Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~kgordon/papers/PS_files/Si_nano_apj.ps.gz Äàòà èçìåíåíèÿ: Thu Apr 24 02:49:35 2003 Äàòà èíäåêñèðîâàíèÿ: Tue May 27 07:56:11 2008 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: reflection nebula

Accepted 1998 May 1; Astrophysical Journal Letters
Preprint typeset using L A T E X style emulateapj
SILICON NANOPARTICLES: SOURCE OF EXTENDED RED EMISSION?
Adolf N. Witt
Department of Physics & Astronomy, The University of Toledo, Toledo, OH 43606
Karl D. Gordon
Department of Physics & Astronomy, Louisiana State University, Baton Rouge, LA 70803
and
Douglas G. Furton
Department of Physical Sciences, Rhode Island College, Providence, RI 02908
Accepted 1998 May 1; Astrophysical Journal Letters
ABSTRACT
We have reviewed the characteristics of the extended red emission (ERE) as observed in many dusty
astronomical environments, in particular, the diffuse interstellar medium of the Galaxy. The spectral
nature and the photon conversion efficiency of the ERE identify the underlying process as highly efficient
photoluminescence by an abundant component of interstellar dust. We have compared the photolumi­
nescence properties of a variety of carbon­ and silicon­based materials proposed as sources for the ERE
with the observationally established constraints. We found that silicon nanoparticles provide the best
match to the spectrum and the efficiency requirement of the ERE. If present in interstellar space with
an abundance sufficient to explain the intensity of the ERE, silicon nanoparticles will also contribute
to the interstellar 9.7 ¯m Si­O stretch feature in absorption, to the near­ and mid­IR nonequilibrium
thermal background radiation, and to the continuum extinction in the near­ and far­UV. About 36% of
the interstellar silicon depleted into the dust phase would be needed in the form of silicon nanoparticles,
amounting to less than 5% of the interstellar dust mass. We propose that silicon nanoparticles form
through the nucleation of SiO in oxygen­rich stellar mass outflows and that they represent an important
small­grain component of the interstellar dust spectrum.
Subject headings: ISM: dust -- radiation mechanisms: non­thermal
1. INTRODUCTION
The recent detection of extended red emission (ERE)
in the diffuse interstellar medium (ISM) of the Galaxy by
Gordon et al. (1998) and the spectroscopic confirmation of
its luminescence band nature by Szomoru & Guhathakurta
(1998) fundamentally changed our perception of the ERE
as an interstellar process in several ways. One, the ERE is
no longer seen as a phenomenon limited to localized special
environments, such as reflection and planetary nebulae.
As a Galaxy­wide process, its characteristics must now be
imposed as an additional observational constraint on mod­
els for interstellar dust in the diffuse ISM. Two, by corre­
lating the ERE intensity with the H I column density in the
diffuse ISM at intermediate and high Galactic latitudes,
and thus with the dust column density, it has become pos­
sible to determine the efficiency with which UV/visible
photons of the interstellar radiation field absorbed by in­
terstellar dust are converted into ERE photons. An ERE
photon conversion efficiency of (10 \Sigma 3)% (Gordon et al.
1998; Szomoru & Guhathakurta 1998) was derived by as­
suming the the ERE agent consumes all photons absorbed
by dust in the 91.2­550 nm wavelength range. To the ex­
tent that we know of the existence of other dust compo­
nents which are not expected to contribute to the ERE,
this assumption is false. The ERE agent, most likely, is not
responsible for the absorption of all absorbed UV/visible
photons. Thus, the derived efficiency is only a lower limit
to the true, intrinsic, efficiency of the photoluminescence
(PL) process seen as ERE. However, even if the intrinsic
efficiency is as high as 50%, a reasonable upper limit for
naturally occurring PL, the ERE agent still needs to ab­
sorb about 20% of all absorbed UV/visible photons in the
diffuse ISM. This can only be, if the ERE agent consists of
a cosmically abundant material which is capable of highly
efficient PL. It appears that only carbon­ or silicon­based
materials fit these criteria. In this letter, we present the
case for silicon nanoparticles being the component of the
ISM responsible for the ERE.
