UV-visible spectroscopy of organic carbon particulate sampled from ethylene/air flames

Article abstract of DOI:10.1016/S0045-6535(00)00241-1

A systematic comparison of spectra obtained with extra and in situ diagnostics in the soot preinception region of rich, premixed ethylene air flames suggests that combustion generated organic carbon (OC) particulate can be extracted from flames and isolated from other flame material for further chemical analysis. Both the trend with height above the burner and the form of UV fluorescence and absorption spectra from extra situ sampled material captured in water agree with those measured in situ. These results show that the OC particulate formed in flames is partially water soluble. However, the collection efficiency can be increased using less polar solvents, like acetonitrile and dichloromethane. The fluorescence spectra from the water samples are comprised both a naphthalene-like component and a broad band UV fluorescence component similar to that observed in situ which is attributed to flame generated OC particulate. The broad band UV fluorescence centered around 320 nm is also observed very early in flames and does not change considerably with increasing flame residence time. These results support previous hypotheses that the UV broad band fluorescence is from carbonaceous material comprised two-ring aromatics, formed earlier than soot in the flame, and is still present along with soot at higher heights or flame residence times.

Full text of DOI:10.1016/S0045-6535(00)00241-1

Chemosphere 42 (2001) 671±680
UV-visible spectroscopy of organic carbon particulate sampled
from ethylene/air ¯ames
a
L.A. Sgro , P. Minutolo
a,b,
a
, G. Basile , A. D'Alessio
a
*
a
Dipartimento di Ingegneria Chimica, Universita degli Studi di Napoli Federico II, Piazzale Tecchio 80, 80125 Napoli, Italy
b
Istituto di Ricerche sulla Combustione, CNR, Piazzale Tecchio 80, 80125 Napoli, Italy
Abstract
A systematic comparison of spectra obtained with extra and in situ diagnostics in the soot preinception region of
rich, premixed ethylene air ¯ames suggests that combustion generated organic carbon (OC) particulate can be extracted
from ¯ames and isolated from other ¯ame material for further chemical analysis. Both the trend with height above the
burner and the form of UV ¯uorescence and absorption spectra from extra situ sampled material captured in water
agree with those measured in situ. These results show that the OC particulate formed in ¯ames is partially water soluble.
However, the collection eciency can be increased using less polar solvents, like acetonitrile and dichloromethane. The
¯
¯
uorescence spectra from the water samples are comprised both a naphthalene-like component and a broad band UV
uorescence component similar to that observed in situ which is attributed to ¯ame generated OC particulate. The
broad band UV ¯uorescence centered around 320 nm is also observed very early in ¯ames and does not change
considerably with increasing ¯ame residence time. These results support previous hypotheses that the UV broad band
¯uorescence is from carbonaceous material comprised two-ring aromatics, formed earlier than soot in the ¯ame, and is
still present along with soot at higher heights or ¯ame residence times. Ó 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Nanoparticles; Premixed ¯ames; Soot; Soot precursors
1
. Introduction
particulate (Wilson and Spengler, 1996; Peters et al.,
997). The toxicological mechanism, though still un-
1
Recent ®ndings from atmospheric science, toxicol-
known, is also thought to involve the ultra®ne particles
(Oberdorster et al., 1995; Donaldson et al., 1998). The
interaction of small particles with water is also of in-
terest to human health since water comprises 90% of
the lung ¯uid, and the dose received from exposure
depends partly on the hygroscopic nature of particles.
Small particles that are partially water soluble may also
signi®cantly a€ect the albedo of our atmosphere be-
cause of their ability to contribute to cloud condensa-
tion (Cruz and Pandis, 1997; Seinfeld and Pandis,
ogy, and epidemiology reveal the need to focus more
research e€orts toward characterizing and understand-
ing the formation processes of the ultra ®ne
ꢀ
d < 0:1 lm† and water soluble fraction of particulate
matter because of their possible e€ects on human
health and the environment. Epidemiological studies
report that the association between morbidity and/or
mortality and particulate matter increases and becomes
more statistically signi®cant moving to smaller-sized
1
998).
In response to the need to better understand the toxic
and radiative e€ects of particulate matter, recent re-
search e€orts in atmospheric chemistry are shifting to-
ward characterizing organic carbon (OC) particulate
rather than the previously well studied elemental carbon
*
Corresponding author. Tel.: +390-81-768-2246; fax: +390-
1-593-6936.
E-mail address: minutolo@irc.irc.na.cnr.it (P. Minutolo).
8
0
045-6535/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ( 0 0 ) 0 0 2 4 1 - 1

6
72
L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
(
EC). EC is black material partially identi®ed with soot
2. Experimental
that absorbs visible radiation and is hydrophobic. OC is
transparent to visible radiation, less characterized
chemically, more hydrophilic, and comprises more of
the ultra®ne fraction of atmospheric particulate than EC
Two laminar, premixed, sooting ethylene air ¯at
¯ames ꢀC=O ˆ 0:77 and 0:92† are studied using in situ
(Fig. 1(a)) and extra situ (Fig. 1(b)) sampling techniques.
The ¯ames are stabilized with a stainless steel plate
(located 29 mm above the burner surface) on a capillary,
water-cooled, burner ꢀdiameter ˆ 5:6 cm†. Other prop-
erties of these ¯ames are listed in Table 1. Flame tem-
peratures are measured using the Na D-line reversal
method employing a calibrated tungsten strip lamp.
