Mon. Not. R. Astron. Soc. 000, 000–000 (0000)
Printed 5 November 2018
(MN LATEX style file v2.2)
arXiv:1105.4631v2 [astro-ph.SR] 25 Jul 2011
On the effects of microphysical grain properties on the
yields of carbonaceous dust from type II SNe
David W. Fallest1⋆ , Takaya Nozawa2, Ken’ichi Nomoto2, Hideyuki Umeda3,
Keiichi
Maeda2, Takashi Kozasa4 and Davide Lazzati1
1
2
3
4
Department of Physics, NC State University, 2401 Stinson Dr., Raleigh, NC 27695-8202
Institute for the Physics and Mathematics of the Universe, University of Tokyo, Kashiwa 277-8583, Japan
Department of Astronomy, School of Science, University of Tokyo, Tokyo 113-0033, Japan
Department of Cosmosciences, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan
5 November 2018
ABSTRACT
We study the role of the unknown microphysical properties of carbonaceous dust particles in determining the amount and size distribution of carbonaceous dust condensed
in type II supernova explosions. We parametrize the microphysical properties in terms
of the shape factor of the grain and the sticking coefficient of gas-phase carbon atoms
onto the grain surfaces. We find that the amount of dust formed is fairly independent
of these properties, within the parameter range considered, though limited by the
available amount of carbon atoms not locked in CO molecules. However, we find that
the condensation times and size distributions of dust grains depend sensitively on the
microphysical parameters, with the mass distributions being weighted toward larger
effective radii for conditions considering grains with higher sticking coefficients and/or
more aspherical shapes. We discuss that this leads to important consequences on the
predicted extinction law of SN dust and on the survival rate of the formed grains
as they pass through the reverse shock of the SN. We conclude that a more detailed
understanding of the dust formation process and of the microphysical properties of
each dust species needs to be achieved before robust prediction on the SN dust yields
can be performed.
Key words: dust, extinction — supernovae: general
1
INTRODUCTION
Interstellar dust, once considered to be little more than
a nuisance to astronomical observations, is one of the
most interesting areas of astrophysical research today
(Li & Greenberg 2003). One aspect of particular interest
is the origin of dust at high redshift (z > 5). Possible
sources of interstellar dust that have been considered
include outflows from asymptotic giant branch (AGB) stars
(Morgan & Edmunds 2003; Zhukovska, Gail, & Trieloff
2008; Valiante et al. 2009), Wolf-Rayet systems (Cherchneff
2010), quasars (Elvis, Marengo, & Karovska 2002), and
supernova
explosions
(Kozasa, Hasegawa, & Nomoto
1989, 1991; Todini & Ferrara 2001; Nozawa et al. 2003;
Bianchi & Schneider 2007; Kozasa et al. 2009; Cherchneff
2010). How much dust can be attributed to each of these
possible sources at such high redshift remains unclear.
Supernovae (SNe) are considered by some to be likely
contributors of much of the dust in the early Universe
⋆
E-mail: dwfalles@unity.ncsu.edu
because their progenitors are quite massive and consequently have short lifetimes (Bromm, Coppi, & Larson
2002; Morgan & Edmunds 2003; Maiolino et al. 2004;
Dwek, Galliano, & Jones 2007). However, such early SNe
have not been observed, making determinations of their
dust contributions difficult. Instead we need to consider
dust yield predictions of more recent SNe, which have
been observed, and then extrapolate the dust yields to
earlier SNe. The theoretical predictions of dust yields for
recent supernovae (Kozasa, Hasegawa, & Nomoto 1991;
Todini & Ferrara 2001; Nozawa et al. 2008), however, are
too large compared to observations, with discrepancies
that can be as high as 3 to 4 orders of magnitude in some
cases (Hoyle & Wickramasinghe 1970; Lucy et al. 1989;
Wooden et al. 1993; Elmhamdi et al. 2003; Meikle et al.
2007; Kotak et al. 2009). Observations of young supernova
remnants (SNR), however, have confirmed dust masses one
to two orders of magnitude greater than SNe. Some of these
SNR have observed dust masses of 0.02–0.054 M⊙ by Spitzer
(Rho et al. 2008), 0.06 M⊙ by AKARI (Sibthorpe et al.
2010), and 0.075 M⊙ by Hershel (Barlow et al. 2010), in
2
Fallest et al.
Cassiopeia A, and 0.04–0.1 M⊙ by Spitzer in the pulsar
wind nebula G54.1+0.3 (Temim et al. 2010). Additionally,
Nozawa et al. (2010) have demonstrated that the observed
spectral energy distribution of Cassiopeia A can be well
reproduced by the calculations of dust formation in Type
IIb SNR and the mass in the SNR is 0.07 M⊙ .
Reconciling the dust yield prediction for local SNe can
be done via one of two channels: improving our understanding of the dust formation process, leading to a substantially decreased prediction, or revising the observational constraints to account for a higher dust yield than so far implied. It is possible that some amount of dust has avoided observation. Dust particles absorb light and re-emit the energy
at infrared wavelengths. However, cold dust at temperatures
of a few tens of Kelvin could escape detection at mid-infrared
wavelengths. Additionally, areas where the dust is optically
thick could obscure some amount of dust, again allowing
some dust to not be detected. Dust clumping may also affect
the estimates of dust mass from absorption (Sugerman et al.
2006), since it is usually assumed that the surface filling
factor of dust is close to unity, while substantial clumping could be present due to the intrinsic inhomogeneity of
the ejecta and Rayleigh-Taylor instabilities in the expanding
ejecta (Wooden 1997; Douvion, Lagage, & Cesarsky 1999).
On the theoretical side, all the estimates for dust production in SNe are based on the so-called classical theory
of nucleation (e.g., Becker & Döring (1935)). It has been argued, however, that the use of classical nucleation theory in
astrophysical environments is questionable (Donn & Nuth
1985; Lazzati 2008). In addition, almost all SN dust nucleation models thus far have considered the formation of
spherical grains, and assumed any atoms/molecules that
contact the grain will adhere to the grain; conditions that
reflect maximally efficient nucleation. It is therefore not entirely surprising that the theoretical estimates of SN dust
yields, based on upper limit of nucleation efficiency, are in
excess of those from observations. It should be noted that
Bianchi & Schneider (2007) have considered less than maximally efficient nucleation by assuming the probability of
atoms/molecules adhering to the grain is less than unity,
resulting in smaller dust yields.