2. CHARACTERISTICS OF THE ERE AND THE IMPLIED
PROPERTIES OF ITS CARRIER
We briefly summarize the observed characteristics of the
ERE and refer to the original sources for the details of the
observations.
1. ERE manifests itself through a broad, featureless
emission band of 60 ! FWHM ! 100 nm, with a peak
appearing in the general wavelength range 610 ! – p !
820 nm. The presence of ERE has been established spec­
troscopically in many dusty astronomical environments,
e.g. the diffuse ISM (Szomoru & Guhathakurta 1998), re­
flection nebulae (Witt & Boroson 1990), planetary nebu­
lae (Furton & Witt 1990,1992), the Orion nebula (Perrin
& Sivan 1992), the high­b dark nebula L1780 (Chlewicki
& Laureijs 1987; Mattila 1979), and the starburst galaxy
M82 (Perrin et al. 1995). The ERE was first clearly recog­
nized in the peculiar reflection nebula called Red Rectangle
by Schmidt et al. (1980).
1

2 ERE from Si nanoparticles?
2. The ERE peak wavelength varies from one environ­
ment to another, and even within a given object, the peak
shifts with distance from the illuminating source. The den­
sity and hardness of the incident radiation field appear to
be the determining factors. The shortest peak wavelength
is seen for the ERE in the diffuse ISM (¸610 nm) in the
case of the relatively weak interstellar radiation field, the
longest (¸820 nm) is found on the Orion nebula bar adja­
cent to the Trapezium stars, where the radiation density
is several orders of magnitude higher.
3. The photon conversion efficiency of the ERE, cal­
culated on the basis that all photons absorbed by dust
in the 91.2­550 nm wavelength range in a given system
are absorbed by the ERE agent, with the absorption
of one UV/visible photon leading to the emission of at
most one ERE photon, has been determined to be near
(10 \Sigma 3)% in the diffuse ISM (Gordon et al. 1998: Szomoru
& Guhathakurta 1998), and it is at least this high in the
Red Rectangle and not much lower in the Orion nebula and
a number of reflection nebulae. This implies that the ERE
agent must absorb a significant fraction of the UV/visible
photons in the diffuse ISM and, simultaneously, possess an
exceedingly high intrinsic PL efficiency, AE10%. Since only
part of the energy of an absorbed photon is converted into
ERE, the energy conversion efficiency of the ERE agent is
only about (4 \Sigma 1)%.
4. No ERE is seen shortward of the wavelength of
540 nm. A sensitive search for PL by dust in several re­
flection nebulae in the 400­500 nm range, carried out by
Rush & Witt (1975), produced a null result. While found
in many dusty environments, ERE is essentially absent in
some, e.g. the Merope reflection nebula. The absence of
ERE in an otherwise normal dusty interstellar environ­
ment has not been explained.
5. Among planetary nebulae, ERE was detected only in
objects currently thought to be carbon­rich, in the sense
that C/O ? 1 (Furton & Witt 1992). This was thought to
be a strong argument in favor of the carbonaceous nature
of the ERE carrier. This argument can no longer be main­
tained in light of the ISO observations of C­rich planetaries
(Waters et al. 1998a) which show the presence of strong
spectral features of crystalline silicates in the mid­IR wave­
length region. In these planetaries, the oxygen­rich dust
appears to be present in the outer parts of the envelopes
where material from earlier mass­loss episodes resides. In
NGC 7027, the only planetary in which spatially resolved
ERE observations have been made, the ERE is found most
strongly enhanced in the outermost envelopes (Furton &
Witt 1990). In a similar way, ISO observations of the Red
Rectangle by Waters et al. (1998b), showing the presence
of oxygen­rich dust there as well, have greatly diminished
the argument that the currently C­rich stellar mass out­
flow implies a carbonaceous nature for the ERE carrier in
that object. One important aspect of the ERE in plane­
tary nebulae is that the ERE band is seen superimposed
upon the atomic continuum only, i.e. there is no evidence
of a scattered light component. This would indicate that
the dust particles in these objects are sufficiently small to
be in the Rayleigh limit, where scattering becomes ineffi­
cient. ERE observations in planetary nebulae, therefore,
imply that the origin of ERE is connected with very small
grains.
