You are on page 1of 11

View Article Online / Journal Homepage / Table of Contents for this issue

PAPER www.rsc.org/faraday_d | Faraday Discussions

Polycyclic aromatic hydrocarbons, carbon nanoparticles


and the diffuse interstellar bands
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

W. W. Duley

Received 17th November 2005, Accepted 22nd December 2005


First published as an Advance Article on the web 23rd May 2006
DOI: 10.1039/b516323d

Observational data on the appearance and properties of the diffuse interstellar bands
(DIBs) are reviewed in the context of a model in which the proposed carriers of these
bands are large carbon molecules and carbon nanoparticles containing between 30 and
several hundred carbon atoms. The abundance of these carriers, as estimated from the
observed strengths of the DIBs, place strong constraints on their rates of formation and
destruction, and suggest that the strongest bands, including that at 4428 Å, could be
produced via the decomposition of larger carbon particles, possibly those particles that
have been postulated to be the source of the 2175 Å extinction feature. Such particles are
of mixed sp2 and sp3 carbon composition, with sizes between that of large molecules and
small macroscopic solids. Any description of their characteristics must combine aspects of
molecular and condensed matter physics, and this is incorporated in the present
discussion. I discuss recent experimental and theoretical data related to these matters.

Introduction
Few problems in astrophysics have captured the attention of astronomers and have been the subject
of as much speculation as that of the diffuse interstellar bands (DIBs). The observational situation
has been well summarized by Herbig1 and in the reports in the proceedings edited by Tielens &
Snow2. A summary and critique of possible carriers of these bands has been given by Snow.3 There
now seems to be general agreement that the DIB carriers are molecular,4 but indications are that
these molecules are relatively large ( Z 10 atoms) primarily because of the requirement that the
carriers be stable in diffuse interstellar clouds, but also because of the linewidth and contour of the
bands themselves.5
Generally, the DIBs are seen only in absorption and there is no indication of re-emission
associated with any of these features, except perhaps in the Red Rectangle where a few similar
bands have been detected.6–9 This detection occurs in a region near the bicone interface where
interstellar material is subject to dissociation10 suggesting that some of the DIB carriers can be
produced by sputtering, photodissociation or chemical reaction involving a precursor molecule or
dust grain. In a discussion of the carrier of the 5800 Å emission band in the Red rectangle and the
associated 5797 Å DIB, Duley11 concludes that the relevant molecule has r30 C atoms and is
formed in situ. The photodissociation probability for this molecule was estimated to be r105 per
absorbed UV photon, while the luminescence efficiency was found to be Z = 102–103. Such
estimates, while very tentative, together with analysis of environmental factors yield some
constraints on the type of molecule that could produce this band and its relation to interstellar
chemistry.
Spectroscopy provides other constraints and will, in the end, be the defining factor in the
identification of the carriers of the DIBs. The classes of molecule currently under consideration
include carbon chains,12–14 fullerenes and fulleranes,15 polycyclic aromatic hydrocarbon (PAH)

Physics Department, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1. E-mail:
wwduley@uwaterloo.ca

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 415
cations16–19 and polycyclic aromatic nitrogen neterocycles (PANH).20 All are carbon molecules
View and
Article Online
share the following general characteristics:
(1) A wide, but limited, range of individual molecules are present, with the most abundant
molecules containing as much as 1% of available carbon in interstellar clouds.21
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

(2) The relative concentration of individual carriers is sensitive to UV intensity and cloud density.
(3) Photodissociation rates must be less than the timescale for replacement of carriers in the
diffuse interstellar medium (DISM).
(4) The DIB carriers are under-abundant in circumstellar shells and in reflection nebulae and do
not appear in carbon-rich proto-planetary nebulae.
(5) There is little correlation between carrier abundance and extinction apart from E(B  V)
although the abundance of individual molecules as evidenced by their DIB features may show
limited correlation with each other.1
The formation and destruction of the DIB carriers is central to the problem of their identification,
and yet has not been dealt with in any general way although there are a number of discussions of the
stability of large carbon molecules in interstellar clouds.5,22–28 There have also been extensive
studies of reactions involving the formation and dispersal of large carbon molecules in dense
interstellar clouds,29,30 circumstellar shells and AGB stellar envelopes.31 This paper examines the
question of the formation and destruction of representative DIB carriers in diffuse clouds and how
these processes can be related to observation. As a specific example, the properties of the carrier of
the strongest DIB at 4428 Å are discussed in detail.

Formation and destruction rates


The density, n(X), of a molecular DIB carrier in a cloud with n hydrogen nuclei will be determined
as a balance between the rate at which the carrier is formed and that of destruction. The relevant
reactions will take the form
(A,A+) + B - X (1)
(A.A ) + hn - X
+
(2)
(A,A+) + c.r. - X (3)
X + hn - X +
+e (4)
X + e - X + hn
(5)
(X,X ,X ) + hn - products
+ 
(6)
X +
+ e - X + hn (7)
(X,X+,X) + c.r. - products (8)
where c.r. is cosmic ray. Rates of these reactions will depend in detail on the type and size of the
molecule involved and the nature of the reaction products. Measured rates are generally unavail-
able, but estimated values are available from a number of sources. These are listed for a
representative sample of large carbon molecules in Table 1. Rates for photodissociation and
photoionization are from Allain et al.23,24 while rates for cosmic ray decomposition have been taken
to be 2.3  1017 Nc s1 , where Nc is the number of carbon atoms in the molecule.32 The rate of
electron–ion recombination has been derived from observations of PAH emission in the diffuse
interstellar medium by Chan et al.33 This value is smaller than previous estimates but is required to
account for the high PAH+/PAH ratio implied by observation.34 Rates for addition reactions (1)
are listed for H atoms combining with neutral and ionized precursors to form a new molecule X,
X+, respectively, while C and C+ reaction rates are taken as 5.0  1010 cm3 s1 for all reactions.
The rates of reactions 6 and 8 are assumed to be the same for neutral and ionic molecules.