Flow rates of the feed gasses to the burner are controlled
with Brooks instruments mass ¯ow controllers. The
position of the ¯ame front is measured by the maximum
(
Hildemann et al., 1991, 1994; Rogge et al., 1993; Saxena
et al., 1995; Havers et al., 1998; Seinfeld and Pandis,
998). The OC/EC ratio is often greater than 1 in urban
1
environments.
In direct studies of combustion sources, organic
particles one order of magnitude smaller than soot
have been previously identi®ed in rich non-sooting
premixed ¯ames and in the region prior to soot in
slightly sooting ¯ames using UV-visible optical diag-
nostics (Minutolo et al., 1994, 1996, 1998, 1999;
Gambi, 1996; D'Alessio et al., 1998). Their optical
properties suggest that these organic particles are mo-
lecular clusters or polymer-like particles, comprised
2
‡
2
OH emission at 306.4 nm ꢀA R ! X P†.
We use the fourth harmonic of a pulsed Nd:YAG
laser ꢀk ˆ 266 nm† as the light source for in situ laser
0
light scattering (LLS) and laser induced ¯uorescence
(LIF) for both in and extra situ samples (Fig. 1(a)). The
laser beam is vertically polarized and focused to the
center of the burner with a cylindrical lens. The laser
pulse is attenuated to avoid photofragmentation or va-
porization, and the energy of the laser is 0.2 mJ with a
2
-ring aromatic and aliphatic bonds. These organic
particles become more aromatic with longer ¯ame
residence time (as observed by a measurable absorption
in the visible), and eventually form primary soot par-
ticles. Studies of combustion systems that use only di-
agnostics in the visible to observe particle formation do
not observe these organic 2±5 nm particles since they
are initially transparent to visible light, and can only be
detected with UV extinction and ¯uorescence. Others
report comparable soot precursor material in di€usion
5
duration of 15 ns and a laser ¯uence of 1:7 Â 10
�
2
W cm , which is well below the ¯uence that causes the
2
appearance of the C swan bands emission signal. A
high-pressure 150 W xenon lamp, ®ltered with a predi-
spersing monochromator to avoid stray light, is used to
¯ames using light scattering techniques (Santoro and
Semerjian, 1984) and thermophoretic sampling fol-
lowed by visualization with transmission electron mi-
croscopy (TEM) (Wesborg et al., 1973; Dobbins and
Subramaniasiavam, 1994; Dobbins et al., 1995; Koylu
et al., 1997).
Recently, extra situ sampling of this ¯ame generated
±5 nm organic particulate in water showed spectral
2
features similar to those measured in situ, and promised
as a method for isolating this material from other species
in ¯ames (Gambi, 1996; Minutolo et al., 1999) and in the
exhausts of engines and industrial ¯ames (Borghese and
Merola, 1998; Merola et al., 1999) for chemical analysis.
While earlier work showed qualitative similarities be-
tween in situ and extra situ spectra (Beretta et al., 1992;
Minutolo et al., 1998), there is a lack of systematic de-
velopment of the technique.
(a)
This paper examines the water solubility of com-
bustion generated organic particulate matter by com-
paring the UV-visible spectra observed in situ in rich
ethylene air ¯ames with the spectra of extra situ sampled
material trapped in water. We investigate the similarity
between spectra obtained in situ and those obtained
from the material sampled extra situ at various heights
above the burner to evaluate the ability of the extra situ
technique to isolate the 2±5 nm carbonaceous material,
identi®ed by broad band UV absorption and ¯uores-
cence features.
Fig. 1. Experimental set up (a) in situ laser light scattering and
absorption and laser induced ¯uorescence for both in situ and
extra situ samples. (b) extra situ isokinetic sampling.

L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
673
Table 1
Flame properties, cold ¯ow velocity ˆ 7:5 cm s�
experiments in this study was 30 min, and 8 ml of solvent
was used in each condenser.
1
C/O ˆ 0.77
U ˆ 2.31
C/O ˆ 0.92
U ˆ 2.76
To determine if the extra situ UV ¯uorescence and
absorption signals are a byproduct of our sampling
procedure, we ran a cold ¯ow of only ethylene and an-
other of air through the burner and the extra situ sam-
pling system. After sampling for the same amount of
time with the same ¯ow rates used for the ¯ame exper-
iments, we found a barely detectable absorption signal
and no ¯uorescence for both the air and ethylene ¯ows.
The UV ¯uorescence and absorption signals reported in
this paper are found only when sampling material pro-
duced in ¯ames.
T
max ˆ 1700 K
T
max ˆ 1675 K
Z
flame front ˆ 0.75 mm
Z
flame front ˆ 1.5 mm
obtain in situ extinction measurements. The light signal,
0° for scattering and ¯uorescence, and line of sight for
9
extinction, is focused on the entrance slit of a f/8.8
spectrometer, which has a focal length of 500 mm. A
vertical polarization analyzer is placed in front of the
entrance to the spectrometer for scattering and ¯uores-
cence measurements. An intensi®ed coupled charged
device (ICCD) camera is aligned to the exit of the
spectrometer. A delay generator controls the ICCD ex-
posure to synchronize the opening of the gate to the
ICCD with the timing of the laser source, and a PC
stores the signal from the ICCD.