6. In clumpy reflection nebulae, e.g. NGC 2023 and
NGC 7023, ERE appears strongly enhanced in filamen­
tary structures, coincident with surfaces of clumps seen
in projection (Witt & Malin 1989). It was suggested that
the ERE in these filaments becomes activated by the expo­
sure of carbonaceous materials to warm atomic hydrogen
and UV radiation in molecular hydrogen photodissocia­
tion zones (Furton & Witt 1993). However, high spatial
resolution observations of H 2 vibrational fluorescence in
NGC 2023 by Field et al. (1994, 1998) showed that the
correlation between ERE and H 2 structures is not strong.
In addition to some clearly correlated structures, there are
ERE filaments without corresponding H 2 filaments. This
lack of correlation is not explainable by optical depth ef­
fects, as is the also observed case of H 2 filaments without
ERE filaments. An analogous result was found when high­
resolution H 2 observations of NGC 7023 were compared
to ERE structures in that nebula (Lemaire et al. 1996).
Frequently, interstellar environments exhibiting ERE also
emit the near­IR emission bands (UIR bands) attributed
to large aromatic molecules, e.g. PAHs. Often (e.g. in
NGC 2023, 7023, 7027, Orion nebula) both emission phe­
nomena are seen in photodissociation zones, but their in­
tensities are poorly correlated. In particular, in the Orion
bar, where both ERE and UIR bands have been observed,
there is no correlation between their respective intensities
(Perrin & Sivan 1992). There is therefore little support to
connect the emitters of the ERE and of the UIR bands, ex­
cept that they both occupy the same general environments
and that they respond to stellar UV illumination.
3. PROBLEMS WITH PREVIOUSLY PROPOSED ERE
CANDIDATES
Published models for materials causing the ERE have
relied on carbon­based substances, either solid­state car­
bonaceous solids or carbon­based molecules. The former
include hydrogenated amorphous carbon (HAC) (Duley
1985), quenched carbonaceous composite (QCC) (Sakata
et al. 1992), and coal (Papoular et al. 1996); the latter
category includes polycyclic aromatic hydrocarbon (PAH)
molecules (d'Hendecourt et al. 1986) and C 60 (Webster
1993). Common problems are the failure to match the
observed ERE spectra with the required PL efficiency.
The most widely advocated model involves HAC (Du­
ley et al. 1997; Furton & Witt 1993); it relies upon amor­
phous carbon, which is already a component of several in­
terstellar dust models. The HAC PL peak varies in wave­
length in response to varying the conditions of production
and subsequent treatment; the absorption spectrum in the
blue and UV is smooth and does not introduce trouble­
some absorption features unobserved in the ISM. It also
explains the 3.4 ¯m C­H stretch feature observed in ab­
sorption along large lines of sight in the diffuse ISM. The
QCC model is similar in all of these respects. Recent de­
tailed laboratory studies of HAC (Robertson 1996: Rusli
et al. 1996, and refrences therein), however, have led to
quantitative results which cast doubts upon HAC as the
ERE carrier. HAC is a highly efficient photoluminescing
material when its bandgap is large, near 4.4 eV. When illu­
minated by UV radiation, this HAC luminesces strongly in
the blue region of the spectrum, which makes it unsuitable
as an ERE analog. Narrowing the bandgap, e.g. by dehy­
drogenation, reduces the PL efficiency of HAC exponen­
tially, so that the HAC PL efficiency drops to a few 10 \Gamma4

Witt, Gordon, & Furton 3
of its maximum value when the bandgap is small enough
to yield PL emission in the wavelength region where ERE
is observed. HAC's principal problem, therefore, is its in­
ability to meet the spectral characteristics of the ERE and
the ERE efficiency simultaneously.
The PAH model (d'Hendecourt et al. 1986) was a com­
peting early suggestion, which attributed the ERE in the
Red Rectangle to luminescence by large PAH molecules.
However, even a large PAH like the 13­ring hexabenzo­
coronene exhibits a sharply structured PL spectrum peak­
ing near 500 nm wavelength, and smaller PAHs luminesce
generally at still shorter wavelengths. To obtain a broad,
unstructured band as observed in the ERE, d'Hendecourt
et al. rely upon a poorly studied process of intramolecu­
lar vibrational energy randomization. The fact that ERE
is seen only with peaks in the 610­820 nm wavelength
range would require substantially larger PAHs than hex­
abenzocoronene, with a total absence of smaller PAHs.