Since dust formation is a highly non-linear phenomenon, understanding the effects of different nucleation
rates on the final dust yields is difficult. To check the effects of different nucleation conditions we have performed a
parametric study, in which we consider carbonaceous dust
production in a SN explosion by varying the shape of the
forming grains as well as the sticking coefficient, i.e., the
probability that an incoming monomer will stick to the
grain rather than bounce off and remain in the gas phase.
Our study is phenomenological and aims at understanding
which conclusions of previous nucleation studies are robust
to changing the parameters, and which may need to be
investigated more thoroughly. A self-consistent nucleation
model in astrophysical conditions will be achieved by involving a kinetic approach (Donn & Nuth 1985; Lazzati 2008;
Keith & Lazzati 2011) and detailed chemistry of precursor
molecules (Cherchneff & Dwek 2009, 2010). Such a detailed
approach, however, is still under development and is not yet
applicable to large scale simulations like the one that we use
here, and that have been used in previous investigations of
e.g., Todini & Ferrara (2001); Nozawa et al. (2003, 2010).
This paper is organized as follows: in Sect. 2 we detail
the nucleation theory that we adopted; in Sect. 3 we describe
the numerical code used for the computations; and in Sect. 4
we describe our results. In Sect. 5 we finally discuss the
implication and limitations of our results and lay out future
perspectives for SN dust studies.
2
NUCLEATION
Nucleation is the first step of a first-order phase transition.
In the case we consider here, the phase transition is from a
gas of carbon atoms to solid clusters of amorphous carbon
dust. There is a phase equilibrium pressure where both the
gas and solid phases are stable within a given volume. The
phase equilibrium pressure depends on the temperature of
the materials in the gas phase within the volume. Nucleation
of clusters of atoms/molecules of the new phase is favoured
when the gas phase is supersaturated (i.e., the pressure of
the gas is higher than the phase equilibrium pressure). The
higher the supersaturation (i.e., the ratio of the pressure
to the phase equilibrium pressure), the smaller the size of
the stable clusters that are able to form. The size of the
smallest stable cluster able to form at a given temperature
and density is called the critical cluster size. Clusters that
are smaller than the critical size will tend to evaporate, while
larger clusters will tend to continue to grow. The goal of any
nucleation theory is to calculate how many critical clusters
form per unit volume per unit time.
The classical theory of nucleation considers nucleation as a thermodynamical process in quasi-equilibrium
(Becker & Döring 1935; Feder et al. 1966; Kashchiev 2000).
Besides the supersaturation level, the physical properties
of the nucleating material affect the size of a critical cluster. The classical nucleation theory assumes that all clusters share the same properties, such as surface tension and
shape, independent of their size. The theory also assumes
that those properties are equal to those of a macroscopic
sample. Moreover, the clusters are assumed to have a uniform equilibrium temperature that is equal to the temperature of the surrounding gas. These basic assumptions are
problematic because macroscopic thermodynamic properties
are not expected to be applicable to clusters of only a few
atoms.
A different approach is provided by the kinetic theory
of nucleation, which is applicable to very small cluster sizes.
The kinetic theory relies on calculating the attachment and
detachment frequencies of monomers to a cluster (Kashchiev
2000). Furthermore, in contrast to the classical nucleation
theory, the kinetic theory follows the formation of clusters
smaller than the critical cluster size. In the framework of the
kinetic theory the critical cluster is the cluster whose attachment and detachment frequencies are equal, and thus it is
stable. The downside of the kinetic theory is that attachment and detachment frequencies for all cluster sizes need
to be calculated in order to determine the overall nucleation
rate. The main aim of this paper is to reveal the dependence
of dust formation processes on the microphysical properties
of grains. In order to achieve this, we adopt the simpler classical nucleation theory, rather than using the kinetic theory
that demands a more complicated treatment.
When the supersaturation level is S > 1, nucleation can
SN dust microphysics
take place because the free energy of the new phase is lower
than that of the old phase (Kashchiev 2000). The change
in the free energy is due to the work necessary to form the
critical size clusters. Energy is released in the formation of
the volume of the cluster, but energy is required in order
to form the surface of the cluster (Kashchiev 2000). The
nucleation rate (Eq. (1)) is given basically by two factors: the
number density of critical clusters and a kinetic factor that
describes the rate at which clusters become large enough to
be stable (i.e., critical size or larger).
The general equation of stationary nucleation is given
by (Kashchiev (2000), their equation 13.39):
∗
JS = A exp (−W /kT ),
(1)
where A is the kinetic factor, W ∗ is the work needed to
form the critical size cluster from the gaseous state, k is
the Boltzmann constant, and T is the gas temperature. The
kinetic factor, A, is given by A = A′ exp (∆µ/kT ), and A′
is:
A′ = γ ∗
c3 σ
18π 2 m0
21
pe v0
kT
C0 ,
where γ is the size dependent sticking coefficient, c is
the shape factor of the cluster, σ is the surface tension,
m0 and v0 are, respectively, the mass and volume of the
monomer of nucleating material, pe is the phase-equilibrium
pressure, and C0 is the concentration of sites where nucleation can occur (equations 13.41 and 13.44, respectively in
Kashchiev (2000)). The concentration of gaseous monomers,
C1 is related to the concentration of nucleation sites by
C1 = C0 exp (−W1 /kT ), where W1 is the work needed to
form a cluster consisting of one monomer (equation 7.5 in
Kashchiev (2000)). We consider the monomer in the gaseous
state to be indistinguishable from the monomer in the condensed state, so that W1 = 0 and C1 = C0 . Using the supersaturation ratio, S = C1 /C1,e = p/pe and ∆µ = kT ln S,
where C1,e = pe /kT is the gaseous monomer concentration
at the phase-equilibrium pressure and p is the partial pressure of gaseous monomers, we obtain
c3 v02 σ
18π 2 m0
12
C12 .
Js = γ
c3 v02 σ
18π 2 m0
21
C12
exp
−4c3 v02 σ 3
27(kT )3 (ln S)2
.
(2)
After nucleation the clusters grow through impingement
of monomers upon the cluster. To find how much the clusters
grow we begin by finding the volume of the newly nucleated
cluster. The clusters nucleate with some critical number of
monomers, n∗ (equation 4.7 in Kashchiev (2000)):
n∗ =
8c3 v02 σ 3
.
27∆µ3
2
dV
= γcV 3 C1 v0
dt
kT
2πm0
21
.