Relative abundance of DIB carriers


A reasonable assumption is that the molecules that are the source of the DIBs represent a
component of interstellar matter in the size range between small molecules such as CH, C2, etc.

416 | Faraday Discuss., 2006, 133, 415–425 This journal is 


c The Royal Society of Chemistry 2006
View Article
Table 1 Representative rates of reactions 1–8 for pyrene, C16H10, coronene, C24H12, ovalene, C32HOnline
14 and a
50 C atom PAH molecule

Reaction Pyrene Coronene Ovalene 50 C atom PAH Reference


Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

a
1 (A + H) 1.7(11)
(A+ + H) 1.0(9) a

(A + C+, A+ + C) 5.0(10)
b
2 (-C2H2) 7.1(11) 4.6(11) 4.9(12) 3.6(18)
b
(-H) 2.0(9) 1.5(9) 1.8(10) 2.3(16)
c
3 3.7(16) 5.6(16) 7.5(16) 1.2(15)
b
4 3.0(9) 6.0(9) 7.0(9) 8.0(9)
d
5 4.8(7) 5.9(7) 6.8(7) 8.5(7)
b
6 7.1(11) 4.6(11) 4.9(12) 3.6(18)
e
7 3.3(7)
c
8 3.7(16) 5.6(16) 7.5(16) 1.2(15)
a 24 b 23 c 32 d
References: Allain et al. Allain et al. Estimated from Duley & Williams 1984. Allamandola
et al.22 e Chan et al.33

and classical dust particles containing in excess of E104 atoms. The observed wavelength range for
the DIBs (4066–13 175 Å)1 constrains the size of these molecules to r103 carbon atoms19,35 unless
they are components of somewhat larger particles. The abundance of a specific DIB carrier will be
determined by the relative rate of formation and destruction according to reactions 1–8 while
variations in these rates can be expected to lead to changes in the strength of individual DIBs. It is
apparent that the selectivity of these reactions, together with the way in which carriers are
destroyed, must be highly effective in reducing the number of carriers to a finite number of species
from the myriad of large molecules that are possible. For example, Salama et al.19 note that there
are 1.2  106 possible isomers containing between 35 and 49 carbon atoms while only a few hundred
molecules are required to give rise to all known DIB transitions.
This high degree of selectivity suggests a model in which only a few very stable carbon-based
molecules exist in the DISM and that the DIB carriers are chemical derivatives of these molecules
that are continuously being created and destroyed. A simple example would involve the reaction
XH+ + hn - X+ + H (9)
where XH+ is the stable precursor and X+ is the DIB absorber. XH+ would be reformed via
X+ + H - XH+ (10)
+
while X could also be photodissociated
X+ + hn - products (11)
If X+ is still hydrogenated then the photodissociation rates in reactions 9 and 11 should be similar
and the relative rates of reactions 9 and 10 will determine the level of hydrogenation in a series of
radicals or radical ions deriving from a fully hydrogenated precursor molecule. The solution to this
equilibrium under a variety of excitation conditions has been studied in detail.23,24,26–28 It is found
that various degrees of hydrogenation are possible in HI clouds for neutral molecules and cations
having fewer than 40 carbon atoms but that larger molecules are fully hydrogenated. In a high
excitation object such as NGC 7027, molecules with fewer than 40 C atoms are almost fully
dehydrogenated and molecules with fewer than E15 C atoms will be destroyed.
The total equivalent width of all known DIBs in a well studied object such as HD 183143 is Wl =
26.9 Å1 and the strongest of these at 4428 Å has Wl = 2.49 Å while the weakest features have Wl o
102 Å. The column density N(DIB) (cm2) of band carriers can be obtained from Spitzer.36
1:13  1020 W
NðDIBÞ ¼ ð12Þ
l2 f

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 417
Table 2 Relative abundance N(DIB)/N(H) of selected DIB carriers View in Article Online
HD 183143 Observational data from Herbig.1 The DIBs chosen are typical
of strong (4428 Å), medium (5797 Å) and weak (5795 Å) features with
Wl = 2.49, 0.238 and 0.062 Å, respectively. f is oscillator strength
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

DIB N(DIB)/N(H)
4428 Å 1.9(9)/f
579.7 1.1(10)/f
579.5 2.8(12)/f

Table 3 Fractional amount of available carbon required in the carriers of representative DIBs in HD 183143
for various molecules. N(C)/N(H) = 2.15  104