3. Results
The light scattering and extinction measurements
obtained as a function of height above the burner in the
C=O ˆ 0:92 ¯ame are reported in Fig. 2. We subtract
from the measured scattering (at 266 nm) and extinction
(at 266 and 530) in situ, respectively, the scattering and
extinction of gas phase components, calculated using
their concentrations predicted by the premix model for
CHEMKIN and their published cross-sections. There-
fore, the excess extinction and scattering is the signal
remaining after that expected from the gas phase species
is subtracted (Fig. 2). The gas phase contribution to
absorption is negligible for these rich ¯ames (on the
We use an isokinetic water-cooled, stainless steel
probe (ID 2 mm) to extract material from the ¯ame that
is transported through Te¯on sampling lines and bub-
bled through a solvent in two water-cooled condensers
connected in series (Fig. 1(b)). Extra situ extinction
measurements are taken using a Hewlett-Packard 8452A
spectrophotometer. Fluorescence spectra of extra situ
samples are obtained with the same optical setup used
for in situ measurements with a standard 1 cm path-
length quartz cell. Fluorescence measurements obtained
by centering the water sample in the probe volume of the
laser compare well with the ¯uorescence spectra ob-
tained by a commercial Perkin-Elmer LS-50B spectro-
�
4
� 1
order of 10 cm in the UV) because there is no con-
siderable amount of CO , which is highly absorbing in
the UV at high temperature. Just above the ¯ame front,
2
an excess UV extinction and scattering are observed.
The extinction increases with height above the burner (z)
while the excess scattering signal remains constant until
z ˆ 5:5 mm, where a large increase in scattering, attrib-
uted to the onset of soot formation, is observed. Visible
extinction is below the limit of detection for z < 3:5 mm.
The ¯uorescence spectra for the two examined ¯ames
at di€erent heights above the burner obtained in situ are
reported in Fig. 3. In situ ¯uorescence curves show a
broad band in the UV centered around 320 nm that is
constant with z. Also, a broad band in the visible
peaking around 530 nm is observed for z > 3:5 mm that
increases strongly as a function of z.
¯uorimeter. Fluorescence and scattering signals are
corrected for extinction (measured at the excitation
wavelength 266 nm), for spectral response of the detec-
tion system measured with the calibrated tungsten strip
lamp, and for changes in laser power using the measured
and theoretical scattering cross-section at 266 nm of
methane at ambient temperature. Therefore, reported
scattering and ¯uorescence signals are in absolute units
� � 1
1
(
cm sr ). A 400 nm long pass ®lter is used when
taking spectra in the visible to remove the 2nd order
signal.
All of the samples reported in this paper were ana-
lyzed within a few hours of collection because we ob-
served that the optical properties of the samples started
changing at longer time. Some experiments were run to
optimize the collection system to capture more material
in less sampling time. For example, using a cone-shaped
condenser rather than a larger, cylindrical one increased
the transfer time for the sample inside the bubble to
di€use to the bubble surface and establish contact with
the solvent. Increasing the amount of solvent also in-
creases the transfer time, but decreases concentration
due to dilution. The sampling time for all extra situ
It is known that laser induced incandescence (LII)
often interferes with LIF measurements (Ni et al., 1994).
However, the low laser ¯uence of our attenuated laser
beam and the shape of the ¯uorescence signal, which has
a maximum at 530 nm for the C=O ˆ 0:92 ¯ames, sug-
gest the signal is unlikely to be incandescence. Further-
more, the in situ signal at 530 nm is also detected in the
presooting region ꢀ<5 mm†, where the absorptivity/
emissivity in the visible of the OC particles is very low,
and it is even less likely due to incandescence. To verify
the nature of the visible laser induced signal, we

6
74
L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
Fig. 3. In situ scattering and ¯uorescence. Numbers near the
curves indicate the height above the burner in mm (onset of
soot formation occurs at z ˆ 5:5 mm).
Fig. 2. In situ scattering (a) and extinction (b) measurements in
excess of the calculated gas phase contribution for the
C=O ˆ 0:92 ethylene/air ¯ame. Note: Both the raw scattering
data, Qvv, and the excess scattering signal after the gas phase
contribution is subtracted are shown in (a). The gas phase
contribution for extinction is negligible in these rich ¯ames.
ples could be due to naphthalene by subtracting the
¯uorescence spectrum from a standard, created by sat-
�
1
urating naphthalene in water (solubility ˆ 30 200 lg l
)
times a subtraction factor. This subtraction factor is
evaluated by the disappearance of the peaks at 320, 328,
and 332 nm from the ¯uorescence spectra. Fig. 5 shows
the ¯uorescence curves of an extra situ sample (a), the
amount of naphthalene estimated in the sample (b), the
subtraction of these two curves (c), and the ¯uorescence
measured in situ (d). There is considerable similarity
between the ¯uorescence peaks in the 320±330 nm range
in the total extra situ material sampled and naphthalene
in water (Figs. 5(a) and (b)). Also, the form of the ¯u-
orescence spectrum after subtraction of naphthalene is
very similar to the ¯uorescence measured in situ in
¯ames (Figs. 5(c) and (d)).
measured the time response of this signal at 530 nm, a
parameter used to distinguish LIF from LII (Bengtsson
and Alden, 1989; Vander Wal et al., 1997), at a height of
1
0 mm above the burner, well into the sooting region of
the ¯ame. The signal decay rate is an order of magnitude
faster than that expected for LII signals, providing fur-
ther evidence that the measured light comes from ¯uo-
rescence rather than incandescence.