(4)
In the same manner as Nozawa et al. (2003) we compute
the depletion of the available nucleation material through
mass conservation:
1−
C1 (t)
= 1 − Y1 =
C̃1 (t)
Z
t
te
J(t′ ) V (t, t′ ) ′
dt ,
C̃1 (t′ ) v0
(5)
where C̃1 is the nominal concentration of monomers – the
concentration expected should nucleation not occur, which
in an expanding shell of volume Vshell can be found using
so that the total number of gas-phase atoms remains constant, and V (t, t′ ) is the volume of a cluster formed at time
t′ and measured at time t. Rather than computing the integral on the right-hand side, we instead calculate C1 (t) as
described in step (iii) in the next section. From Eqs. (2)
and (4), we see that Js and dV /dt are simply proportional
to the sticking coefficient γ. Thus, we expect that reduced
sticking coefficients will suppress both nucleation and grain
growth. The dependence of Js on the shape factor c is more
complicated, since it appears in both the kinetic factor and
the exponential term. The shape factor in the exponential
term, however, will dominate the nucleation rate equation
and we expect that increased shape factors will suppress nucleation. On the other hand, grain growth (dV /dt) is simply
proportional to c, and thus an increase in the shape factor
will increase the cluster growth rate.
3
While the sticking coefficient may depend on the size of the
cluster, we consider it to be constant, so that γ ∗ = γ = constant. Finally, using the work to form a critical sized cluster
W ∗ = 4c3 v02 σ 3 /27∆µ2 (Kashchiev (2000), equation 4.8), we
find our stationary nucleation rate equation to be:
concentration of monomers C1 , the volume of the monomer
v0 , and the average relative speed of the monomers with
respect to the cluster. In this paper we consider clusters
that can nucleate with aspherical shapes through the use
of the shape factor c. The shape factor is a dimensionless
quantity that relates the surface area Σ of an object to its
volume V by: c = Σ/V 2/3 (Kashchiev 2000). Thus we find
the change in volume over time to be:
C̃1 (t = tn )Vshell (t = tn ) = C̃1 (t = t0 )Vshell (t = t0 ),
∗
A = γ∗
3
(3)
The volume of the critical cluster is then, v ∗ = n∗ v0 . The
change in the cluster’s volume over time depends on the
sticking coefficient γ, the surface area of the cluster Σ, the
SIMULATIONS
We concentrate on the formation of carbonaceous grains
(clusters) from carbon atoms in the expanding material of
a core-collapse supernova. Our simulations are based on the
hydrodynamic results and elemental composition for the
unmixed ejecta of a core-collapse supernova of a 20 M⊙
progenitor star with metallicity Z = 0 and an explosion
energy of 1051 ergs by Umeda & Nomoto (2002) (see also
Nomoto et al. (2006)).
Table 1 describes the data necessary for the calculation
of carbon grain formation. To compute the supersaturation
of the expanding gas, we find the phase equilibrium pressure
by: pe = p0 e−A/T +B , where p0 is the standard pressure, T
is the temperature, and the values of A and B, listed here
in Table 1, are taken from Table 2 of Nozawa et al. (2003).
As the material ejected by the core-collapse supernova expands it also cools. Figures 1 and 2 show the evolution of
the density and temperature, respectively, of the expanding
and cooling material for two of the enclosed mass subshells
that we refer to in the rest of this work.
We study four different values of the unknown sticking
Fallest et al.
4
10-5
6.0
10-7
0
☞2
(number fraction)
10-8
10-9
10-10
10-11
10-12
log
·
density (g cm 3 )
M✁
M✂
4.96
10-6
10-13
400
600
time (days)
800
106
M✄
M☎
2
3
4
5
6
enclosed mass (M ✆)
7
8
Figure 3. Number fraction of helium (green dash-dot), carbon
(black dashed), oxygen (red dotted), and silicon (magenta dashdot) atoms for enclosed masses from 2.45 to 8.7 M⊙ . The solid
(blue) line is the carbon number fraction after the formation of
CO molecules.
4.96
temperature (K)
105
104
103
102
200
400
600
time (days)
800
1000
Figure 2. Temperature evolution for 4.96 (black) and 6.0 (red)
M⊙ enclosed mass subshells up to 1000 days after the SN explosion.
coefficient, γ = 1.0, 0.1, 0.01, and 0.001, neglecting any dependence of γ on the temperature of the gas and the size
of the cluster. For each value of the sticking coefficient, we
study six different values for the shape factor, c = (36π)1/3 ,
5.4, 6.0, 7.0, 9.0, and 12.0, corresponding to shapes ranging from a sphere to a flattened cylinder similar to a coin.
We therefore performed 24 simulations in total. A sticking
coefficient of γ = 1.0, and a shape factor for a sphere of
c = (36π)1/3 are the usual parameters used in previous nucleation works.
As in Nozawa et al. (2003), we assume that the stable
formation of CO molecules occurs prior to grain nucleation
and that carbon grains will form only where the initial num-
ber fraction of carbon is higher than that of oxygen. Under
this assumption, the number fraction of free carbon atoms
available for dust formation is obtained simply from the
initial number fraction of carbon minus the number fraction oxygen. We divide the expanding gases into a series of
enclosed mass shells beginning at ∼ 4.93 M⊙ and ending
at ∼ 6.21 M⊙ . In this range of enclosed masses the number fraction of carbon atoms, after the formation of CO,
is highest; ranging between 2 × 10−1 and 8 × 10−9 . Figure
3 shows the number fraction of carbon and oxygen atoms
in the expanding ejecta from the hydrodynamic results of
Umeda & Nomoto (2002). The solid (blue) line indicates the
number fraction of carbon left over after the formation of CO
molecules. It should be noted for completeness that grain
nucleation can occur at enclosed masses larger than we consider here, but is extremely inefficient due to low carbon
number fractions, ∼ 10−14 and below.
For each mass subshell, our code starts by following the
evolution of the density and temperature (see Figures 1 and
2, respectively, for examples) of the gas until the condition
of supersaturation is satisfied. From that point on, at each
time step the code performs three operations.
(i) First, the code computes the number of critical clusters that are formed given the saturation, temperature, and
partial pressure of the carbon atoms (according to Eq. (2)).
(ii) Second, the code grows any pre-existing grain formed
at earlier times according Eq. (4).
