246 A. H. DELSEMME and D. C. MILLER

purposes. Finally, the dotted lines show the behavior of very large and very small hypo-
thetical icy grains. The two horizontal lines show the envelopes of the halos for 50°K
and 200°K respectively.
Table 6 expresses the extent of the icy halo in km for different grain sizes and different
vaporization rates.

TABLE 6. EXTENT OF ICY HALO IN km FOR DIFFERENTGRAIN SIZES AND FOR DIFFERENTVAPORIZATIONRATES

(11 Icy grain initial radius a,,@)


r,
(a.u.) log Z’“’ 1OP lOO,u lmm 1 cm 10 cm

3.46 14 5200 0 0 0 0
3.12 15 13,800 5200 0 0 0
2.75 16 5400 13,800 5200 0 0
2.16 17 1800 5400 13,800 5200 0
1.17 18 590 1800 5400 13,800 5200
0.412 19 190 590 1800 5400 13,800
0.141 20 52 190 590 1800 5400

(1) Typical heliocentric distance with x = 5 and hydrate or water grains.


(2) Vaporization rate for a rotating nucleus with A = 0.1 at distance (1).
(3) The extent of the icy halo is nil for the grains whose size n, > a, because cometary gravity is larger
than gas drag forces, and they cannot leave the nucleus. The maximum size of the icy halo is reached with
log Z - log a, = 17.5 cm-S set-’ for R = 2 km.

It is worthwhile mentioning that for a nuclear radius of 2 km, the maximum size of the
icy halo is reached for log Z/q, = 17.5 while the gravitational cutoff lies at 16.9. Most of
the grains with &a,,, < a,, -=ca, move very slowly, as seen in Fig. 4.

11. PHOTOMETRIC PROFILE OF THE CONTINUUM REFLECTED BY THE ICY HALO


As the optical depth effects cannot be found outside a region which is rarely distinguish-
able from the apparent photometric nucleus for average seeing conditions, they are going
to be neglected here. Let us call n, the number density of the grains near the nucleus, n’
their density at distance r. One has:
R2
n’=n,
0 -
r
.

Let us call a, the initial cross sectional area of one grain, and o its cross sectional area
during vaporization. One has:
2

o=a,
( l-f
H 1
with r < R,, R, being the largest extension of the icy halo. Let us call r’ the projection
on the sky of the radius r, and 4 the angle between r and r’. The surface brightness B(8)
of the icy halo, at a distance r’ from the cometary nucleus, is proportional to the cross-
sectional area of all the grains, which is given by integrating all the contributions of the
different parts of the halo along the line of sight:

Boc2
s 0
40
an’r’ sec2 + d$. (41)
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1247

The integration limit is the edge of the halo: & = arc cos r’/RH. if we introduce x =
r’JRHI by using (39) and (40) into ($I), we get:
arc cos x 1 -b (I - x‘qria
- 2log f (1 - $)I/2 (42)
lzcc x x
x is used as a dimensionless parameter, expressing the projected distance on the sky as a
fraction of the radius RR of the halo used as a unit. Table 7 gives log 3 as a function of
log X, as well as the exponent ra in the formula:
B 4: x-”
n expresses the brightness gradient of the continuum, that is the slope of the curve log
B =f(log X} as in Fig. 6 where a scale of brightness as well as a scale of astronomical

I t I
05 0 0,5

Fraction of halo radius

magnitudes illustrate the shape of the continuum reflected by the icy halo. To limit the
central peak, a simulated seeing of x = 0.01 has been adopted, This corresponds, for a
halo radius of lo4 km, to a 0:3 seeing when the comet is at a geocentric distance of O-4au.
ft is clear that the brightness per unit area of the sky falls ofI?very sharply in the hato. This
simulates rather well the photometric shape of the continuum as it is seen on spectra, as
well as the central peak of the photometric nucieus, as it is seen through telescopes.