4428 Å 5797 Å 5795 Å


PyreneX, C16 1.4(4)/f 8.2(5)/f 2.1(7)/f
CoroneneX, C24 2.1(4)/f 1.2(4)/f 3.1(7)/f
OvaleneX, C32 2.8(4)/f 1.6(4)/f 4.2(7)/f
50 C atom PAH 4.4(4)/f 2.6(4)/f 6.5(7)/f

where f is oscillator strength, and l and W are in Å. In HD 183143 eqn. (12) gives abundances
relative to hydrogen ranging from N(DIB)/N(H) = 1.9  109/f for the 4428 Å band to o1012/f
for the weakest features. A summary of N(DIB)/N(H) for representative DIBs in HD 183143 is
given in Table 2. The fraction of carbon tied up in each DIB carrier for several sizes of molecule is
listed in Table 3. This shows that the strongest DIB could be produced by a transition with f = 0.1
in a 16 atom molecule containing as little as 0.14% of available carbon. With f = 0.1, the weakest
DIBs could be produced by a similar molecule containing o2  106 of available carbon.
Since most molecules of this size will have spectral features with f Z 103 in the 4000–10 000 Å
range, the fact that only several strong DIBs, rather than many such features, are observed is further
evidence for selectivity in the formation and destruction of these species. On the other hand, there
does appear to be a forest of very weak lines suggesting that numerous additional molecules are
present in diffuse clouds at the N(DIB)/N(H) = 109–1011 level. A deep search for candidate
molecules such as C4 and C5 in z Oph has, however, been negative.37

Formation and destruction mechanisms


The low densities (n r 500 cm3) and high UV flux in diffuse interstellar clouds precludes the in situ
synthesis of large molecules32 implying that the molecules responsible for the DIBs have either been
imported into these objects, created there through the destruction of larger particles, or formed by
the action of photons or H atoms on larger stable molecules. The first scenario has been discussed
by Bettens & Herbst30 in a dispersive model in which large non-PAH hydrocarbon molecules are
formed in dense clouds that then evolve into diffuse objects. They find that seed hydrocarbon and
carbon cluster molecules containing up to 64 C atoms can be transferred from a dense to a less dense
environment, mostly in the form of fullerenes and fulleranes. It was suggested that large molecules
adsorbed on dust grains and gradually desorbed in diffuse clouds would be an effective way to form
the requisite concentration of DIB carriers. This scenario may not, however, be relevant as PAH
molecules are more closely linked to the regions around evolved stars.22
It is unrealistic to attempt to incorporate all of these effects into a single model, particularly since
so little is known about the chemical composition of any DIB carrier. Instead, a simple hierarchical,
or ‘‘top-down’’, model will be used in which it will be assumed that the carbon atom structures of
molecules or molecular fragments responsible for the DIBs were once components of larger
particles. This scenario is supported by recent observations of the spatial distribution of molecules
and small grains in a variety of objects.38–41 In this model, the survivability of a DIB carrier under
DISM conditions derives from the number of C atoms in the carrier, and the bonding of these
atoms. This approach obviously ignores the effect of hydrogenation and dehydrogenation on the

418 | Faraday Discuss., 2006, 133, 415–425 This journal is 


c The Royal Society of Chemistry 2006
abundance of specific DIB carriers, but it provides a basis for estimating the size of Article
View putative DIB
Online
carriers. For example, this approach assumes that, if a particular DIB arises from a pyrene radical
cation such as C16H8+, then the primary limitation on the abundance of this carrier will be
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

determined by the survivability of the C16 core of this molecule.


The formation of DIB carriers via decomposition of a precursor molecule is then postulated to
follow the sequence
XY + (hn, shock, c.r.) - X + Y (13)
X + hn - products (14)
giving n(X)/n(XY) = k13/k14 where k13 (cm3 s1) is the rate constant for reaction 13. Assuming an
equivalence between column and space density for a given line of sight i.e. N(X)/N = n(X)/n, n(X)/
n Z 109 for the carriers of the strongest DIBs (Table 2). Then
nðXYÞ 109 k14
¼ ð15Þ
n fk13
If b(XY) is the fraction of carbon in XY, and NC is the number of carbon atoms in XY, then eqn.
(15) can be rewritten
4:65  106 NC k14
bðXYÞ ¼ ð16Þ
fk13
Solutions to eqn. (16) are summarized in Table 4, using data from Table 1. For simplicity in
notation, all derivatives having four fused aromatic rings as in pyrene will be referred to as
‘‘pyreneX’’. When formation occurs by photodissociation of a larger precursor, k13 is taken to be
that of the next largest molecule in Table 1, e.g. coroneneX would yield pyreneX. The result is that
photodissociation can create only molecules having NC r 24 atoms at the required rate, since b is
limited to r0.3 in the DISM.47 However, this does not exclude the possibility of the photodesorp-
tion of small molecules from larger particles.30 It is apparent that, a requirement for the appearance
of a DIB in the DISM is that the concentration of a precursor molecule such as coroneneX must be
maintained. The desorption of molecules from larger particles by cosmic ray impact would be
ineffective in maintaining the concentration of molecules with NC r 32 at the requisite level because
of the high photodissociation rate of these smaller species. This rate decreases dramatically for
larger molecules (Table 1) becoming less than the rate of cosmic ray excitation. This suggests that
molecules having in excess of 50 carbon atoms could be produced by cosmic ray desorption from
larger particles although the yield (molecules desorbed per cosmic ray impact) is uncertain. A
number of studies have demonstrated that large intact organic molecules can be readily desorbed
from the surface of solids under MeV ion impact.42,43 A schematic representation of the coroneneX
to pyreneX conversion is shown in Fig. 1.