(iii) Finally, the code subtracts from the carbon in the
gas phase the amount of carbon that has been locked in
the solid phase by the processes in steps (i) and (ii). The
concentration of gas-phase carbon is evolved according to:
C1 (tn )
Table 1. Carbon properties
8.64726
✟10
1000
6.0
A/104
(K)
✠8
✝14
200
Figure 1. Density evolution for 4.96 (black) and 6.0 (red) M⊙
enclosed mass subshells up to 1000 days after the SN explosion.
101 0
He
C
O
Si
C minus O
✡6
✞12
10-14
10-15 0
☛4
B
σ
(erg·cm−2 )
v0
(10−24 cm−3 )
m0
(10−23 g)
19.0422
1400
8.805
1.995
=
C1 (tn−1 ) −
−
∆Vgrains,n
v0 Vshell,n
C1 (tn−1 )
C̃1 (tn−1 ) − C̃1 (tn ) ,
C̃1 (tn )
where ∆Vgrains,n is the total change in volume of grains from
the previous time step to the current time step. Then, the
SN dust microphysics
✑5
✜
(cm
✛5
logJ
logJ
(cm
3
✒ ✏10
3
·
✢
·
✓
0
s 1)
s 1)
0
✎15
✚10
✔ =1 0
✕ =0 1
✖ =0 01
✗ =0 001
.
✍20
.
350
355
time (days)
360
365
.
.
.
345
✣ =1 0
✤ =0 1
✥ =0 01
✦ =0 001
.
.
✌25340
5
✙154.8
370
Figure 4. Nucleation rates for spherical grains ( c = (36π)1/3 ),
for the four considered sticking coefficients of γ = 1.0, 0.1, 0.01,
and 0.001. Filled circles indicate the time and rate of maximum
nucleation. Rates are calculated at an enclosed mass coordinate
4.96M⊙ .
.
5.0
5.2
5.4
5.6
enclosed mass (M ✘)
5.8
6.0
Figure 5. Maximum nucleation rates for spherical carbon grains
at enclosed masses < 6.2 M⊙ for four sticking coefficients.
✯4
This process is repeated for each mass shell until the concentration of gas-phase carbon is reduced to 1 per cent of
its original value. The dust distributions from all shells are
then summed together to produce the final dust yield of each
particular set (γ, c).
4
4.1
RESULTS
Spherical Grains
After the supernova explosion, the hot gases expand and
cool. The cooling allows the gases to reach supersaturation
conditions. Once the supersaturation level becomes greater
than unity, nucleation can occur. The supersaturation level
continues to rise and the nucleation rate increases over time
until depletion of available material becomes significant and
the supersaturation level begins to drop, after which the
nucleation rate falls off quickly. Eventually the gas is no
longer supersaturated and nucleation ceases. However, grain
growth is still possible.
Figure 4 shows the nucleation rate for spherically
shaped grains at enclosed mass coordinate 4.96 M⊙ for the
four sticking coefficients. We chose this particular enclosed
mass shell because it contains the highest abundance of carbon atoms, after CO formation, of all our mass shells. The
solid curve corresponds to c = (36π)1/3 and γ = 1.0, the parameters generally used for nucleation studies. The dashed,
dash-dot, and dotted curves correspond to γ = 0.1, 0.01,
and 0.001, respectively. To be consistent with the works of
Kozasa & Hasegawa (1987); Kozasa, Hasegawa, & Nomoto
(1989, 1991); Nozawa et al. (2003), we consider the time at
✮6
s 1)
✭8
·
✱ ✬10
(cm
3
✰ ✫12
logJ
first term on the right-hand side is the concentration of gasphase carbon from the previous time step, the second term
is the change in the concentration of solid-phase carbon (it
is a summation of all grain changes, including formation of
new critical clusters and the growth of existing grains), and
the final term accounts for the decrease in concentration of
gas-phase carbon due to the expansion of the shell.
✪14
✲ =1 0
✳ =0 1
✴ =0 01
✵ =0 001
✩16
.
.
★18
✧20350
.
.
400
450
500
time (days)
550
600
Figure 6. Nucleation rates for spherical carbon grains at an enclosed coordinate of 6.00 M⊙ for four sticking coefficients. Condensation times are indicated by filled circles.
which the nucleation rate is at its maximum to be the condensation time of the grains. We show these condensation
times as filled circles at the maxima of the nucleation rates
in the figure.
At early times, the reduced sticking coefficient makes
the formation of critically sized grains more difficult and
results in a suppressed nucleation rate. In the absence of
strong nucleation, carbon atoms are not depleted from the
gas and the saturation continues to increase. The reduced
sticking coefficient thus causes the time at which nucleation
is at its maximum to be delayed, and the nucleation to take
place at higher saturation levels. As a consequence, a larger
number of critical clusters can form with much smaller size
(Eq. 3).
Figure 5 shows the maximum nucleation rates for spherical grains as a function of enclosed mass for our four sticking
coefficients. For enclosed masses up to Mr ∼ 5.87 M⊙ , the
nucleation rate maxima for reduced sticking coefficients exhibit similar behaviour as in Figure 4. At greater enclosed
masses, however, the behaviour is inverted and a reduced
Fallest et al.
6
600
550
✷ =1.0
✸ =0.1
✹ =0.01
✺ =0.001
55
✼ ✽ 3. 5
r
dN/dr)
time (days)
500
log(
450
45
400
350
40
5.2
5.4
5.6
enclosed mass (M ✶ )
5.8
6.0
Figure 7. Condensation times of spherical carbon grain as a
function of enclosed mass for four sticking coefficients.
sticking coefficient results in depressed maximum nucleation
rates. This difference in behaviour is due to the reduction
in the number fraction of carbon available, from ∼ 10−2.5
to ∼ 10−5 (see Figure 3), which corresponds to a reduction
in the concentration of carbon monomers. Since the nucleation rate (Eq. (2)) is proportional to the concentration of
monomers squared, the drop in the carbon concentration
consequently drops the nucleation rate. Thus, the nucleation
rates do not peak as strongly (Figure 6), drawing out nucleation to later times for all sticking coefficients, so that
a catastrophic reduction in the available material does not
occur. Thus, the nucleation rates for reduced sticking coefficients remain depressed throughout the simulation time for
higher enclosed mass shells.