A further problem with all organic luminescing materials
is that their absorption spectra are highly structured as
well, which should lead to observable absorption bands in
the blue and UV spectral regions of highly reddened stars,
which so far remains undetected. On the other hand, the
PAH model has had considerable success in explaining the
near­IR emission bands at 3.3, 6.2, 7.7, 8.6, and 11.3 ¯m.
The C 60 model (Webster 1993) can produce reasonable
spectral matches with the observed ERE, especially if mix­
tures of differently sized fullerenes were used, but its fatal
weakness is the measured PL efficiency of 8:5 \Theta 10 \Gamma4 for the
C 60 fluorescence (Kim et al. 1992). The ERE efficiency re­
quirement is particularly difficult to match for any molec­
ular/organic luminescent material which does not exhibit
a broad continuous absorption spectrum covering the 90­
550 nm spectral range. The ERE requires the presence
of a material with a bandgap near 2 eV, which naturally
absorbs photons of higher energy and which converts part
of the absorbed energy of these photons into ERE photons
with an efficiency of not less than (10 \Sigma 3)%.
4. SILICON NANOPARTICLES AS A POTENTIAL ERE
SOURCE
In this section, we shall discuss the PL characteristics of
silicon nanoparticles and how these meet the requirements
posed by astronomical observations of ERE. In the next
section, we will explore whether such particles can form
under astronomical conditions and whether cosmic abun­
dances will permit the existence of sufficient numbers of
particles of this type.
In order for materials to be efficient photoluminescing
sources, two requirements must be met: The electronic
excitation resulting from the absorption of a photon must
be confined spatially, and possibilities for non­radiative re­
combinations must be minimized. In organic luminescent
materials such as PAHs the confining units are individ­
ual molecules, and photon yields can be as high as 99%
in some instances (Krasovitskii & Bolotin 1988). In the
case of HAC, small aromatic islands embedded in a sp 3 ­
coordinated amorphous carbon matrix are the absorbing
entities. The much higher band gap energy of the sur­
rounding matrix confines the excited electrons to the aro­
matic islands, where the PL can then occur (Robertson
1992, 1996). The observation of ERE in planetary nebu­
lae, which occurs in the absence of visible scattering, con­
firms that the luminescing particles are small compared to
conventional interstellar grains.
Silicon nanoparticles, which consist of crystalline silicon
cores of 1­3 nm diameter, surrounded by a SiO 2 mantle,
are known to be remarkably efficient PL emitters in the
1.5­2.0 eV energy range (e.g. Wilson et al. 1993). The
quantum confinement in such zero­dimensional crystallites
is responsible both for shifting the bandgap from a value
near 1.1 eV in bulk crystalline silicon to values of 2.0 eV
and above (Brus 1986, 1994; Delley & Steigmeier 1993;
Delerue et al. 1993) and for greatly enhancing the PL ef­
ficiency (Efros & Prigodin 1993) to laboratory­measured
values as high as 50% at temperatures around 50 K (Wil­
son et al. 1993). A series of recent investigations using
size­selected silicon nanostructures (Schuppler et al. 1994,
1995; Ehbrecht et al. 1997; Lockwood et al. 1996) have
established the correlation between the structure size and
the wavelength of peak PL emission. According to these
studies, silicon nanoparticles in the 1.5­5 nm diameter
range luminesce strongly in the 600­850 nm wavelength
range. The reported maximum PL efficiencies ocurred for
sizes near 1.6 nm, corresponding to 664 nm wavelength, for
two­dimensional silicon nanostructures (Lockwood et al.
1996) and for sizes near 3.9 nm, corresponding to 726 nm
wavelength, for zero­dimensional nanoparticles (Ehbrecht
et al. 1997), and both studies reported a rapid drop in
PL efficiency when going to both larger and smaller size
parameters.
Fig. 1 -- The observed ERE spectrum of NGC 2327 (Witt
1988) is plotted along with the photoluminescence of film
B at room temperature from Guha (1997). Film B had
a porosity of 70% which corresponds to a particle size of
4 nm.