12. COMPARISON OF THE MODEL WITH COMET BURNHAM


Comet Burnham (1960 11) was selected to assess the validity of the model, because the
photometric profile of its head was studied in several monochromatic bands, including
C, and the continuum, by different authors: O’Dell (I961), F. Miller (1961 and 1967),
O’Dell and Osterbrock (1962) and Malaise (1966 and 1968).
On the ather hand, it has a rather weak continuum, giving the hope that the phenomena
linked with snow particles would not be hidden by a very large amount of light reflected
by fine dust.
1248 A. H. DELSEMME and D. C. MILLER

It is also on comet Burnham that Dossin (1962) observed the rare phenomenon of an
optical depth effect near a cometary nucleus.
Finally, among the six comets studied by Malaise (1970), Burnham is the only one
showing a pure fluorescence spectrum. The absence of collisional excitations suggests that
in the inner coma a large fraction of the gases could still be stored in icy grains.

13. OPTICAL DEPTH EFFECTS NEAR THE NUCLEUS

It was established in Section 5 that there are about 1Or6gr in the coma. This number
is small enough to see almost all of them, and optical depth effects can occur only very
near the nucleus. More specifically, let us consider here the model where the icy grains
move isotropically from a point source at a constant radial speed which is their terminal
velocity. The solar flux F reaching a distance r from the nucleus is dimmed by an amount
expressed by its ratio to the initial solar flux F,,:

E=exp(-n!L!)
FO
where rio is the grains’ flux in number at the surface of the nucleus, uT their terminal velocity
and 0 the geometric cross sectional area of one grain. In this formula, VTplays an important
role. For a given heliocentric distance (constant Z), Fig. 4 gives the distribution of the
terminal velocities as a function of the grain radius. The small grains with large velocities
(18 < log Z/a0 < 23) do not play any role in the optical depth build up. Specifically
when rio = 1 g cm+! se&, UT = lo4 cm set-l and CT= 1O-3cm2 (0.6 mm dia.), one gets
for a nucleus of R = 2 km:
F = 0*98F,,

which shows that, at the nuclear surface, only 2 per cent of the incoming flux is lost.
On the contrary, the large grains are very much influenced by gravitation and have
very small terminal velocities. In particular those grains whose size is near the gravitational
cutoff correspond to the lowest part of the dashed curve marked t+o on Fig. 4. For instance,
for Qa, -c a, < a,, one has 0 < r+o < 2 m se+. Specifically, the same example as
before, but with a velocity VT = lo2 cm set-l, becomes:
F = 0.135Fe

which is no more negligible.


This, however, cannot happen in the exact direction of the Sun, because any blanketing
of the nucleus by a layer of hail grains or snowflakes has a strong feedback action: the
blanketing slows down the vaporization of the nuclear snows, which stops the gas drag on
the grains until the sunshine returns to the nuclear surface. If we add the fact that the
vaporization at the subsolar point is about four times as large as the vaporization averaged
over the whole nucleus, the well known fountain effect acts sunwards by keeping grain
velocities large and therefore stopping any optical depth build up in this particular direction.
This of course does not rule out a dissymetrical distribution of optical depth, which stands
as the only easy explanation of an isolated observation like that of Dossin (1962).
Let us discuss in detail this important observation of comet Burnham. A star dimmed by
1 magnitude at 600 km from the nucleus implies that 60 per cent of the original flux was
subtracted from the beam by its passage through the head’s diameter. Because of the large
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1249