Table 4 Fraction of carbon, b(XY), in a precursor molecule required to maintain a


potential DIB carrier X at a relative abundance consistent with the strength of the most
intense DIBs in DISM clouds (eqn. (16)). Rates are from Table 1 and it is assumed that
the precursor for a molecule X is the molecule with the next largest number of C atoms.
In all examples except that for formation of 50 atom PAH via cosmic ray impact, the
dominant destruction mode for X is photodissociation. Only values of b r 1 are of
physical significance

f b (XY)
Molecule, X Photodissociation Cosmic ray Shock
PyreneX 1.7(4) 1.4(1) 2.5(1)
CoroneneX 1.4(3) 9.1(1) 2.1(1)
OvaleneX 3.2(2) 9.4(1) 3.5(2)
50 atom PAH 9.0(4)a 4.7(4)
a
Destruction by cosmic ray impact.

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 419
View Article Online
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

Fig. 1 Illustrating the hierarchical connection between coroneneX (left structure) and pyreneX (right
structure). The arrow represents the effect of a shock on coroneneX. Other fragments are not shown. The open
hexagons are non-aromatic rings.

To estimate the possible role of interstellar shocks in the formation of the DIB carriers, it will be
assumed that the interval between weak (10 km s1) shocks is 106 years.44 The passage of a low
velocity shock can lead to the shattering of grains as well as to sputtering35 resulting in the release of
molecular components.46 The overall effect is to transform large particles and molecules into
smaller species. This can be modeled in the present context as the conversion of ovaleneX to
coroneneX, coroneneX to pyreneX, etc. Values of b(XY) for all candidate molecules (Table 4) are
compatible with the available abundance of carbon in diffuse clouds suggesting that shocks may be
a significant source of DIB carriers if the precursor molecule, XY, can be maintained at the required
density in the DISM.
If M is a large carbon particle and Z is the number of molecules of XY released from this particle
per shock, then

Z n(M) g = n(XY) LXY (17)


1 1
where LXY (s ) is the loss rate of XY, and g (s ) is the rate of shock excitation. It is implicitly
assumed that Z can represent only a small fraction of the possible molecules contained in M. The
loss of carbon from M will be compensated, at least in part, through accretion of C and C+.

M + C, C+ - MC (18)

that will occur at the kinetic rate23,24 so that the timescale for accretion, ta = (k18 n(C))1 where
n(C) is the gas-phase carbon density. The number of C atoms accreted in the interval between
shocks is then (gta)1.
Taking LXY = 4.6  1011 s1 (Table 1, coroneneX), and with g = 3.17  1014 s1 one obtains
n(M) = 1.45  103 n(coroneneX)/Z = 2.15  104 b (M) n/NM where NM is the number of carbon
atoms in M. If a pyreneX-like molecule is the source of one of the stronger DIBs and this molecule is
formed through the photodissociation of coroneneX, then b(coroneneX) = 1.7  104/f (Table 4)
and n(coroneneX)/n = 1.5  109/f cm3. The resulting constraint is that NM = 102 bZf which
precludes the proposed sequence M - coroneneX - pyreneX as NM must c24Z while f o 1 and
b o 0.3. This tells us that, while the required abundance of a pyreneX-like carrier of a strong DIB in
a diffuse cloud can be obtained through the photodissociation of a coroneneX-like precursor, the
density of this precursor cannot be maintained through shock-induced decomposition of a larger
particle. A similar analysis shows that the sequence M - ovaleneX - coroneneX, where a
coroneneX-like molecule is a carrier of one of the strongest DIBs, is also invalid as the density of
ovaleneX cannot be maintained at the requisite level. This does not preclude the possibility that
carriers of the weaker DIBs are formed in this way.
Large molecules are much less subject to destruction by photodissociation (Table 1) but may still
be destroyed, or modified, by shocks. When shocks dominate in both the formation and destruction
of a DIB carrier, g E LX in eqn. (17) and n(M) = n(X)/Z. For the DIB at 4428 Å this becomes n(M)
= 2.15  104 b(M)n/NM = 1.9  109 n/fZ and NM = 1.13  105 b(M)Zf. In this case, a number of
solutions are allowed satisfying both abundance and intensity constraints. For example, with

420 | Faraday Discuss., 2006, 133, 415–425 This journal is 


c The Royal Society of Chemistry 2006
f = 0.1, and b = 0.1, NM = 1.13  103 Z and the DIB at 4428 Å could be produced by a molecule
View Article Online
containing 50 C atoms from a precursor having NM E 104 carbon atoms if Z = 10 molecules per
shock. This would correspond to a mass loss of 5% per shock, a loss rate that is in agreement with
theoretical estimates.46
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