The ¯uorescence spectra from the extra situ material
possess only a UV component which is more structured
than the broad band UV ¯uorescence measured in situ,
containing a series of sharp peaks at 320, 328, and 332 nm
Comparing the ¯uorescence spectrum of naphthalene
(Fig. 5(b)) with those from the extra situ material
(Figs. 5(a) and 4) reveals that naphthalene does not
produce ¯uorescence at 300 nm, while both the sub-
tracted (extra situ material ± naphthalene) and the in situ
curves do ¯uoresce at 300 nm. The monochromatic
¯uorescence at 300 nm can be tracked to determine the
trend with height of the extra situ sampled material that
(Fig. 4). Also, unlike the UV ¯uorescence detected in
situ, which does not change considerably with z, the
extra situ ¯uorescence increases strongly as a function
of z. The peaks at 320, 328, and 332 nm in the UV ¯uo-
rescence spectra from the extra situ material are similar
to those of naphthalene (Karcher, 1992). We estimate
how much of the ¯uorescence from the extra situ sam-

L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
675
Fig. 5. Fluorescence spectra of total extra situ sampled material
at z ˆ 3:5 m, C=O ˆ 0:92 (a), naphthalene estimated in this
sample (b), total material ± naphthalene (c), and ¯uorescence
measured in situ at the same height and ¯ame conditions (d).
Fig. 4. Extra situ ¯uorescence. Numbers near the curves indi-
cate the height above the burner (onset of soot formation oc-
curs at z ˆ 5:5 mm).
¯uoresces similar to that observed in situ without the
additional naphthalene-like ¯uorescence. Fig. 6 plots
¯
¯
uorescence at 300 nm along with that at 330 nm (where
uorescence is observed also for naphthalene) as a
function of z. Fluorescence of the samples in the ®rst
condenser at 300 nm is relatively constant, showing a
trend similar to the UV component of ¯uorescence
observed in situ, but the 330 nm ¯uorescence shows a
much stronger increase with z.
The material collected at z ˆ 2 mm shows a broad
band extinction in the UV that decays with wavelength
similar to that observed in situ (Fig. 7). Also, the in-
tensity of the absorption spectrum increases with height
above the burner similarly for both in situ and extra situ
measurements, providing evidence that the sampling
procedure is able to isolate the material responsible for
Fig. 6. Extra situ monochromatic ¯uorescence at 300 nm (open
symbols) and 330 nm (®lled symbols) vs. height above the
burner for the ®rst (squares) and second (circles) condensers.

6
76
L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
Fig. 8. Measured extinction coecient,
K
extꢀk ˆ 266 nm†,
0
Fig. 7. Extinction spectra of total extra situ sampled material at
di€erent heights above the burner for C=O ˆ 0:77. Also shown
on the graph is the in situ extinction at 5 mm.
from material collected in the ®rst and second condenser
compared with in situ measurements. Extra situ Kext is mea-
sured from material sampled for 30 min in 8 ml of water.
the UV absorption in the ¯ame. This is seen by the
monochromatic extinction coecients at 266 nm, which
are reported in Fig. 8. The in situ and extra situ curves in
Fig. 8 have di€erent scales because of the di€erent par-
ticle concentration. In the case of the water sample, the
concentration depends on sampling time, dilution in
water, and collection eciency, which is in the order of a
few percent.
not all, can be attributed to naphthalene. In fact, a low
concentration of naphthalene, with a low level absorp-
tion, can produce a relatively large ¯uorescence signal
because of its rather large quantum eciency (QE)
ꢀQE ˆ 0:23† (Karcher, 1992).
The peaks in the extra situ extinction curves could be
due to other PAHs also trapped in the water, but not to
PAHs with more than two rings since we did not observe
¯uorescence in the visible in any of our samples. The
spectra of the commonly reported species from chemical
analyses of ethylene ¯ames (Beretta et al., 1985; Petarca
and Marconi, 1989; Ciajolo et al., 1994; Olten and
Senkan, 1999) could not account for these peaks from
our combined absorption and ¯uorescence measure-
ments. The peaks at 236 and 246 nm may be due to
similar material that produces a hump in the extinction
spectra at 230 nm observed in situ in ethylene (Minutolo
et al., 1998) and methane (Buchta et al., 1995) ¯ames
and at 240 nm in benzene ¯ames (Minutolo et al., 1996).
Increased spectral resolution of the in situ extinction
spectra is needed to better resolve this 230±240 nm
hump.