We conclude from the published experimental results
that silicon nanoparticles luminesce extremely efficiently
(up to 50%) in the 600­850 nm range when they occur in
a very limited size range from 1.5­5.0 nm, and not other­
wise. The width of the PL band depends on the width of
the size distribution present, and the wavelength of max­
imum PL intensity is determined by the dominant size
within the distribution. The published PL spectra of sil­
icon nanoparticles match those observed in ERE sources

4 ERE from Si nanoparticles?
extremely closely (see Fig. 1). Environmental effects, such
as vaporization of the smaller particles in intense radiation
fields, will lead to a gradual shift of maximum PL toward
longer wavelength, as observed. On the other hand, the
diffuse ISM with the lowest radiation densities will allow
the existence of the smaller particles as well, with a resul­
tant maximum efficiency and a very broad PL spectrum.
5. SILICON NANOPARTICLES AS A COMPONENT OF
INTERSTELLAR DUST
Any interstellar dust component candidate must provide
positive answers to two fundamental questions. Can the
grains form naturally in an astronomical environment? Is
the required total grain mass consistent with the cosmic
elemental abundances of the constituents?
Silicon monoxide (SiO) is one of the most abundant
and most strongly bonded molecules in the outflows from
oxygen­rich stellar sources; virtually all silicon in the gas
phase is initially locked up in this form (Gail & Sedlmayr
1986). The nucleation of silicates is thought to be pre­
cipitated by the initial formation of SiO clusters (Nuth
1996), a conclusion supported by laboratory evidence re­
garding the condensation of SiO (Nuth & Donn 1982),
which showed the condensate to contain elemental silicon
plus silicon oxide SiO x with x ¸ 1:5. This can be un­
derstood by the fact that SiO 2 is the energetically pre­
ferred form of silicon oxide in the solid phase, while SiO
is preferred in the gas phase. The competition for addi­
tional oxygen atoms in the condensing cluster will leave
half of the silicon atoms ultimately without a partner, if
the change of state can complete itself. Since this conden­
sation occurs in a high­temperature environment (¸500­
1000 K), we postulate that the resulting annealing of the
condensing SiO clusters will lead to a separation of the
elemental silicon into a core and the SiO 2 into a mantle,
to form the basic structure of silicon nanocrystals with
oxygen passivation (Littau et al. 1993), i.e. in which the
dangling bonds at the surface of the silicon core are con­
nected to oxygen atoms in the SiO 2 mantle. As we will
show, only 5% of the total dust mass needs to remain in
the form of silicon nanoparticles; the overwhelming frac­
tion of the condensing particles can, therefore, remain in­
volved in the condensation of various silicates and metal
oxides, as supported by astronomical observations. We
conclude, therefore, that silicon nanoparticles are a prod­
uct of the initial dust formation process occurring in one of
several oxygen­rich outflow environments, e.g. M­type su­
pergiants, WN stars, type II supernovae, and AGB stars.
In addition to producing the ERE, these silicon nanopar­
ticles will be subject to temperature fluctuations, result­
ing from absorptions of individual UV photons, and thus
contribute to the near­ and mid­IR thermal background
radiation received from the diffuse ISM. The temperature
increases expected from the absorption of individual 10 eV
photons (the absorption peak of the Si cores is near 125
nm) by Si nanoparticles in the 1.5 ­ 5 nm size range are
170 K ­ 60 K (Purcell 1976). These are upper limits be­
cause part of the absorbed energy reemerges as ERE pho­
tons, and the actual particle mass is larger than assumed
here by the amount of SiO 2 contained in the mantle. The
Si nanoparticles required for efficient ERE production are
therefore not hot enough to cause significant emission in
the 9.5 ¯m Si­O band from their SiO 2 mantles. That would
require temperatures in excess of 230 K. Our proposal is
therefore not in conflict with the conclusion of Desert et
al. (1986) that the smallest interstellar grains are most
likely carbonaceous in nature. However, the SiO 2 mantles
should contribute to the Si­O features at 9.5 ¯m and 20
¯m in absorption, which are generally referred to as the in­
terstellar silicate features (Whittet 1992). To conclude, in
contrast to some other ERE candidates, silicon nanopar­
ticles are not expected to produce spectral features which
are not already observed in the diffuse ISM.