For a particle containing 104 carbon atoms the rate constant for redeposition of carbon, k18 E
108 cm 3 s1 and each particle will accrete E30 n carbon atoms in the interval between shocks,
assuming n(C,C+) = 104 n cm3. This suggests that an equilibrium may exist in diffuse clouds in
which carbonaceous material deposited on larger (E104 atom) particles between shocks is returned
as DIB carriers after the passage of a shock.
This analysis, based on the abundance of carbon in interstellar clouds, the sensitivity of smaller
molecules to photodissociation and the observational constraints provided by the strength of the
most intense DIBs, implies that the strongest DIBs are likely associated with the largest molecules. It
appears that the carriers of the strongest bands have NC Z 40 since only molecules in this size
range are sufficiently resistant to photodissociation in the DISM.23,24,26–28

Discussion
The presence of a DIB carrier in the DISM is made possible only by achieving a balance between
formation and destruction in an equilibrium that allows the abundance of this carrier to increase to
the point where its strongest electronic transitions produce an optical depth Ac Z 0.01. The details
of the reactions that create a specific carrier and the rate at which this carrier is destroyed are
uncertain, but these can still be defined in general terms. This leads to the conclusion (Section 4) that
only carriers with greater than E40 carbon atoms can achieve equilibrium densities sufficient to
produce the strongest DIBs. The carbon core of smaller molecules, such as pyrene and its simple
derivatives, while attractive candidates for some of the DIBs,45,48 cannot be produced at a sufficient
rate under DISM conditions to compensate for photodissociation if it is assumed that these
molecules are separate gas-phase species. However, this constraint can be relaxed if these molecules
are constituents of somewhat larger species. Spectroscopically, electronic spectra of large hydro-
carbon molecules have been found to reproduce those of any core aromatic structures.49
In the defected graphite (DG) model,50 a significant fraction of interstellar carbon is contained in
nanoparticles ranging in size from dehydrogenated pyrene (16 carbon atoms) to clusters containing
several hundred carbon atoms. These particles are suggested to be the source of the 2175 Å
extinction feature which arises as a p–p* plasmon.51 In this model, smaller components gain
stability by incorporation in somewhat larger particles, and a similar mechanism may exist for the
DIB carriers. An example of the way in which this can occur is shown in Fig. 1, where a larger
particle separates into two independent smaller aromatic domains as defects are introduced. The
cluster size distribution will depend on the number of atoms in the original nanoparticle and is
highly constrained when NC is small. For example, a particle with NC = 50 can contain no more
than two pyreneX structures or one substituted coroneneX molecule. It is proposed that a number
of the DIB carriers are such separated aromatic domains in carbon nanoparticles containing up to
several hundred atoms. The proportion of small clusters in this kind of particle is expected to
depend on the UV absorption rate, leading to dehydrogenation and the formation of CQC bonds,
as well as to the rate of rehydrogenation. The spectrum of individual aromatic clusters should be
similar to that of derivatives of the corresponding core aromatic carbon structure. This effect has
been observed experimentally.49,52 For example, Colmsjo et al.49 have shown that sharp-line spectra
of a variety of hydrogenated PAH’s are determined by transitions of the core aromatic entity, a
‘‘molecule in a molecule’’ scenario.
The survival of small molecules in the harsh environment of the DISM could then be enhanced,
since delocalization of vibrational excitation produced in one aromatic region redistributes energy
over the entire nanoparticle effectively inhibiting the rate of decomposition. This is similar to the
model proposed by Allain et al.24 Energy sharing strongly affects the decomposition of a small
molecule as the decomposition rate decreases dramatically when NC increases.23,24 For example, the
photodissociation rate of an independent pyreneX component like C16 in the DISM is calculated to
be 7.1  1011 s1 while that of a 50 atom PAH is 3.6  1018 s1 (Table 1). A C16 component
within a 50 atom particle would then have a much higher survival rate under these conditions. It
seems likely, from the studies of Colmsjo et al.49 and others, that electronic transitions associated

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 421
with this component could be observed while the structure has enhanced View stability through
Article Online
incorporation in a somewhat larger particle.
Applying the DG model to a nanoparticle consisting of a mixture of aromatic PAH-like
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