Figure 7 shows the condensation times corresponding to
the maximum nucleation rates shown in Figure 5. The condensation times of the solid curve (c = (36π)1/3 , γ = 1.0)
are in good agreement with the condensation times for carbon grains reported in Nozawa et al. (2003). Here the lower
sticking coefficients result in delayed condensation times up
to Mr ∼ 5.87 M⊙ , outside which the condensation time is
only slightly delayed, or no longer delayed, when compared
to the case of γ = 1.0. The more noticeably delayed condensation times, as well as the similarity of condensation times
for all our sticking coefficients, at higher enclosed mass shells
are also due to the drawn out nucleation process already discussed above.
Reducing the sticking coefficient makes both nucleation
and grain growth more difficult. A reduced sticking coefficient results in a larger number of smaller grains, fewer larger
grains, and a smaller maximum grain radius. In Figures 8
and 9, respectively, we show the size and mass distributions
for spherical grains. For the case of γ = 1.0, the maximum
grain size achieved is between 2 and 3 µm. When the sticking coefficient is reduced to 0.001, the maximum grain size
is decreased to less than 0.01 µm.
Since the reduced sticking coefficient causes larger numbers of small grains to form, the small grains contain more
relative mass than the larger grains, as can be seein in Figure 9. However, the total mass of the grains is relatively
robust. The total masses of dust grains for the spherical
✾ =1.0
✿ =0.1
❀ =0.01
❁ =0.001
10
-3
10
-2
10
-1
✻m)
10
0
10
1
radius (
Figure 8. Size distribution of carbon grains for γ = 1.0,
0.1, 0.01, 0.001, with c = (36π)1/3 . For reference, the solid
(red) line represents the power-law distribution with the form
of Nr ∝ r −3.5 , which has been suggested as that of interstellar
grains (e.g., Mathis, Rumpl, & Nordsieck (1977)).
2
❇ =1.0
❈ =0.1
❉ =0.01
❊ =0.001
0
dM/dr)
5.0
log(
4.8
50
❆2
❅4
❄6
❃8
10
-3
10
-2
10
-1
❂m)
10
0
10
1
radius (
Figure 9. Mass distribution of carbon grains for γ = 1.0, 0.1,
0.01, 0.001, with c = (36π)1/3 .
case are shown in Table 2, along with the aspherical cases
which are discussed in the next section. Even though the
onset of nucleation is delayed due to the reduced sticking
coefficient at enclosed masses less than 5.87 M⊙ , where the
majority of available carbon is contained, the subsequent
grain growth consumes almost all of the carbon atoms for
γ > 0.001. Thus, the total mass of carbon dust is principally
determined by the mass of pre-existing carbon atoms.
4.2
Non-spherical Grains
We also calculated nucleation rates for aspherical grains. We
choose a range of shape factors from c = 5.4 to c = 12.0.
While c = 6.0 is the shape factor of a cube, each shape
factor can correspond to a number of different grain shapes.
The shape factor can be thought of as a deviation from the
spherical case, the bigger the shape factor, the larger the
SN dust microphysics
■5
7
5.0
▲ =1 0
.
4.8
·
M
1
❑
)
s )
4.6
❍10
4.2
Mass (10
●15
logJ
(cm
4.4
2
◆
3
❏
❖
❋20
340
350
360
370
time (days)
380
▼
390
3.8
P =1 0
◗ =0 1
❘ =0 01
❙ =0 001
1/3
=(36 )
c =5.4
c =6.0
c =7.0
c =9.0
c =12.0
c
4.0
.
3.6
.
.
3.4
.
400
3.2
5
7
6
8
9
10
11
12
shape factor c
Figure 10. Nucleation rates as a function of time for six shape
factors with γ = 1.0. Filled circles indicate maximum nucleation
rate. Rates shown are for nucleation within a shell of enclosed
mass of ∼ 4.96 to ∼ 4.97 M⊙ .
Figure 15. Total mass of carbon dust formed for all simulations.
deviation from a sphere and the larger the surface for a
given volume.
In Figure 10 we show the nucleation rates for the same
enclosed mass shell as shown in Figure 4, for increasing
shape factors with γ = 1.0. Since the shape factor appears
in the exponential term of the nucleation rate equation as
Js ∝ exp (−c3 ) (Eq. (2)), the higher shape factors reduce
the nucleation rate. Therefore, the higher supersaturation
levels at later times need to be attained so that the nucleation rate becomes high enough for significant depletion of
the gas due to the growth of newly formed grains. However,
the maximum nucleation rates for increasing shape factors
are decreased. This is due to the fact that with a larger surface to volume ratio aspherical grains tend to grow faster,
producing a sizeable depletion of the gas-phase carbon even
for moderate nucleation rates. For all mass coordinates, increasing the shape factor leads to lower nucleation rates and
delayed condensation times (Figures 11a and 12a, respectively).
In Figure 11b–d we show the nucleation rate maxima
for all shape factors for each of the other three sticking coefficients. For any sticking coefficient, the shape factor has
the same effect of suppressing the nucleation rate. All condensation times, shown in Figure 12b–d, increase with the
increased shape factor. Furthermore, for reduced sticking coefficients, greater shape factors lead to greater delays in the
condensation time.
The shape factor c = (36π)1/3 is the only shape factor
that has just one associated shape, the sphere. The other
shape factors we consider do not necessarily have a unique
grain shape. To find a size distribution that is easily comparable to the spherical case, we define a volume equivalent
radius for the aspherical grains as:
Table 2. Total mass of carbon grains formed
reff =
3V
4π
1/3
,
(6)
where V is the grain volume.
Figures 13a–d show the size distributions for all shape
factors and sticking coefficients. As the shape factor is increased, the size and number of larger grains also increases.
Mass (10−2 M⊙ )
c
(36π)1/3
5.4
6.0
7.0
9.0
12.0
γ = 1.0
γ = 0.1
γ = 0.01
γ = 0.001
4.95
4.94
4.92
4.89
4.82
4.64
4.93
4.91
4.89
4.86
4.76
4.52
4.89
4.87
4.84
4.79
4.65
4.27
4.79
4.75
4.70
4.62
4.34
3.32
We find that the largest maximum grain radii are formed
with c = 12.0 and γ = 1.0, and have a volume equivalent radius of almost 18 µm. On the other hand, the smallest maximum grain radii (∼ 0.008 µm) are formed with c = (36π)1/3
and γ = 0.001. As the sticking coefficient is reduced, the
maximum grain radii for each shape factor are also reduced,
and the number of smaller grains is increased. Even though
there is a much larger number of small grains than large
grains, the majority of the dust mass is contained within
intermediate sized grains. In Figure 14a (γ = 1.0), most of
the dust mass is contained in grains with volume equivalent
radii between 0.01 and 0.5 µm.