At higher heights above the burner the intensity of
the broad band absorption spectrum increases and three
peaks at 220, 236 and 246 nm and a broader hump at
2
60±280 nm are present (Fig. 7). Using the estimated
concentration of naphthalene in our samples, obtained
from the samplesÕ ¯uorescence spectra and the quantum
yield and molar absorptivity of naphthalene obtained
from the literature at 266 nm (Karcher, 1992), we can
subtract the extinction expected from the naphthalene in
the samples from the extinction spectra of the extra situ
samples. Subtracting the naphthalene contribution
changes the shape of the extinction spectra less notice-
ably than the ¯uorescence spectra. Just the small hump
from 260 to 280 nm and part of the 220 nm peak, but

L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
677
The (QE) can be calculated from measured extinction
and ¯uorescence by
the peaks superimposed on the background UV ab-
sorption are less evident with respect to the ®rst con-
denser (Fig. 7). These results suggest that the ®rst
condenser traps both materials that ¯uoresce similar to
that observed in situ and a naphthalene-like ¯uorescing
material. However, only the ®rst kind of material is not
completely collected by the ®rst condenser and is also
trapped in the second condenser.
R
k0
Q
dk
fk
k0
QE ˆ 4p
;
K
ext
k0
where Q is the ¯uorescence measured at wavelength, k,
fk
k0
ext
excited at k
0
, and K is the measured extinction coe-
We ran additional extra situ experiments using less
polar solvents, acetonitrile and dichloromethane, in
which we found extinction and ¯uorescence spectra
similar to the hydrosol samples. No new peaks in the
cient at k . We can calculate the QE of the UV broad
0
band ¯uorescence in situ at z ˆ 2 mm, where the sam-
pled material does not ¯uoresce in the visible. The QE of
the extra situ material is generally higher than that
measured in situ, and it increases with height above the
burner, approaching the QE of naphthalene, which is
two orders of magnitude higher than that observed in
situ. After subtracting the naphthalene from the extra
situ spectra, the quantum yield no longer increases with
z, and is one order of magnitude lower, closer to the
value observed in situ (Table 2).
¯uorescence spectra were found using di€erent solvents,
but the ¯uorescence intensities for the less polar solvents
were higher. The QEs for the material collected by dif-
ferent solvents, water included, are similar when calcu-
lated after the subtraction of naphthalene. Therefore,
also less polar solvents can collect the two kinds of
materials collected by water: a naphthalene-like one and
the broad band UV absorbing particles also detected in
situ. The relative amounts of these two components
collected depend on the solvent. This is shown in Table 2
from the ratio of the ¯uorescence intensity at 300 and
The ¯uorescence in the visible detected in situ that
increases strongly as a function of height is not present
in any of the extra situ samples. Visible ¯uorescence was
also not detected in experiments using dichloromethane
or acetonitrile as a solvent in the condensers instead of
water. We ¯ushed the lines and sample probe with dif-
ferent solvents, and analyzed the material trapped in the
probe lines. Water ¯ushes out material that absorbs/
3
30 nm. These results suggest that while UV absorbing
particles can be trapped in water in high enough con-
centration to be analyzed, less polar solvents can be used
to increase collection eciency.
¯uoresces mostly in the UV, but less polar solvents
collected material that absorbs/¯uoresces both in the UV
and the visible from the walls of the probe. The ¯uo-
rescence of the total material trapped in the lines shows
signi®cant ¯uorescence in the visible in addition to the
UV similar to that measured in situ, but with more
structured peaks.
4
. Discussion
The main purpose of this study was to examine the
ability of the sampling method to isolate organic parti-
cles formed in ¯ames from other ¯ame generated ma-
terial and to examine the interaction of these particles
with water. Elaboration of the in situ scattering and
extinction measurements (Fig. 2) estimates a diameter of
We also analyzed the material trapped in the second
condenser connected in series. For both the ¯ames, the
¯uorescence spectra from the second condenser do not
2
±3 nm for the particles, which are transparent in the
visible and possess a broad band UV absorption and
uorescence spectra, in agreement with past results
Minutolo et al., 1998). The shape of the extra situ ex-
exhibit the strong peaks similar to naphthalene observed
in the spectra from material collected in the ®rst con-
denser, and the ¯uorescence at 300 and 330 nm are
relatively constant at various heights (Fig. 6). Similarly,
the extinction spectra in the second condenser show the
same trend with wavelength as the in situ spectra, and
¯
(
tinction spectra is similar to those observed in situ (Fig. 7),
and also the pro®les of intensity of extinction spectra
versus z show the same trend (Fig. 8). These results
strongly suggest that the 2±3 nm organic particles are
captured in the samples.
Table 2
Di€erences in material collected with di€erent solvents
Comparing the UV portion of the ¯uorescence curves
of in and extra situ samples (Figs. 5 and 6) further
suggests the 2±3 nm organic particles are isolated in the
water samples with only one other interfering species
whose ¯uorescence coincides with naphthalene. The in
situ ¯uorescence in the UV is relatively constant as a
function of height above the burner, while visible ¯uo-
rescence increases strongly. After subtracting the naph-
thalene contribution, the extra situ ¯uorescence spectra
closely resemble the in situ spectra in the UV (both
Solvent
Fluores-
cence at
QE (calculated
after subtracting
the naphthalene
component)
3
00/330 nm
Dichloromethane
Acetonitrile
Water
0.97
1.33
0.51
0.083
0.056
0.053
0.003
0.23
In situ material
Naphthalene

6
78
L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
shape and trend with z), even in ¯ame conditions where
visible ¯uorescence is observed in situ. This result fur-
ther demonstrates the ability of the sampling procedure
to isolate the particles, even when more aromatic ma-
terial and also soot particles are present.
tures. The fact that naphthalene peaks in ¯uorescence
are seen only in the ®rst condenser suggests it is more
eciently collected than the 2±3 nm particles, which
break through to the second condenser in series and
have a sampling eciency on the order of a few percent.