molecules permits a description of the spectroscopic properties of carbon particles in the transition
region between single PAH-like species and graphite. This question has been discussed by Salama
et al.19 in the context of the PAH model. The DG model can also be utilized to describe the
properties of HAC as a p-plasmon resonance has been observed at 5.1 eV in as-deposited solids53
shifting to the 6.5 eV peak of graphite on dehydrogenation.54 DG also retains a bandgap that can be
identified with the UV absorption edge in PAH molecules.55 The bandgap energy, Eg E 5.8/(NR)0.5
where NR E 0.4 NC is the average number of aromatic rings in the material. A nanoparticle with 50
atoms can then consist of a single aromatic domain or several smaller aromatic components joined
together by sp2 or sp3-bonded species. The connectivity of this structure will determine if the
p electron resonance extends across the particle or is localized at one or more locations in individual
aromatic clusters. Fig. 1 shows a schematic representation of two such possible structures in a larger
(91 ring) nanoparticle.64 In Fig. 2a, the removal of 7 C atoms accompanied by the formation of CH
bonds adjacent to the vacancy retains the original aromatic sp2 coordination. The particle continues
to look like a single large aromatic structure.
In Fig. 2b, 7 atoms have been removed but hydrogenation is more complete and carbon atoms
adjacent to the vacancy are sp3-bonded. This has destroyed the overall p-electron connectivity and
the single large aromatic domain has been converted into two smaller structures. The bandgap
energy of each of the two small structures will in general be larger than that of the single domain in
Fig. 2a. It is apparent that smaller fused ring structures analogous to pyrene and coronene could
appear in this way as components of a larger nanoparticle. Chemically, since these are to some
extent bonded into the network, they would look like derivatives of the parent compound.49
In an extended network of sp2 and sp3-bonded carbon atoms, separation into aromatic and non-
aromatic domains is a natural consequence of thermodynamics and the geometrical constraints
imposed by bonding. This has led to the description of HAC and other amorphous carbons as
random covalent networks56–58 and is also the model proposed by Robertson & O’Reilly55 to
characterize the structure of these materials. In the Robertson & O’Reilly description, aromatic
domains of various sizes extending from single benzene-like rings to 10 ring ovalene-like structures,
are present in HAC while aromatic domains containing 20–50 rings are components of dehydro-
genated amorphous carbons. Recently, we have shown that coroneneX is a significant component
of HAC under certain conditions.51 These components have been observed to devolve from HAC in
laser ablation experiments.59
While a wide range of domain size is possible in a macroscopic solid, this is not feasible in
nanoparticles where surface effects are important and the maximum domain size is limited by the

Fig. 2 Representation of a 91-ring graphitic cluster from which seven carbon atoms have been removed at
random.64 Hydrogen atoms are shown as open circles. In (a) the carbon atoms adjacent to the vacancy remain
sp2 coordinated, forming CH bonds. In (b) these atoms bond to two hydrogens to form CH2 groups. These
atoms are then sp3 coordinated. It is apparent that the cluster in (a) continues to look like a single aromatic
domain since individual rings remain connected and p electrons are delocalized. This delocalization is lost in (b)
as the cluster separates into two smaller aromatic domains. The electronic spectrum of (b) will appear as that of
two substituted aromatic components; a molecule in a molecule configuration.

422 | Faraday Discuss., 2006, 133, 415–425 This journal is 


c The Royal Society of Chemistry 2006
number of carbon atoms in the particle. For a 50 atom particle one might expect
View Articlethat the
Online
introduction of a few defects would lead to the formation of a coroneneX domain together with
one or more small ring structures. Spectroscopically, electronic transitions in a distribution of
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

aromatic domains in a macroscopic solid results in continuous absorption at photon energies E Z


Eg and an Urbach tail at E r Eg.60 This should not be the case for nanoparticles having a limited
range of domain sizes, where spectra will be dominated by electronic transitions in only a few
relatively small aromatic components. In general, discrete transitions would be evident at
wavelengths typical of these components. Thus the visible absorption spectrum of a sample of 50
atom particles would more closely reflect that of its constituent domains than that of a macroscopic
piece of amorphous carbon. This conclusion is indicated from the laboratory studies of large
substituted PAH molecules.49,52 These particles may then represent an intermediate phase in the
evolution between large single PAH molecules and a macroscopic HAC solid (Fig. 3).
The structure of individual aromatic domains in interstellar carbon nanoparticles is expected to
reflect the enhanced stability of compact ring clusters including the fully benzenoid hydrocar-
bons.55,61,62 In these compounds, the p-electrons are delocalized, so that each carbon atom shares
the stability of the aromatic bond. Fully benzenoid PAH molecules include benzene, C6H6,
triphenylene, C18H12, and hexaperibenzocoronene, C42H18, all of which have high thermodynamic
stability due to the low energy of their HOMOs. Other highly stable compact clusters include pyrene
C16H10, coronene, C24H12, ovalene, C32H14 and circumcoronene, C54H18.62 Linear chains of rings, for
example napthalene, C10H8, anthracene, C14H10, and tetracene, C16H12, may also be expected, but are
less favored thermodynamically.55 Viewed as domain components of an interstellar carbon nanopar-
ticle, all of these molecules would contribute their carbon skeleton but would in general have different
peripheral terminations. This often would involve substituting C–C for C–H bonds (see Fig. 2b).
Robertson & O’Reilly55 estimate that the defect creation energy in an extended carbon network is
related to the difference in stability between clusters containing odd and even numbers of carbon
atoms. This is approximately Ed E 0.2–0.3 eV for clusters with 50–100 atoms decreasing to E0.1 eV
in larger structures implying that carbon nanoparticles are likely to be defected, with the defect
density increasing as the bandgap energy, Eg, decreases. Defects can take a variety of forms; for
example, the loss of a hydrogen atom from a terminal methyl (CH3) group results in a dangling

Fig. 3 C53H49, a chemical derivative of coronene, C24H12, containing a number of non-aromatic structures.
This is a radical containing an odd number of electrons and would have electronic transitions in the spectral
region of the DIBs. The electronic spectrum of this molecule would be dominated by transitions involving the p-
electrons in coronene. A structure of this kind would be more resistant to destruction than that of the parent
molecule (coronene).