size of the grains and of Babinet’s principle, half of the flux lost was only slightly diffracted,
while half of it was stopped by the geometrical cross-section of the grains. Therefore,
along a radius, the optical depth at 600 km from the nucleus is defined by F/F0= 0435.
To illustrate the possibilities and the limitations of the present model, let us delineate the
values of the parameters able to give such an optical depth. If an albedo of O-1 is accepted
for the nucleus, Roemer’s (1966) photometric data imply a nucleus of 3 km radius for
Burnham, that is a gravitational cutoff for log Z/a0 = 17 (with the usual assumption of
d = 1 g cm-3 for the nucleus, which stands as a compromise between O-3 to 05 g cm-2 for
the snows and 5 g cm+? for the meteoritic material). As 2 = 17.5 for water or clathrate
hydrates evaporating from a rotating nucleus at 1 au., a,,, = 1.6 cm. On the other hand,
the mean terminal velocity for the grains between a, and ia, is about 1 m set-l. Let us
introduce it with d, = Qa, = 1.2 cm and F/F0= 035 at r = 600 km in formula (38), and
let us solve for P&. We get
ti, = 7.5 x 1O-5gr cm-2 see-l

at the surface of the nucleus. Hence, to explain the observations, it must be admitted that
the particle number between a, and $a, is of the order of lo-” the total distribution.
On the other hand, if we assume that the log normal distribution used in paper I to describe
the observations, can be extended to 5 times its standard deviation, we should have only
a fraction of the order of tom6 of the size distribution between am and $a,.
It is of course impossible to go further without knowing the actual size distribution
of the grains in the centimeter range. In particular, the mechanism of concentration of
particles by the gravitational cutoff, which acts like a sieve by keeping the large particles
grounded until they vaporize or break into fragments, could easily provide enough grains
of the proper size.
Other mechanisms than the one proposed here are difficult to conceive. As Huebner
and Weigert (1966) have already shown, no fine grain component could explain an optical
depth so far away from the nucleus. A purely isotropic and homogeneous absorption
layer could not explain it either because it would stop the evaporation. In such a low gravity
field, a convection mechanism to transport the energy flux seems very difficult to justify.
In the present model, an anisotropic coma is easy to explain, with slow grains accumulating
elsewhere than in the Sun’s direction. Finally, the behavior of the small fraction of large
particles that are expected to exist near the gravitational cutoff can easily explain Dossin’s
observations.
14. PHOTOMETRIC PROFILE OF THE CONTINUUM OF COMET BURNHAM
Probstein (1968) has shown that small particles reach their terminal velocity very quickly.
For intermediate and large particles, Fig. 2 establishes that they move more slowly, but that
the gas-grain interaction is finished in any case at a distance of the order of 100 km from the
nucleus. Hence for all practical purposes, the grains move with a constant velocity at all
distances that are resolved by telescopes in the cometary head. Therefore, the number
density of the grains falls off according to a law in r-z for all observable distances.
If the continuum comes from a non evaporating dust expanding isotropically, its profile
is therefore given by a law in T-” where n = 1 at all distances for which the optical depth is
negligible. From now on, n will be called the logarithmic brightness gradient, or in short,
the gradient (from the relationship log B = --nlog r).
When approaching the nucleus, if the optical depth is no longer negligible, then the
4
1250 A. H. DELSEMME and D. C. MILLER

gradient becomes smaller than 1. If isotropy is conserved and if the grains do not evaporate
or decay in some way, it is impossible ever to observe a gradient larger than 1.
The continuum of comet Burnham was observed by O’Dell (1961). Dr. O’Dell kindly
sent us his photometric recordings. A new analysis of his data is represented by the small
circles drawn in Fig. 7. The dimensionless profile of the surface brightness of the continuum
reflected by icy grains (Fig. 6 and Table 7) is the solid line in Fig. 7. As it is expressed as a

-0.4 -0.2

log fraction of halo radius

I I I
41 43 45

log km from nucleus

I T 3’ 4’

Arc mln

FIG. 7. CO~NUUM OF COMET BURNHAM COMPARED WITH MODELS OF ICY GRAINS AND DUST
PARTICLES.