It is also not surprising that naphthalene is the only
PAH identi®ed in the spectra of our extra situ samples
since the solubility in water of naphthalene is one order
of magnitude higher than other PAHs reported in high
concentration in ethylene air ¯ames (Lee et al., 1981).
The experimental evidence can also be explained by
the hypothesis that the PAHs seen in extra situ sampling
separate or disaggregate during the sampling from the
particles, while in the ¯ame they are attached as a part of
a polymeric structure of aromatic material joined by
aliphatic bonding. These carbonaceous structures could
be held together by weak bonds of aggregation rather
than molecular bonds as is the case for asphaltenes
(Sheu and Mullins, 1995). It may be that a similar
bonding forms in the rich combustion of hydrocarbons
at high temperatures, and these aggregates disassociate
in solution at room temperature whereas in the ¯ame, at
high temperature, the reactions to form aggregates are
faster than the destruction reactions. This point of view
can explain both the large excess scattering signal ob-
served in situ that, combined with UV extinction, comes
from some 2±3 nm scatterer, with aliphatic and aromatic
functionalities as well as the fact that the extra situ
spectra contain signatures of naphthalene that are not
observed in situ.
To separate the 2±5 nm particles from other ¯ame
generated material, we make use of the previous result
that these particles have a very low coagulation rate in
combustion conditions (Minutolo et al., 1999). Because
of their small size and low coagulation rate (or sticking
probability) this material would not be well captured by
®lters and may not stick to the walls of a cold trap. Since
the cool walls of the probe and sample lines are, in e€ect, a
sort of ®lter removing sticky material from our samples,
we collect only the material that does not stick to them
but is captured in the condensers. In this way, we isolate
the 2±3 nm organic particles produced in ¯ames, and
analyze their spectral features. In studies that capture all
condensable material formed in ¯ames, the spectral fea-
tures of the more sticky material that remains in the lines
and probe mask the spectra of these nanoparticles.
By using water as a solvent, we are looking speci®cally
at the material that is partially soluble in water. The fact
that the water traps collect material with spectra similar
to that in ¯ame suggests the 2±3 nm organic particles
formed in ¯ame are partially water soluble. Furthermore,
since water is able to ¯ush the material that absorbs/
¯uoresces in the UV but not in the visible from the probe
sampling lines, we conclude that the 2±3 nm particles
associated with the broad band UV extinction/¯uores-
cence are more water soluble than the material that ab-
sorbs and ¯uoresces in the visible which is associated
with the more aromatic or soot-like particles.
The other main di€erence between in situ and extra
situ spectra is that the in situ ¯uorescence spectra pos-
sess visible ¯uorescence which is not observed in any of
the extra situ samples collected in water or with less
polar solvents. At high temperatures, ¯uorescence bands
can be shifted to higher wavelengths. Speci®cally, the
¯uorescence peak of naphthalene is shifted about 50 nm
into the visible going from room to ¯ame temperature
(Ossler, 1999). However, it is unlikely that the ¯uores-
cence in the visible observed in situ is solely a temper-
ature e€ect because the UV ¯uorescence is also observed
simultaneously with the visible ¯uorescence in situ and
they have di€erent trends with z. Also, the shift between
the UV and visible ¯uorescence is too large (200 nm) to
be an e€ect of temperature. The visible feature of the in
situ spectra can instead be attributed to more aromatic
particles that are ®ltered by the sampling line and are
not collected in the condensers.
The main di€erence between in situ and extra situ
uorescence spectra is the presence in the extra situ
¯
spectra of a banded structure, which is coincident with
that of naphthalene and absent in the spectra obtained
in situ at high temperature. This feature is seen only in
the ®rst condenser.
One possible hypothesis for this spectral feature is
that the increase of QE at low temperature with respect
to the ¯ame temperature is much larger for naphthalene
than for the 2±3 nm particles, and the capturing e-
ciency for naphthalene is higher. We observed that the
QE for the broad band UV ¯uorescence attributed to the
2
±3 nm particles is one order of magnitude lower in situ
at high temperature with respect to that observed extra
situ. It is dicult to measure directly the change in QE
of a pure PAH since it is not thermally stable and de-
composes at temperatures much lower than ¯ame tem-
peratures. However, on the basis of a recent study that
measured the lifetime of ¯uorescence (which is linearly
related to the QE) of naphthalene in a wide range of
temperature (Ossler, 1999), it seems probable that the
QE of naphthalene would decrease more than one order
of magnitude changing from room to ¯ame tempera-
5. Conclusions and implications
In this paper, we show that extra situ sampling can
isolate the 2±3 nm sized organic particulate, which is
UV-absorbing and transparent to visible radiation, from
other material generated in ¯ames. Both the form and

L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
679
the trend with height of the UV absorption and ¯uo-
rescence spectra of extra situ samples are similar to the
spectra observed in situ from rich ethylene air ¯ames.