The question of abundance is of particular importance,
especially in view of the suggested reductions in the in­
terstellar heavy element abundances relative to hydrogen
(Snow & Witt 1996), compared to formerly used solar
abundances. Silicon absorbs strongly at energies above
3 eV, with an average absorption coefficient of 1:5 \Theta
10 6 cm \Gamma1 throughout the UV (Landolt­Bornstein 1982).
With a density of 2.42 g cm \Gamma3 for crystalline silicon, we
derive a cross­section per Si­atom of 2:9 \Theta 10 \Gamma17 cm 2 Si \Gamma1 .
Silicon nanoparticles luminesce with an intrinsic efficiency
of about 50% (Wilson et al. 1993). Coupled with the ob­
served lower limit to the ERE photon conversion efficiency
of (10 \Sigma 3)% (Gordon et al. 1998), this suggests that silicon
nanoparticles are responsible for 20% of the UV/visible
absorption in the ISM. Hence, along a typical interstellar
sightline of 1 kpc with an extinction of 3 mag kpc \Gamma1 in
the UV and a typical dust albedo of 0.5, silicon nanopar­
ticles contribute about 0.3 magnitudes of UV/visible ab­
sorption. This absorption and the cross­section per atom
leads to a density of silicon atoms required for nanoparti­
cles of 3 \Theta 10 \Gamma6 Si cm \Gamma3 , while the hydrogen density for the
chosen environment is about 1 H cm \Gamma3 . A similar number
of silicon atoms and twice that number of oxygen atoms
will be needed for the SiO 2 mantles. The total amount of
silicon in dust is 1:7 \Theta 10 \Gamma5 relative to hydrogen; thus, by
number only 36% of the interstellar silicon atoms present
in interstellar dust need to be present in the form of sil­
icon nanoparticles in order for the ERE observations to
be explained. The total mass of the silicon nanoparticles
with mantles, when compared with the total mass of de­
pleted interstellar atoms (Snow & Witt 1996) amounts to
less than 5%.
6. SUMMARY
We have examined the observed characteristics of the
ERE, a photoluminescence phenomenon associated with
interstellar dust and seen in a wide variety of astronom­
ical environments as excess radiation in the 600­850 nm
wavelength range. In the diffuse ISM, about 10% of ab­
sorbed UV/visible photons contribute to the ERE. The
ERE carrier must be a cosmically abundant material ex­
hibiting highly efficient photoluminescene properties. Sil­
icon nanoparticles containing a few hundred silicon atoms
each, surrounded by a SiO 2 shell, match the constraints re­
garding the spectrum and efficiency of the ERE. Interstel­
lar silicon abundances easily suffice to provide the needed
material quantities, and the nucleation of SiO in oxygen­
rich stellar outflows provides a likely source for their for­
mation. We suggest that silicon nanoparticles are an abun­
dant component of the interstellar dust spectrum.
ANW acknowledges stimulating and fruitful exchanges
of ideas with Drs. H.­ P. Gail, S. Guha, F. Huisken, K.

Witt, Gordon, & Furton 5
D. Kolenbrander, D. Lockwood, and E. Werwa. We also
acknowledge constructive comments from an anonymous
referee. This work was supported by grants from NASA
to The University of Toledo.
NOTE: After submission of this paper, we received an
advance copy of a paper by Ledoux et al. (1998), which
arrives at conclusions similar to ours.
REFERENCES
Brus, L. E. 1986, J. Quant. Electronics, QE­22, 1909
Brus, L. 1994, J. Phys. Chem., 98, 3575
Chlewicki, G., & Laureijs, R. J. 1987, in Polycyclic Aromatic
Hydrocarbons and Astrophysics, eds. A. Leger, L. d'Hendecourt,
& N. Boccara, (Dordrecht: Reidel), 335
d'Hendecourt, L. B., Leger, A., Olofson, G., & Schmidt, W. 1986,
A&A, 170, 91
Delerue, C., Allan, G., & Lannoo, M. 1993, Phys. Rev. B, 48, 11024
Delley, B., & Steigmeier, E. F. 1993, Phys. Rev. B, 47, 1397
Desert, F. X., Boulanger, F., Leger, A., Puget, J. L., & Sellgren, K.