function of the logarithm of the distance from the nucleus, a horizontal translation could be
used to fit it with the observations. It gives the extent of the icy halo:
log R, = 4.65 (in km).
This value is somewhat larger than the results of Section 10, and suggests the existence of
all grain sizes up to the gravitational cut off. In particular, grains of the centimeter size
should exist. The shape fitting seems to be very gratifying, especially if compared with
the slope of an isotropic dust coma (dashed line). The deviation of the upper left part of
the observations from the solid line, as visible in Fig. 7, can be explained. O’Dell’s dia-
phragm was 1 arc min in diameter, and it is easy to show that the average surface brightness
integrated over a circle of such a diameter is sizeably lower than the actual surface bright-
ness, when the center of the diaphragm is at less than 1.5 arc min from the nucleus. (In
PHYSICO-CHEMICAL PHENOMENA IN COMETS-HI 1251

TABLET. INTEGRATEDBRIGHTNESS B INARBITRARYUNITS,


ANDGRADIENTnOFTHEBRIGHTNESSPROFILEOFTHELIGHT
REFLECTED BY THE ICY HALO, AS A FUNCTION OF THE
REDUCED DISTANCE X FROM THE NUCLEUS (X = r/R,)

logx log B n

-2.2 5.375 1.04


-2.0 5.166 1.06
-1.8 4.951 1.09
-1.6 4.731 1.14
-1.4 4.503 1.19
-1.2 4.255 1.25
-1.0 3.990 1.36
-0.9 3.842 1.49
-0.8 3.685 1.64
-0.7 3.513 I.82
-0.6 3.321 2.04
-0.5 3.101 2.39
-0.4 2.839 2.77
-0.3 2.510 3.79
-0.2 2.056 5.89
-0.1 1.292 12.93
-0.05 0.544 -20

particular, it is obvious that the central peak visible in Fig. 6 cannot be correctly detected
with a diaphragm diameter of the order of one fourth the halo radius.)
However, this curve fitting implies also a vertical adjustment, as the brightness is known
in arbitrary units only. One might presume that the final fitting implies little more than to
find the position where the curvature of the observations satisfies the theoretical curve.
For this reason, a more exacting technique was also used. It implies the use of gradient
n only. O’Dell’s (1961) comments on the observed gradients have been by and large
confirmed by the new treatment of his data presented here. For instance, he writes that the
gradient is >7*0 for r > 315. On May Ist, A = O-245 a.u. for Burnham, therefore 315 =
37,000 km (and not 13,000 km as misquoted by Malaise (1966) p. 201.)
For our present purpose, the two slight bumps of uncertain origin visible in Fig. 8 have
been first smoothed out by drawing the solid line. The underlying assumption is that they
may represent deviations from the steady state if they are not spurious. Hence a ‘smoothed’
gradient was computed from the solid line; its values are the small circles on Fig. 9. The
‘smoothed’ gradient reaches 4 near 21,000 km and 7 near 28,000 km. O’Dell’s statement
quoted before is therefore still verified.
Two possibilities are left to explain such high gradients. Either the grains evaporate, or
isotropy is no more conserved. The latter hypothesis seems at first glance the right one,
because the interaction of the coma with the solar field can distort the fine component of
the dust head. After all, this phenomenon seems to give the correct explanation of the dust
tails, as it was again carefully substantiated by Probstein’s analysis (1968). However, no
simple model using non evaporating dust can give the observed very high values of the
gradient. For instance, Mocknatsch (1938), using a single ejection velocity for the grains,
introduces a constant acceleration acting radially away from the Sun to take into account
the deformation introduced by the radiation pressure. However, he still finds a law in r-l,
whatever the direction, at least when the line of sight is perpendicular to the radius vector
sun-comet. Wallace and Miller (1958) have generalized Mocknatsch’s findings, and show
that, with a single ejection velocity, the law remains in r-l and the isophotes remain circles
1252 A. H. DELSEMME and D. C. MILLER

45

log kmfrom nucleus

I I I
I’ 2’ 3

FIG. 8. CONTINUUM OF COMET BURNHAMAS OBSERVEDBY O’DELL.