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 423
bond and an extra p-electron. A study of non-bonding states in carbon networks hasArticle
View shown that
Online
these states are localized at the periphery of aromatic clusters.63
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

Conclusions
The DIB spectrum contains a vast amount of information about conditions in the DISM, but this
information will only become available if the carriers of these bands can be unambiguously
identified. This identification continues to be a challenge to theoretical, observational and
laboratory astrophysics. The primacy of experimental studies in resolving this unique problem is
apparent, and firm identifications will come only when the requisite chemical species have been
produced under conditions that simulate those in the ISM. To reach this point, it is important to
focus experimental studies on the correct set of materials and, in this regard, theory and observation
can provide some useful insights. In this paper, an attempt has been made to integrate chemical
constraints with those provided by observation. Furthermore, it seems likely that the carriers of the
DIBs must be chemical derivatives or decomposition products of the carbon particles that give rise
to the 2175 Å extinction bump. This implies that the carriers of the strongest DIBs are primarily
aromatic structures, but does not exclude the possibility that some carriers appear as the non-
aromatic decomposition products of larger, more abundant species. Some general conclusions are
as follows:-
(1) Photodissociation severely limits the equilibrium abundance of small molecular carriers and
implies that the carriers of the strongest DIBs must be larger species, but this constraint can be
relaxed if carriers are components of particles containing at least 50 carbon atoms.
(2) It appears that the weakest DIBs can be produced by small independent molecules if these are
formed by shock-induced desorption from particles containing several thousand carbon atoms.
(3) Some carbon nanoparticles may segregate internally into distinct aromatic domains with
carbon structures identical to the fused rings in PAH molecules. This process would be facilitated by
the formation of defects. For example, a 50 atom nanoparticle could divide into C16 (pyrene-like)
and C24 (coronene-like) aromatic domains connected by sp3-bonded carbon atoms. Other particles
could consist of aromatic molecules containing an extensive network of non-aromatic substituents
(Fig. 2).
(4) Electronic spectra of the components of such defected graphite particles should be predictable
from the core molecular structure.

Acknowledgements
This research was supported by the Natural Sciences and Engineering Research Council of Canada.

References
1 G. H. Herbig, Annu. Rev. Astron. Astrophys., 1995, 33, 19.
2 A. G. G. M. Tielens and T. P. Snow, The Diffuse Interstellar Bands IAU, Colloq. 137, Kluwer, Dordrecht,
1995.
3 T. P. Snow, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G. G. M. Tielens and T. P. Snow,
Kluwer, Dordrecht, 1995, p. 379.
4 A. G. G. M. Tielens, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G. G. Tielens and T. P. Snow,
Kluwer, Dordrecht, 1995, p. 395.
5 S. Leach, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G. G. M. Tielens and T. P. Snow,
Kluwer, Dordrecht, 1995, p. 281.
6 R. F. Warren-Smith, S. M. Scarrott and P. Murdin, Nature, 1981, 292, 317.
7 S. F. Fossey, PhD Thesis, University of London, 1990.
8 P. J. Sarre, Nature, 1991, 351, 356.
9 H. Van Winckel, M. Cohen and T. R. Gull, Astron. Astrophys., 2002, 390, 147.
10 T. H. Kerr, M. E. Hurst, J. R. Miles and P. J. Sarre, Mon. Not. R. Astron. Soc., 1999, 303, 446.
11 W. W. Duley, Mon. Not. R. Astron. Soc., 1998, 301, 955.
12 A. E. Douglas, Nature, 1977, 269, 130.
13 P. Thaddeus, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G. G. M. Tielens and T. P. Snow,
Kluwer, Dordrecht, 1995, p. 369.
14 J. Maier, P. Freivogel, J. Fulara and D. Lessen, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G.
G. M. Tielens and T. P. Snow, Kluwer, Dordrecht, 1995, p. 199.

424 | Faraday Discuss., 2006, 133, 415–425 This journal is 


c The Royal Society of Chemistry 2006
15 A. S. Webster, Mon. Not. R. Astron. Soc., 1993, 262, 831. View Article Online
16 W. W. Duley and J. D. McCullough, Astrophys. J., 1977, 211, L145.
17 G. P. Van der Zwet and L. J. Allamandola, Astron. Astrophys., 1985, 146, 76.
18 A. Leger and L. B. d’Hendecourt, Astron. Astrophys., 1985, 146, 81.
Published on 23 May 2006. Downloaded by Heinrich Heine University of Duesseldorf on 15/12/2013 09:29:48.