The masses of carbon grains formed (Figures 14a–d)
are dominated by the relatively small numbers of larger sized
grains as the shape factor is increased. However, as the sticking coefficient is reduced (Figures 14b–d), even the masses
of the grains formed with the largest shape factor become
dominated by the smaller sized grains. The total mass of
carbon grains formed for increased shape factors is given in
Table 2 and shown in Figure 15. Again, we notice that despite the difference in the size distribution due to different
microphysical parameters, the total mass of dust that condenses is fairly robust (within a factor of 1.5) and constitutes
almost the total amount of carbon that was left in the gas
phase after the creation of CO molecules.
Fallest et al.
8
(a)
❨ =1 0
(b)
❩ =0 1
(c)
❛ =0 01
(d)
❞ =0 001
.
0
1
s )
❳
.
❱5
logJ
(cm
3
·
❲
❯10
❚15
.
0
1
s )
❵
.
❫5
logJ
(cm
3
·
❴
❪10
c
❜
=(36 )
=5.4
c =6.0
1/3
=7.0
=9.0
c =12.0
c
c
c
❭15
4.8 5.0 5.2 5.4 5.6 5.8 6.0 4.8 5.0 5.2 5.4 5.6 5.8 6.0
enclosed mass (M ❬ )
enclosed mass (M ❝)
Figure 11. Maximum nucleation rates for all six shape factors and four sticking coefficients.
the discrepancy between the mass of dust grains predicted
in SN explosions and the observed dust mass in local type
II SNe (see Section 1 for a more thorough discussion and
references).
❡ =1 0
❢ =0 1
❣ =0 01
❤ =0 001
.
14
.
.
12
.
RV
10
8
6
4
2
0
5
6
7
8
9
10
11
12
shape factor c
Figure 17. RV values for 23 simulations. Not shown is RV = 41
for c = 12.0 and γ = 1.0.
5
DISCUSSION
We have shown that varying the microphysical properties
of dust grains has important effects on the condensation
times, nucleation rates, and size distributions of carbon dust
grains from type II supernova explosions. However, the total
mass of dust is only modestly affected by the changes in
the grain properties. An inadequate choice of the shape or
sticking coefficient is not therefore a viable explanation for
We find that a larger saturation is necessary to achieve
efficient nucleation with either a small sticking coefficient
(γ < 1) or for aspherical grains (c > (36π)1/3 ). For that reason, all our simulations show that the condensation time
grows when sticking coefficients less than unity or grain
shapes that are different from spherical are adopted. However, differences can be found in the nucleation rates and
final size distribution of the grains. Simulations with a low
value of the sticking coefficient show a delayed nucleation
but very high nucleation rates, thereby producing large
quantities of small grains. This is due to the fact that low
sticking coefficients inhibit both nucleation and grain growth
and, therefore, all the carbon remains in the gas phase until a high saturation level is reached. At that point, many
small grains are nucleated and the atomic carbon is quickly
depleted. Bianchi & Schneider (2007) found similar effects
when calculating dust nucleation with a sticking coefficient
of γ = 0.1. Asphericity of the grains, on the other hand,
inhibits nucleation but enhances grain growth. As a consequence, even if fewer grains are nucleated, they grow fast
and the result is a grain size distribution characterized by
less numerous, larger grains.
We find that the total mass of carbonaceous dust
formed remains relatively stable even with sticking coefficients as low as 0.001. For the spherical case only, we ex-
SN dust microphysics
600
❥
=(36 )
=5.4
c =6.0
c =7.0
c =9.0
c =12.0
c
time (days)
550
1/3
✐ =1.0
c
500
❦ =0.1
450
400
350
600
(b)
(a)
♠ =0.01
550
time (days)
9
♦ =0.001
500
450
400
350
(c)
(d)
4.8 5.0 5.2 5.4 5.6 5.8 6.0 4.8 5.0 5.2 5.4 5.6 5.8 6.0
enclosed mass (M ❧ )
enclosed mass (M ♥)
Figure 12. Condensation times for six shape factors and four sticking coefficients.
5
s)
c =(36
1/3
M
)
4
3
Mass (10
2
q
r
2
1
0
10
-9
10
-8
10
-7
10
-6
10
-5
♣
10
-4
10
-3
10
-2
10
-1
10
0
Figure 18. Total mass of dust formed for spherical carbonaceous
grains with sticking coefficients down to γ = 10−9 .
plored the possibility of even smaller sticking coefficients,
down to γ = 10−9 . We find that the mass of carbon dust
formed becomes significantly reduced for sticking coefficients
of γ = 10−7 and below (see Figure 18). With sufficiently low
values of sticking coefficient (below 10−8 ) there is virtually
no dust formation, but the required sticking coefficients seem
unphysically low.
In terms of the observable properties of the SNcondensed dust, we find that the quantity that is most af-
fected is the extinction curve (see Figure 16). Not surprisingly, simulations with small sticking coefficient (which, as
explained above, produce large amounts of small grains), result in a very steep extinction curve at far-UV wavelengths,
with RV values between 3 and 3.5, shown in Figure 17. We
make special note that the RV values for γ = 1.0 are relatively high, even though the size distribution of the grains
is consistent with that of interstellar grains (see Figure 8),
because we account for only carbon grains and that including other grains, such as silicates, could decrease the RV
values. On the other hand, simulations with very aspherical
grains and relatively high sticking coefficients produce larger
grains and, consequently, grey extinction curves. Without
the knowledge of the values of the sticking coefficient and of
the shape factor it is therefore impossible to predict the extinction curve of SN-condensed dust. This is a particularly
worrying conclusion since the extinction curve is relatively
easy to measure, even at high redshift, and could be used as
an observational constraint for the origin of dust in the various environments. For example, Maiolino et al. (2004) compared the extinction curve measured in a quasar at z = 6.2
to the extinction curve calculated using the dust model
by Todini & Ferrara (2001). They find that the data and
the theoretical prediction are in good agreement and conclude that the dust observed in SDSSJ104845.05+463718.3
is indeed condensed in SN explosions (see also Stratta et al.
(2007)). In light of our results, such conclusions need confirmation once a complete theory of SN dust nucleation is
obtained.