log km from nucleus

FIG. 9. BRIGHTNESS GRADIENTS OF THE CONTINUUM OF COMET BURNHAM COMPARED WITH


(A) ICY GRAIN MODEL AND (B) COMA OF DUST PARTICLES DISTORTED BY RADIATION PRESSURE.
The dashed curves are not corrected for resolving power. The solid curves correspond to a
resolving power of 1 arc min.
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1253

whatever the position of the line of sight. However, with a dispersion in the ejection
velocity (their model (b)) they show that the gradient grows to a larger and larger value
when nearing the parabolic envelope of all particles. The only direction where the gradient
remains constant and equal to 1 is the direction radially away from the Sun. It is worth-
while mentioning here that Haser’s (1966) generalization of the model of Wallace and
Miller applies only to exponentially decaying particles and is therefore non applicable for
solid dust. Model (b) of Wallace and Miller is to our knowledge, the only model with
isotropic initial ejection of solid dust which leads to high values of the gradient; but these
high values depend on the orientation of the radius concerned, and disappear opposite to
the Sun.
To compare model (b) with observations, it is therefore necessary to know which radius
of the coma has been observed. O’Dell’s recording of comet Burnham’s profile in the
continuum on May lst, 1960 was obtained by driving at the sidereal rate and allowing
the comet to move itself out by its own motion. The angle of the direction of the recording
with the direction of the Sun is therefore defined. It is about 65”. Using model (b) of
Wallace and Miller the brightness profile at 65” from the Sun was fitted to the observations.
Curve B on Fig. 9 shows the brightness gradients predicted for a good adjustment of the
parameters, (a= O-24 cm/se?, u, = 4.95 x IO4cmfsec, vertex of the envelope ao2/2a =
IO47 km). The fitting technique is now very exacting because it is a translation along the
x axis only. Curve A shows the type of profile given by Table 7, shifted to a vaporizing icy
grain halo whose radius is 1046Bkm. These two theoretical curves could not be compared
with observations without one more adjustment taking space resolution into account.
Indeed when n varies within the space resolution of the photometer diaphragm, an average
# must be used. This averaging procedure diminishes n mainly for the smallest values of the
distance from the nucleus, as r? drops to zero at the nucleus. Figure 9 shows how the two
curves A and B compare with O’Dell’s observations when a space resolution of lo4 km is
introduced; this resolution coincides with O’Dell’s diaphragm. It is obvious here that the
evaporating icy grain model (curve A) fits the observations better than the best adjustment
of a coma deformation for the non evaporating dust model. The major difficulty of the non
evaporating grain model appears clearIy in Fig. 8 and 9; the slope of the gradient curve
(that is, the second derivative of log B) does not vary quickly enough at the crucial distance
when grains are not decaying. A slightly better fit of curve B might be reached only for very
unrealistic values of the parameters, like ejection velocities much larger than 1 km/set for
the grains, coupled with accelerations larger than 1 cm/sec2,
Of course, more conclusive observations of future comets could be easily obtained by
photometry along several diameters or, more simply, radially away from the Sun, where
the gradient of the dust, following the theory of Wallace and Miller (1958), should remain
constant and equal to unity, whatever the acceleration from the radiation pressure.
In the present state of knowledge of the continuum of Comet Burnham, the existence of a
halo of icy particles seems to be the fairest hypothesis to explain the observations.

15. SUMMARY
1. Photodissociation of the water molecule by the solar ultraviolet seems to be able to
explain the large amounts of OH, 0 and H observed in the cometary spectra.
2. The amount of water needed can be vaporized by the solar radiation absorbed by a
nucleus whose effective diameter is, depending on its albedo, somewhere between 4 and 10
km for an average comet.
1254 A. H. DELSEMME and D. C. MILLER