19 F. Salama, E. L. O. Bakes, L. J. Allamandola and A. G. G. M. Tielens, Astrophys. J., 1996, 458, 621.
20 D. M. Hudgins, C. W. Bauschlicher and L. J. Allamandola, Astrophys. J., 2005, 632, 316.
21 D. A. Williams, Astron. Geophys., 2003, 6, 14.
22 L. J. Allamandola, A. G. G. M. Tielens and J. R. Barker, Astrophys. J., Suppl. Ser., 1989, 71, 733.
23 T. Allain, S. Leach and E. Sedlmayr, Astron. Astrophys., 1996, 305, 602.
24 T. Allain, S. Leach and E. Sedlmayr, Astron. Astrophys., 1996, 305, 616.
25 H. W. Jochims, H. Baumgartel and S. Leach, Astrophys. J., 1999, 512, 500.
26 M. H. Vuong and B. H. Foing, Astron. Astrophys., 2000, 363, L5.
27 V. Le Page, T. P. Snow and V. M. Bierbaum, Astrophys. J., Suppl. Ser., 2001, 132, 233.
28 V. Le Page, T. P. Snow and V. M. Bierbaum, Astrophys. J., 2003, 584, 316.
29 E. Herbst, in The Diffuse Interstellar Bands IAU, Colloq. 137, ed. A. G. G. M. Tielens and T. P. Snow,
Kluwer, Dordrecht, 1995, p. 307.
30 R. P. A. Bettens and E. Herbst, Astrophys. J., 1996, 468, 686.
31 M. Frenklach and E. D. Feigelson, Astrophys. J., 1989, 341, 372.
32 W. W. Duley and D. A. Williams, Interstellar Chemistry, Academic Press, London, 1984.
33 K. W. Chan, T. L. Roellig, T. Onaka, M. Mizutani, K. Okumura, I. Yamamura, T. Tanabe, H. Shibai,
T. Nakagawa and H. Okuda, Astrophys. J., 2001, 546, 273.
34 D. M. Hudgins and L. J. Allamandola, Astrophys. J., 1999, 513, L69.
35 D. C. B. Whittet, Dust in the Galactic Environment, IOP, Bristol, 1992.
36 L. Spitzer, Physical Processes in the Interstellar Medium, Wiley, New York, 1978.
37 J. Maier, G. A. H. Walker and D. A. Bohlender, Astrophys. J., 2002, 566, 332.
38 M. Goto, W. Gaessler, Y. Hayano, M. Iye, Y. Kamata, T. Kanzawa, N. Kobayashi, Y. Minowa,
D. J. Saint-Jacques, N. Takami, N. Takato and H. Terada, Astrophys. J., 2003, 589, 419.
39 M. Rapacioli, C. Joblin and P. Boissel, Astron. Astrophys., 2005, 429, 193.
40 J. Pety, D. Teyssier, D. Fosse, G. Gerin, E. Roueff, A. Abergel, E. Habart and E. Cernicharo, Astron.
Astrophys., 2005, 435, 885.
41 Ph. Brechignac, M. Schmidt, A. Masson, T. Pino, P. Parneix and C. Brechignac, Astron. Astrophys., 2005,
442, 239.
42 A. Hedin, D. H. Fenyo, P. Hakansson, G. Jonsson and B. W. R. Sundquist, Nucl. Instrum. Methods Phys.
Res., Sect. B, 1989, 40-41, 275.
43 S. Della-Negra, Y. LeBeyec, B. Monart, K. Standing and K. Wien, Nucl. Instrum. Methods Phys. Res., Sect.
B, 1988, 32, 360.
44 P. Martin, Cosmic Dust: Its Impact on Astronomy, OUP, Oxford, 1978.
45 W. W. Duley, Q. J. R. Astron. Soc., 1986, 27, 403.
46 A. P. Jones, A. G. G. M. Tielens, D. J. Hollenbach and C. F. McKee, Astrophys. J., 1994, 433, 797.
47 T. P. Snow and A. N. Witt, Astrophys. J., 1996, 468, L65.
48 F. Salama, G. A. Galazutdinov, J. Krelowski, L. J. Allamandola and F. A. Musaev, Astrophys. J., 1999,
526, 265.
49 A. Colmsjo, Y. Zehuhr and C. Ostman, Chem. Scr., 1984, 23, 185.
50 W. W. Duley and S. S. Seahra, Astrophys. J., 1998, 507, 874.
51 W. W. Duley and S. Lazarev, Astrophys. J., 2004, 612, L33.
52 E. Clar and W. Schmidt, Tetrahedron, 1979, 35, 2673.
53 J. Biener, A. Schenk, B. Winter, U. A. Schubert, O. Lutterloh and J. Kuppers, Phys. Rev. B: Condens.
Matter, 1994, 49, 17307.
54 J. Fink, Th. Muller-Heinzerling, J. Pfluger, B. Scheerer, B. Dischler, P. Koidl, A. Bubenzer and R. E. Sah,
Phys. Rev. B: Condens. Matter, 1984, 30, 4713.
55 J. Robertson and E. P. O’Reilly, Phys. Rev. B: Condens. Matter, 1987, 35, 2946.
56 J. Angus and F. Jansen, J. Vac. Sci. Technol., A, 1988, A6, 1778.
57 A. P. Jones, Mon. Not. R. Astron. Soc., 1990, 247, 305.
58 G. Dadswell and W. W. Duley, Astrophys. J., 1997, 476, 184.
59 A. Scott, W. W. Duley and G. Pinho, Astrophys. J., 1997, 489, L193.
60 T. Datta, J. A. Woolham and W. Notohamiprodjo, Phys. Rev. B: Condens. Matter, 1989, 40, 5956.
61 E. Clar, The Aromatic Sextet, Wiley, London, 1972.
62 J. Aihara, Bull. Chem. Soc. Jpn., 1987, 60, 3143.
63 I. Varga and J. Pipek, Phys. Rev. B: Condens. Matter, 1990, 42, 5335.
64 M. A. Tamor and C. H. Wu, J. Appl. Phys., 1990, 67, 1007.

This journal is 
c The Royal Society of Chemistry 2006 Faraday Discuss., 2006, 133, 415–425 | 425

You might also like