1986, A&A, 159, 328
Duley, W. W. 1985, MNRAS, 215, 259
Duley, W. W., Searah, S., & Williams, D. A. 1997, ApJ, 482, 866
Ehbrecht, M., Kohn, B., Huisken, F., Laguna, M. A., & Paillard, V.
1997, Phys. Rev. B, 56, 6958
Efros, Al. L., & Prigodin, V. N. 1993, Appl. Phys. Lett. 62, 3013
Field, D., et al. 1994, ApJ, 286, 909
Field, D., et al. 1998, A&A, 333, 280
Furton, D. G., & Witt, A. N. 1990, ApJ, 364, L45
Furton, D. G., & Witt, A. N. 1992, ApJ, 386, 587
Furton, D. G., & Witt, A. N. 1993, ApJ, 415, L51
Gail, H.­ P., & Sedlemayr, E. 1986, A&A, 166, 225
Gordon, K. D., Witt, A. N., & Friedmann, B. C. 1998, ApJ, 498, (in
press)
Guha, S. 1997, Thin Solid Films, 297, 102
Kim, D., Lee, M., Suh, Y. D., & Kim, S. K. 1992, J. Am. Chem.
Soc., 114, 4429
Krasovitskii, B. M., & Bolotin, B. M. 1988, Organic Luminescent
Materials, (Weinheim: VCH Verlagsgesellschaft)
Landolt­Bornstein 1982, Numerical Data and Functional
Relationships in Science and Technology, 17, ed. O. Madelung,
(Berlin: Springer Verlag), 387
Ledoux, G., et al. 1998, A&A, 333, L39
Lemaire, J. L., et al. 1996, A&A, 308, 895
Littau, K. A., Szajowski, P. J., Muller, A. J., Kortan, A. R., & Brus,
L. E. 1993, J. Phys. Chem., 97, 1224
Lockwood, D. J., Lu, Z. H., & Baribeau, J.­ M. 1996,
Phys. Rev. Lett., 76, 539
Mattila, K. 1979, A&A, 78, 253
Nuth, J. A. 1996, in The Cosmic Dust Connection, ed. J. M.
Greenberg (Dordrecht: Kluwer Sci. Publ.), 205
Nuth, J. A., & Donn, B. 1992, J. Chem. Phys., 77, 2639
Papoular, R., Conard, J., Guillois, O., Nenner, I., Reynaud, C., &
Rouzaud, J.­ N. 1996, A&A, 315, 222
Perrin, J.­ M., & Sivan, J.­ P. 1992, A&A, 255, 271
Perrin, J.­ M., Darbon, S., & Sivan, J.­ P. 1995, A&A, 304, L21
Purcell, E. M. 1976, ApJ, 206, 685
Robertson, J. 1992, Philos. Mag. B, 66, 199
Robertson, J. 1996, Phys. Rev. B, 53, 16302
Rusli, Robertson, J., & Amaratunga, G. A. J. 1996, J. Appl. Phys.,
80, 2998
Sakata, A., et al. 1992, ApJ, 393, L83
Schuppler, S., et al. 1994, Phys. Rev. Lett., 72, 2648
Schuppler, S., et al. 1995, Phys. Rev. B, 52, 4910
Schmidt, G. D., Cohen, M., & Margon, B. 1980, ApJ, 239, L133
Snow, T. P., & Witt, A. N. 1996, ApJ, 486, L65
Szomoru, A., & Guhathakurta, P. 1998, ApJ, 494, L93
Waters, L. B. F. M., et al. 1998a, A&A, 331, L61
Waters, L. B. F. M., et al. 1998b, Nature, 391, 868
Webster, A. 1993, MNRAS, 264, L1
Whittet, D. C. B. 1992, Dust in the Galactic Environment, (Bristol:
IOP Publ. Ltd)
Wilson, W. L., Szajowski, P. F., & Brus, L. E. 1993, Science, 262,
1242
Witt, A. N. 1988, in Dust in the Universe, ed. M. E. Bailey & D. A.
Williams (Cambridge: Cambridge Univ. Press), 1
Witt, A. N., & Boroson, T. A. 1990, ApJ, 355, 182
Witt, A. N., & Malin, D. F. 1989, ApJ, 347, L25