3. If free helium and hydrogen are assumed to have been lost by the comet a long time
ago, the vaporizations of other gases more volatile than water could multiply the total
density of molecules present in the coma by a factor of three to six, with ten as an unlikely
limit. On the other hand vaporization of hydrogen would multiply the gas density by two
orders of magnitude and helium by three orders. However, the recent observations of Ly a
by the OAO and Malaise’s (1970) results on the densities compatible with the observed
mixture of fluorescence and Bolzmann equilibria rule out the vaporization of these two
gases from the nucleus of several comets.
4. Assuming the presence of water snows as a major constituent of the nucleus, the
probable stripping of icy grains by vaporizing gases was established in paper I. It has been
shown here that cometary gravity introduces a sharp cutoff limiting the maximum size of
grains leaving the nucleus. This cutoff excludes the possibility of icy grains larger than 10 ,U
at distances larger than 3.5 a.u., but would allow enormous chunks, if they can exist, to
leave the nucleus at distances smaller than 1 au.
5. Experiments as well as observations point to a mass flow rate of grains that could be
of the same order of magnitude as the mass flow rate of the vaporizing gas.
6. The vaporizing gas, expanding into space, goes from fluid dynamics conditions near
the nucleus, to pure effusion in vacuum at some 10 km near 3 a.u., but at some lo5 km near
1 a.u. In the transition region, the equipartition of energy disappears first for the rotational
and vibrational energies, while the translational energy remains available somewhat longer
for collisions.
7. The expanding gases accelerate icy grains to a terminal velocity which is usually
reached within the first 100 km of the trajectory.
8. The terminal velocity is very much reduced by gravity for the grains whose initial
size is near the maximum size allowed by the gravitational cutoff. These sizes are likely to
be very abundant in the size distribution of particles, because the gravitational cutoff acts
as a sieve, keeping the particles grounded until they vaporize or break into pieces of the
right size to be stripped away.
9. For standard conditions, the life time of grains of the maximum permissible size
is about 200 days whatever the heliocentric distance; the lifetime of the grains reaching the
edge of the halo and setting its size is about 50 days.
10. The largest possible extent of the icy halo cannot ever reach lo5 km whatever the
heliocentric distance; large halos correspond to unrealistic chunks of snow for small
distances. If the size distribution observed in paper I is assumed, the halo appears near
3.5 a.u., reaches its maximum extent at 3 a.u. and shrinks slowly to some lo4 km at 1 a.u.,
3 x lo3 km at O-3 a.u., 100 km at O-1a.u. It could explain the well known shrinking of the
cometary head nearing the Sun.
Il. The photometric profile of the continuum reflected by the icy halo presents a sharp
central peak that simulates rather well the ‘photometric’ nucleus. Its shoulders are much
fainter and should be accessible by photometry only.
12. Comet Burnham seems a proper choice to test the model: it has a rather weak con-
tinuum, not many molecular collisions in the head; however, it showed the dimming of
stars near its nucleus.
13. This optical depth effect observed in comet Burnham can be explained by the
present model. It requires icy grains moving radially with a velocity near 1 m set-l at
600 km from the nucleus; this slow velocity corresponds to grains near the gravitational
cutoff, At 1 a.u. for the standard values of the parameters, this would imply hail grains or
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1255