10
Fallest et al.
t =1.0
✉ =0.1
50
log(
dN/dr)
55
45
40
(a)
(b)
✇ =0.01
② =0.001
50
log(
dN/dr)
55
c =(36③) /
c =5.4
c =6.0
c =7.0
c =9.0
c =12.0
1 3
45
40
(c)
10
(d)
-3
10
-2
10
-1
✈m)
10
0
radius (
10
1
10
-3
10
-2
10
-1
①m)
10
0
10
1
radius (
Figure 13. Size distribution of carbon grains for six shape factors and four sticking coefficients.
The dust that condenses, however, is not the dust that
is ejected into the ISM. Dust produced in a CCSN has to
travel through the reverse shock before being released into
the ISM. The reverse shock can destroy most of the dust,
in particular the smaller dust grains (Draine & Salpeter
1979; Nozawa et al. 2007; Nath, Laskar, & Shull 2008;
Silvia, Smith, & Shull 2010). If astrophysical dust formation is indeed characterized by small values of the sticking coefficient, it is likely that the amount of dust formed
is reduced significantly by the reverse shock. Conversely, increased shape factors allow for the formation of larger grains
which would survive shock processing. The microphysical
properties of dust grains can therefore affect the mass of dust
that is injected in the interstellar medium, even though they
affect only marginally the dust that is condensed during the
early stages of the explosion.
Grain nucleation with non-spherical shapes may be
more complicated than we considered here. We have assumed that the shape factors of the grains do not change
as the grains grow. Since we assume here that grains grow
by the addition of monomers, the shape of small clusters
can change as monomers attach, in turn altering the shape
factor. Another route to take could be to nucleate grains at
an arbitrary shape factor and then allow the grains to grow
into spheres. For example, one may turn grains into spheres
when the number of monomers is larger than a given value.
In this case, increased shape factors may not lead to such
large grains as we show in our size distributions (Figure 13).
This work is purely a parametric study that aims to
show that the microphysical properties of the grains are important, but not to point to any specific values of γ and
c to be used in nucleation calculations. Therefore, we have
chosen to neglect important factors that need be considered
in a complete SN dust nucleation model. These include the
choice of progenitor model, the presence of other dust species
(see Nozawa et al. (2003)) as well as charged molecules
that may interfere with carbonaceous dust condensation,
and the destruction of CO molecules, due to photodissociation or collisons with fast electrons and charged particles
(Petuchowski et al. 1989; Lepp, Dalgarno, & McCray 1990;
Liu, Dalgarno, & Lepp 1992; Clayton, Deneault, & Meyer
2001), that can inject additional carbon atoms into the available monomer concentration (see Todini & Ferrara (2001);
Bianchi & Schneider (2007)).
A zero-metallicity 20 M⊙ CCSN progenitor model was
chosen because SNe at high redshift are expected to have
zero metallicity. In general, the relative abundances of major
elements in the He core are not significantly different among
the SN models with different metallicities. Thus, the species
of dust formed do not depend on the metallicity of the SN
progenitor star (Nozawa et al. 2010). Additionally, the gas
density and temperature in the He core are almost independent of the progenitor mass and metallicity as long as the
kinetic energy of the explosion is the same (Nozawa et al.
2003). Therefore, SN models with non-zero metallicities, or
with different progenitor masses, are expected to show similar effects on carbon grains formation as we see here. This
means that the mass of carbon dust formed in the SN ejecta
SN dust microphysics
⑧ =1.0
2
11
⑨ =0.1
log(
dM/dr)
0
⑦2
⑥4
⑤6
④8
(b)
(a)
❺ =0.01
2
❼ =0.001
log(
dM/dr)
0
❹2
❸4
c =(36❽) /
c =5.4
c =6.0
c =7.0
c =9.0
c =12.0
1 3
❷6
❶8
(c)
10
(d)
-3
10
-2
10
-1
⑩m
radius (
10
0
)
10
1
10
-3
10
-2
10
-1
❻m
radius (
10
0
10
1
)
Figure 14. Mass distribution of carbon grain for six shape factors and four sticking coefficients.
is rather insensitive to changes in the sticking coefficient and
shape factor and is purely determined by the mass of carbon
atoms available for dust formation in the He layer. However,
to confirm such expectations, additional progenitor models
would need to be investigated.
Perhaps more important than the choice in progenitor
model is the dissociation process of CO molecules. In
this paper we assumed the formation of CO molecules
to be complete and considered only the condensation process of C grains in the He layer where C/O
> 1. In the expanding ejecta, CO molecules could
be destroyed through interactions with fast electrons
from radioactively decaying 56 Co and charged particles such as He+ and Ne+ (Petuchowski et al. 1989;
Lepp, Dalgarno, & McCray 1990; Liu, Dalgarno, & Lepp
1992;
Clayton, Liu, & Dalgarno
1999;
Kwong, Chen, & Fang 2000; Clayton, Deneault, & Meyer
2001; Deneault, Clayton, & Meyer 2006), allowing for more
free carbon (and oxygen) atoms to be available for grain
formation than we consider here. However, the number
abundance of silicon atoms is too small for most of the
enclosed mass regions (M= 4.93–6.21 M⊙ ) in this work
(see Figure 3), so that the formation of SiC and silicate
gains cannot be expected. Therefore, the dissociation of
CO molecules due to interactions with He+ only results
in a slight enhancement of the final mass of carbon grains
and never changes our conclusion on the dependence
of formation process of C grains on the microphysical
properties.
On the other hand, Clayton, Liu, & Dalgarno (1999)
and Deneault, Clayton, & Meyer (2006) show that CO dissociation enables carbon dust grains to form even in O-rich
layer where C/O < 1. Given that the abundance of silicon
atoms in the O-rich layer is higher than in the He layer, the
formation of SiC grains could be expected there. However,
as discussed in Nozawa et al. (2003), even if free carbon and
silicon atoms coexist abundantly, the nucleation theory does
not predict the formation of SiC grains. The formation process of large SiC grains as appeared in presolar grains, as
well as formation efficiency of molecules is to be pursued in
more sophisticated studies of dust formation. Furthermore,
in the O-rich layer, the formation of silicate grains is also
feasible. Bianchi & Schneider (2007) show that the formation of silicate grains is more sensitive to changes in sticking
coefficient than carbon grains, and the mass of silicate grains
formed can be reduced for even γ = 0.1. The inclusion of
silicate grains could affect the resulting extinction curves.
We have adopted the thermodynamic approach for this
study because it involves the simplest nucleation equations,
however, use of the kinetic theory of nucleation should be
considered in the future. The kinetic theory still needs to
take the sticking coefficient into account, but makes consideration of an evolving shape factor unnecessary, because the
shape of the grain from a complex solid (for a few molecules)
to a sphere (for ∼ 100 molecules) is intrinsically taken into
account.