The intensity of the broad band UV ¯uorescence spec-
trum associated with 2 nm OC particulate does not
change substantially with height. It is present even after
the formation of soot, which implies that these OC
particles can persist past the presooting regions of pre-
mixed ¯ames and may be present in the exhaust of
combustion systems. This result agrees with the ®ndings
of material with similar spectra in the exhaust of engines
and industrial sized ¯ames.
Beretta, F., Cincotti, V., D'Alessio, A., Menna, P, 1985.
Ultraviolet and visible ¯uorescence in the ¯uorescence in the
fuel pyrolysis regions of gaseous di€usion ¯ames. Combust.
Flame 61, 211±218.
Beretta, F., D'Alessio, A., D'Orsi, A., Minutolo, P., 1992. UV
and visible laser excited ¯uorescence from rich premixed
and di€usion ¯ames. Combust. Sci. Technol. 85, 455±470.
Black, A.P., Christman, R.F., 1963. Characteristics of colored
surface waters. J. Am. Water Works Assoc. 55, 753±770.
Borghese, A., Merola, S.S., 1998. Detection of extremely ®ne
carbonaceous particles in the exhausts of diesel and spark-
ignited IC engines, by means of broadband extinction and
scattering spectroscopy in the UV band 190 nm±400 nm.
In: Proceedings of the 27th International Symposium on
Combustion. The Combustion Institute, pp. 2101±2109.
Buchta, C., D'Alessio, A., D'Anna, A., Gambi, G., Minutolo,
P., Russo, S., 1995. The optical characterization of high
molecular mass carbonaceous structures produced in pre-
mixed laminar ¯ames across the soot threshold limit. Planet.
Space Sci. 43, 1227±1232.
These combustion generated organic particles are
more soluble in water than both soot particles and the
more mature aromatic particles transforming into soot.
The fact that this material interacts with water has im-
plications for both the e€ects of this material on human
health and on the radiative balance of the atmosphere.
UV extinction and ¯uorescence spectra similar to those
we observe here have been observed in atmospheric
studies and in atmospheric water (Black and Christman,
Ciajolo, A., D'Anna, A., Barbella, R., Tregrossi, A., 1994. The
formation of aromatic carbon in sooting ethylene ¯ames.
In: Proceedings of the 25th International Symposium on
Combustion. The Combustion Institute, pp. 679±685.
Cruz, C.N., Pandis, S.N., 1997. A study of the ability of pure
secondary organic aerosol to act as cloud condensation
nuclei. Atmos. Environ. 31, 2205±2214.
1
963; Perry et al., 1984; Havers et al., 1998; Zappoli
et al., 1999). Recently this UV absorption/¯uorescence
in water was attributed to humic-like material with size
ꢁ
3000 amu coming from biomass burning (Zappoli
D'Alessio, A., D'Anna, A., Gambi, G., Minutolo, P., 1998. The
spectroscopic characterization of UV absorbing nanoparti-
cles in fuel rich soot forming ¯ames. J. Aerosol Sci. 29, 397±
et al., 1999). The result that we can isolate combustion
generated material collected in water with similar spec-
troscopic features implies that direct emissions of these
409.
2
nm OC particles from combustion sources other than
Dobbins, R.A., Subramaniasivam, H., 1994. Soot precursor
particles in ¯ames. In: Bockhorn, H. (Ed.), Soot Formation
in Combustion. Springer, Heidelberg, pp. 290±301.
biomass may be an unaccounted source for the humic-
like substances found in atmospheric waters.
Dobbins, R.A., Fletcher, R.A., Lu, W., 1995. Combust. Flame
A recent study using a similar collection technique to
estimate the amount of OC particulate in the exhausts of
engines and stationary sources shows that motor vehi-
cles are likely an important source (Merola et al., 1999).
Because of their small size and low tendency to coagu-
late, the detection of this material is dicult. Atmo-
spheric studies using ®lters or mechanical devices such as
the particle mobility analyzers that measure well parti-
cles with diameters greater than 10 nm would not detect
this material.
100, 301±309.
Gambi, G., 1996. Cinetica di Formazione di Strutture Car-
boniose di Alta Massa Molecolare in Fiamme, Ph.D. Thesis.
Universita degli Studi di Napoli, Federico II, Napoli.
Havers, N., Burba, P., Lambert, J., Klockow, D., 1998.
Spectroscopic characterization of humic-like substances in
airborne particulate matter. J. Atmos. Chem. 29, 45±54.
Hildemann, L.M., Markowski, G.R., Cass, G.R., 1991. Chem-
ical composition of emissions from urban sources of ®ne
organic aerosol. Environ. Sci. Technol. 25, 744±759.
Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit,
B.R.T., 1994. Seasonal trends in Los Angeles ambient
organic aerosol observed by high-resolution gas chroma-
tography. Aersol Sci. Technol. 20, 303±317.
Acknowledgements
Karcher, W., 1992. Spectral Atlas of Polycyclic Aromatic
Compounds. Kluwer, Dordrecht.
Koylu, U.O., McEnally, C.S., Rosner, D.E., Pfe€erle, L.D.,
Support for L.A. Sgro by the US Fulbright Fellow-
ship is acknowledged. The authors thank Anna Ciajolo
and Rosalba Barbella for helpful discussions regarding
the experiments and the use of the spectrometer and
spectro¯uorometer, and Angela Violi for help with
CHEMKIN.