snowflakes of a size between 1 and 2 cm. A fraction of about 1O-4 of these very large
grains in the distribution observed in paper I could explain the observations.
14. Although the model has been developed as far as it was possible, the compounded
uncertainty on all the parameters involved does not allow more than the prediction of an
order of magnitude for the extent of the halo. It is lo4 km near 1 a.u., if we accept a nuclear
radius of l-10 km, a nuclear temperature of 200”K, and the existence of icy grains in the
centimeter range. It could be somewhat larger only for a larger nucleus, or for a higher
nuclear temperature.
15. O’Dell observed the photometric profile of the continuum of comet Burnham.
It shows a gradient becoming very large at distances greater than 20,000 km. A very good
fit of the shape of the photometric profile is given by the evaporating grain model by using
a halo radius of 104.65km. This radius could be somewhat smaller if the observed con-
tinuum was still to some degree polluted by the light of C, or reflected by some non-
evaporating dust, as is not unlikely for the faintest measurements (O’Dell 1961).
16. A further boost in the efficiency of the process could be had if grains of the most
volatile ices were also launched. These would evaporate close to the nucleus and add to the
gas stream acting on the H,O grains, thus increasing the terminal velocities of the H,O or
clathrate ice grains of the largest sizes initially launched. This should help to remove the
remaining discrepancy between theory and observation.
17. It is interesting to note that no good fit can be found for the photometric profiles
of comet Burnham, even by using a deformation of the coma, if non-evaporating dust is
used only.
In a forthcoming paper IV, the influence of the icy halo on the photometric profile of
the molecular emission bands is studied and applied to the profile of C, in comet Burnham.
A general discussion of the results is given at the end of paper IV.
Acknowledgements-We want to thank Dr. Alfan F. Cook who made several substantive suggestions and Dr.
Swings who read the manuscript and offered some helpful comments.
Dr. O’Dell was kind enough to lend us his original photometric recordings of comet Bumham. An
exchange of letters with Dr. Probstein was helpful to understand a point in his approach that was obscure
to us. Dr. Freeman Miller mentioned several useful references.
This research was made possible by the grant No. GP-17712 of the National Science Foundation.

REFERENCES
BIERMANN,L. and TREFFTZ,E. (1964). 2. Astrophys. 59, l-28.
DELSEMME,A. H. (1966). Proc. CoNoques Universitkde Liige, Belg. 37,77; also in M&m. Sot. R. Sci. Likge
12, 77 (1966).
DELSEMM~,A. H. and MILLER, D. C. (1970). Planet. Space Sci. 18,717.
DELSEMM~?, A. H. and SWINGS,P. (1952). Am& Astrophys. l&l.
DELSEMME,A. H. and WENGER,A. (1970). Planet. Space Sci. 18,709.
DOSS~N,F. (1962). J. Obsrs 45, 1.
FINSON, M. L. and PROBSTEIN,R. F. (1968). Astrophys. J. 154,327 and 353.
GOODY, R. M. (1964). Atmospheric Radiation, p. 42. Oxford University Press.
HASER,L. (1966). Proc. CoNoques Universitt!de Li@e, Belgium, 37,233; also in Mkmoires Soci&& Royale des
Sciences de Li&e, 12,233.
HUEBNER,W. F. (1965). 2. Astrophy. 63,22.
HUEBNER,W. F. and WEIGERT,A. (1966). 2. Astrophys. 64,185.
JOHNSON,V. J. (1960). I. Rroperties of~a~ds (Hydrogen and Helium). Edit. U.S. Dept. of Commerce,
PB 17X 618, Washington, D.C.
KENNARD,E. H. (1938). Kinetic Theory of Gases, p. 77. McGraw-Hill, New York.
KNUDSEN, M. (1910). Ann& Physik 34, 593.
LANGMUIR,I. (1913). Phys. Rev. 2,329.
LANGMUIR,I. (1917). Phys. Rev. 3,141.
LYTTLETON,R. A. (1968). Mysteries of the Solar System, p. 168. Oxford University Press, London.
1256 A. H. DELSEMME and D. C. MILLER