12
Fallest et al.
10
A/AV
8
6
c =(36❿) /
c =5.4
c =6.0
c =7.0
c =9.0
c =12.0
❾ =1.0
1 3
➀ =0.1
4
2
(a)
0
10
(b)
➄ =0.01
➈ =0.001
A/AV
8
6
4
2
0
(d)
(c)
0
2
4
6
➃
1/➁ (➂m )
1
8
0
2
4
6
➇
1/➅ (➆m )
1
8
Figure 16. Extinction curves for each sticking coefficient and shape factor.
ACKNOWLEDGEMENTS
We would like to thank the reviewer for their insightful comments and suggestions for the improvement of this work. We
would also like to thank Raffaella Schneider and Andrea Ferrara for their comments. Thanks go to the Institute for the
Physics and Mathematics of the Universe (IPMU), University of Tokyo, Kashiwa, Japan, for their generous hospitality, and to the NSF East Asia and Pacific Summer Institute
(EAPSI) Japan 2010 program (award #1015575) and the
Japanese Society for Promotion of Science for their generous support. This work has also been supported in part
by World Premier International Research Center Initiative,
MEXT, Japan.
REFERENCES
Barlow M. J., et al., 2010, A&A, 518, L138
Becker R., & Döring W., 1935, Ann. Phys., 24, 719
Bianchi S., Schneider R., 2007, MNRAS, 378, 973
Bromm V., Coppi P. S., Larson R. B., 2002, ApJ, 564, 23
Cherchneff I., 2010, ASPC, 425, 237
Cherchneff I., Dwek E., 2009, ApJ, 703, 642
Cherchneff I., Dwek E., 2010, ApJ, 713, 1
Clayton D. D., Deneault E. A.-N., Meyer B. S., 2001, ApJ,
562, 480
Clayton D. D., Liu W., Dalgarno A., 1999, Sci, 283, 1290
Deneault E. A.-N., Clayton D. D., Meyer B. S., 2006, ApJ,
638, 234
Donn B., Nuth J. A., 1985, ApJ, 288, 187
Douvion T., Lagage P. O., Cesarsky C. J., 1999, A&A, 352,
L111
Draine B. T., Salpeter E. E., 1979, ApJ, 231, 438
Dwek E., Galliano F., Jones A. P., 2007, ApJ, 662, 927
Elmhamdi A., et al., 2003, MNRAS, 338, 939
Elvis M., Marengo M., Karovska M., 2002, ApJ, 567, L107
Feder J., Russell K. C., Lothe J., Pound G. M., 1966, Adv.
Phys., 15, 111
Hoyle F., Wickramasinghe N. C., 1970, Nature, 226, 62
Kashchiev D., 2000, Nucleation: Basic Theory With Applications, Butterworth-Heinemann, Oxford
Keith A. C., Lazzati D., 2011, MNRAS, 410, 685
Kotak R., et al., 2009, ApJ, 704, 306
Kozasa T., Hasegawa H., 1987, PThPh, 77, 1402
Kozasa T., Hasegawa H., Nomoto K., 1989, ApJ, 344, 325
Kozasa T., Hasegawa H., Nomoto K., 1991, A&A, 249, 474
Kozasa T., Nozawa T., Tominaga N., Umeda H., Maeda
K., Nomoto K., 2009, ASPC, 414, 43
Kwong V. H. S., Chen D., Fang Z., 2000, ApJ, 536, 954
Lazzati D., 2008, MNRAS, 384, 165
Lepp S., Dalgarno A., McCray R., 1990, ApJ, 358, 262
Li A., Greenberg J. M., 2003, ssac.proc, 37
Liu W., Dalgarno A., Lepp S., 1992, ApJ, 396, 679
Lucy L. B., Danziger I. J., Gouiffes C., Bouchet P., 1989,
LNP, 350, 164
Maiolino R., Schneider R., Oliva E., Bianchi S., Ferrara A.,
Mannucci F., Pedani M., Roca Sogorb M., 2004, Nature,
431, 533
SN dust microphysics
Mathis J. S., Rumpl W., Nordsieck K. H., 1977, ApJ, 217,
425
Meikle W. P. S., et al., 2007, ApJ, 665, 608
Morgan H. L., Edmunds M. G., 2003, MNRAS, 343, 427
Nomoto K., Tominaga N., Umeda H., Kobayashi C., Maeda
K., 2006, NuPhA, 777, 424
Nozawa T., Kozasa T., Umeda H., Maeda K., Nomoto K.,
2003, ApJ, 598, 785
Nozawa T., Kozasa T., Habe A., Dwek E., Umeda H., Tominaga N., Maeda K., Nomoto K., 2007, ApJ, 666, 955
Nozawa T., et al., 2008, ApJ, 684, 1343
Nozawa T., Kozasa T., Tominaga N., Maeda K., Umeda
H., Nomoto K., Krause O., 2010, ApJ, 713, 356
Nath B. B., Laskar T., Shull J. M., 2008, ApJ, 682, 1055
Petuchowski S. J., Dwek E., Allen J. E., Jr., Nuth J. A.,
III, 1989, ApJ, 342, 406
Rho J., et al., 2008, ApJ, 673, 271
Sibthorpe B., et al., 2010, ApJ, 719, 1553
Silvia D. W., Smith B. D., Shull J. M., 2010, ApJ, 715,
1575
Stratta G., Maiolino R., Fiore F., D’Elia V., 2007, ApJ,
661, L9
Sugerman B. E. K., et al., 2006, Sci, 313, 196
Temim T., Slane P., Reynolds S. P., Raymond J. C.,
Borkowski K. J., 2010, ApJ, 710, 309
Todini P., Ferrara A., 2001, MNRAS, 325, 726
Umeda H., Nomoto K., 2002, ApJ, 565, 385
Valiante R., Schneider R., Bianchi S., Andersen A. C., 2009,
MNRAS, 397, 1661
Wooden D. H., Rank D. M., Bregman J. D., Witteborn
F. C., Tielens A. G. G. M., Cohen M., Pinto P. A., Axelrod
T. S., 1993, ApJS, 88, 477
Wooden D. H., 1997, AIPC, 402, 317
Zhukovska S., Gail H.-P., Trieloff M., 2008, A&A, 479, 453
13