1997. Simultaneous measurements of soot volume fraction
and particle size/measurements in ¯ames using thermopho-
retic sampling technique. Combust. Flame 110, 494±507.
Lee, M.L., Novotny, M.V., Bartle, K.D., 1981. Analytical
Chemistry of Polycyclic Aromatic Compounds. Academic
Press, New York.
References
Merola, S.S., Allouis, C., Gambi, G., Beretta, F., Borghese, A.,
D'Alessio, A., 1999. Di€erential analysis of exhaust emitted
by reciprocating engines and stationary burners, by means
Bengtsson, P., Alden, M., 1989. Application of a pulsed laser for
soot measurements in premixed ¯ames. Appl. Phys. B 48, 155.

6
80
L.A. Sgro et al. / Chemosphere 42 (2001) 671±680
of UV-visible extinction and scattering spectroscopy. In: ultra®ne particles. Am. J. Respir. Crit. Care Med. 155,
Proceedings of the Sixth International Congress on Toxic
Combustion Byproducts, Karlsruhe, Germany, 27±30 June.
1376±1383.
Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R.,
1993. Sources of ®ne organic aerosol. 2. noncatalyst and
catalyst-equipped automobiles and heavy-duty diesel
trucks.. Environ. Sci. Technol. 27, 636±651.
Minutolo, P., Gambi, G., D'Alessio, A., 1996. The optical band
gap model in the interpretation of the UV-visible absorption
spectra of rich premixed ¯ames. In: Proceedings of the 26th
International Symposium on Combustion, The Combustion
Institute, pp. 951±957.
Santoro, R.J., Semerjian, H.G., 1984. Soot formation in
di€usion ¯ames: ¯ow rate, fuel species, and temperature
e€ects. In: Proceedings of the 20th International Symposium
on Combusiton. The Combustion Institute, pp. 997±1006.
Saxena, P., Hildemann, L.M., McMurry, P.H., Seinfeld, J.H.,
1995. Organics alter hygroscopic behavior of atmospheric
particles. J. Geophys. Res. 100, 18755±18770.
Minutolo, P., Gambi, G., D'Alessio, A., 1998. Properties of
carbonaceous nanoparticles in ¯at premixed
2 4
C H /Air
¯
ames with C/O ranging from 0.4 to soot appearance limit.
In: Proceedings of the 27th International Symposium on
Combustion.
Minutolo, P., Gambi, G., D'Alessio, A., Carlucci, S., 1999.
Spectroscopic characterisation of carbonaceous nanoparti-
cles in premixed ¯ames. Atmos. Environ. 33, 2725±2732.
Minutolo, P., Gambi, G., D'Alessio, A., D'Anna, A., 1994.
Optical and spectroscopic characterization of rich premixed
Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and
Physics: From Air Pollution to Climate Change. Wiley,
New York.
Sheu, E.Y., Mullins, O.C., 1995. Asphaltenes Fundamentals
and Applications. Plenum Press, New York.
¯ames across the soot formation threshold. Combust. Sci.
Technol. 101, 311±325.
Ni, T., Gupta, S.B., Santoro, R.J., 1994. In: Proceedings of the
Vander Wal, R.L., Jensen, K.A., Choi, M.Y., 1997. Simulta-
neous laser-induced emission of soot and polycyclic aro-
matic hydrocarbons within
Combust. Flame 109, 399±414.
a gas-jet di€usion ¯ame.
25th International Symposium on Combustion. The Com-
bustion Institute, pp. 585±592.
Wesborg, B.L., Howard, J.B., Williams, G.C., 1973. Physical
mechanism in carbon formation in ¯ames. In: Proceedings
of the 14th International Symposium on Combustion. The
Combustion Institute, pp. 929±940.
Olten, N., Senkan, S., 1999. Atmospheric pressure ethylene
di€usion ¯ame. Combust. Flame 118, 500±507.
Ossler, F., 1999. Laser diagnostics in combustion ± elastic
scattering and picosecond laser-induced ¯uorescence, Ph.D.
Thesis, Lund University, Lund, Sweden.
Wilson, R., Spengler, J.D., 1996. Particles in our Air: Concen-
trations and Health E€ects. Harvard University Press,
Cambridge, MA.
Perry, M.E., Elzerman, A.W., Overcamp, T.J., 1984. Solubility
of atmospheric particulate matter. In: Durham, J.L. (Ed.),
Chemistry of Particles, Fogs, and Rain. Butterworths,
Stoneham, MA.
Zappoli, S., Andracchio, A., Fuzzi, S., Facchini, M.C.,
Gelencser, A., Kiss, G., Krivacsy, Z., Molnar, A., Mesza-
ros, E., Hansson, H.-C., Rosman, K., Zebuhr, Y., 1999.
Inorganic, organic and macromolecular components of ®ne
aerosol in di€erent areas of Europe in relation to their water
solubility. Atmos. Environ. 33, 2733±2743.
Petarca, L., Marconi, F., 1989. Fluorescence spectra polycyclic
aromatic specie in a N-heptane di€usion ¯ame. Combust.
Flame 78, 308±325.
Peters, A., Wichmann, H.E., Tuch, T., Heinrich, J., Heyder, J.,
1
997. Respiratory e€ects are associated with the number of