LYUBITOV,Y. N. (1967). MolewEar Flow in Vessels (Consultants Bureau Translation)Plenum Publ. Corp.,
New York.
MCNESBY,5. R. and OKABE,H. (1964). Advames in Photochemistry (Ed. W. A. Noyes et al.), Vol. 3, p. 157.
Interscience, Wiley, New York.
MALAISE,D. (1966). Proc. Colloques Uniuersitd de Likge, 37, 199; also in Memoires Societe Royafe des
Sciences de Liege, 12, 199.
MALAISE,D. (1968). Ph.D. Thesis, Universite de Liege, Belgium.
MALAISE,D. (1970). Astron. and Astrophys. 5,209.
MARKOVICH,M. A. (1963). Bull. Commission on Comets and Meteors USSR 8, 11 (in Russian).
MASSEY,H. S. W. and BURHOP, E. H. S. (1952). Elecrronic and Ionic Impact Phenomena, p. 460. Oxford
University Press.
MILLER, F. D. (1961). Astrophys. J. 134,1007.
MILLER, F. D. (1967). AstronomicalJ. 72,487.
MOCKNATSCH,D. 0. (1938). Leningrad State Univ. Annals, Astr. Series, Issue 4.
O’DELL C. R. (1961) Publ. Astran. Sot. of the Pacific 73, 35.
O’DELL, C. R. and OSTERBROCK, D. E. (1962). Astrophys. J. 136,559.
OVERSTREET,R. and GIAUQUE,W. F. (1937). J. Amer. Chem. Sot. 59,254.
PROBSTEIN, R. N. (1968). Problems of Hydrodynamics and ~onfinuam Mechanics, p. 568. Edit. Society for
Industrial Mathematics, Philadelphia.
ROEMER,E. (1966). Proc. Collogues Universite’de LiPge, Be&am 37,23; also in Memoires SociCte Royale des
Sciences de Liege 12,23.
ROEMER,E. (1969). Personal communication.
SHULMAN,L. M. (1967). Summarized by G. Guigay for Commission 15 Meeting, I.A.U., Prague.
UNG, A. Y. M. and BACK R. A. (1964). Can. J, Chem. 42,753.
VENUGOPALAN,M. and JONES,R. A. (1968). Chemistry of Dissociated Water Vapor arrd Related Systems,
p. 200. Interscience, Wiley, New York.
WALKER, J, F. (1964). Formaidehyde. Reinhold Publishing Corp. New York.
WALLACE,L. V. and MILLER, F. (1958). Astronom. J. 63,213.
WASHBURN,E. W. (1928). International Critical Tables 3,210. McGraw Hill, New York.
WATSON,K., MURRAY, B. and BROWN, H. (1961). J. geophys. Res. 66,3033.
WEIGERT,A. (1959). Astron. Nachrichten 281, 117.
WENGER,A. (1968). Master’s Thesis, The University of Toledo, Toledo, Ohio.
WHIPPLE, F. (19SO). Astrophys. J. 111, 375.
WHIPPLE, F. (1951). Astrophys. J. 113,464.
WHIPPLE, F. (1955). Astrophys. J. 121,750.
ZIEGLER,W. T. and MULLINS,J. C. (1963). Vapor Pressures and Latent Heats V. Nitrogen and Fluorine. Edit.
U.S. Dept. of Commerce, PB168896, Washington, D.C.
ZIEGLER,W. T., MULLINS, J. C. and KIRK, B. S. (1962). Vapor Pressures and Latent Heats III. CH,. Edit
U.S. Dept. of Commerce PB172347, Washington, D.C.

APPENDIX 1

Momentum Traasfer to a Spherical Grain


Symmetry considerations show that the only momentum transfer that is to be taken
into account is the component aIong the direction 2 of the radial veIocity of the gas; the
gas can therefore be represented by the simplest model of a stream of particles all moving in
direction 2 with the mean radial velocity u,. The radius of the grain is a, and the point of
molecular impact is defined by the angle M.between the impact radius and the 2 directed
particles.
As the collisions are supposed to be elastic, when a molecule bounces back from the
spherical grain, the transfer of momentum along the 2 axis is

pnv,(l + cos Zcr)

(instead of 2mv, for the flat disk model).


The cross sectional area 2rra da of the ring of radius a = a, sin CCis 2rra02 sin a cos a du
and the total transfer of momentum is therefore (if PO = 2~a,mv,,Z is the momentum transfer
PHYSICO-CHEMICAL PHENOMENA IN COMETS-III 1257

to a flat disk of radius a*),

P = PO "(1+ cos24 sin a cos ccda = *PO.


s0
When the spherical grains reflect gas molecules diffusely in all directions, the mean momen-
tum transferred by the gas is therefore exactly one half the momentum transferred onto a
flat disk of the same radius.

You might also like