Residence time effect on fullerene yield in butadiene-based laser pyrolysis flame

Article abstract of DOI:10.1016/j.cplett.2003.08.010

A new route for fullerene synthesis by CO₂-laser pyrolysis of gas phase mixture is proposed. Small hydrocarbon molecules which absorb the laser radiation, such as butadiene, are mixed with nitrous oxide (N ₂O) as oxidizer. Such a mixture allows avoiding the use of a photosensitizer as SF₆ which causes contamination of the reaction zone and possibly influences the growth of fullerenic structures. This Letter also confirms the strong influence of the C/O atomic ratio in the mixture on the fullerene yield, and shows that residence time of the reactants in the pyrolysis flame and pressure influence dramatically the fullerene formation.

Full text of DOI:10.1016/j.cplett.2003.08.010

Chemical Physics Letters 379 (2003) 40–46
www.elsevier.com/locate/cplett
Residence time effect on fullerene yield in butadiene-based
laser pyrolysis flame
b
a
a
a,
*
F. Tenegal a,1, I. Voicu , X. Armand , N. Herlin-Boime , C. Reynaud
ꢀ ꢀ
a
Laboratoire Francis Perrin (CEA-CNRS URA 2453), Service des Photons, Atomes et Moleꢀcules, DSM, CEA Saclay,
91191 Gif Sur Yvette Cedex, France
b
National Institute for Lasers, Plasma and Radiation Physics, P.O. Box MG-36, Bucharest-Magurele, Romania
Received 2 April 2003; in final form 6 August 2003
Published online: 2 September 2003
Abstract
A new route for fullerene synthesis by CO2-laser pyrolysis of gas phase mixture is proposed. Small hydrocarbon
molecules which absorb the laser radiation, such as butadiene, are mixed with nitrous oxide (N2O) as oxidizer. Such a
mixture allows avoiding the use of a photosensitizer as SF6 which causes contamination of the reaction zone and
possibly influences the growth of fullerenic structures. This Letter also confirms the strong influence of the C/O atomic
ratio in the mixture on the fullerene yield, and shows that residence time of the reactants in the pyrolysis flame and
pressure influence dramatically the fullerene formation.
Ó 2003 Elsevier B.V. All rights reserved.
1. Introduction
was reported [7]. In particular, laser pyrolysis ap-
pears efficient for the production of higher fulle-
renes, which could be even more attractive for
applications.
Since their discovery in carbon cluster beams
[1], fullerenes have attracted much attention be-
cause of their potential use in medical, storage or
electronic applications. An efficient synthesis
method is still an important objective. Fullerenes
are mainly produced by graphite vaporization [2]
or from sooting flame [3–5]. Formation of fulle-
renes was also demonstrated by CO2-laser pyro-
lysis [6] and synthesis of macroscopic quantities
Laser pyrolysis is a flow method allowing the
formation of nanopowders by homogeneous nu-
cleation and growth in a reaction flame [8]. This
method differs from combustion by the heat sup-
ply to the reactants: a continuous energetic trans-
fer is ensured by the resonance between the laser
radiation and the infrared absorption band of one
of the precursors. In the case of fullerene produc-
tion, the precursor mixture usually contains a
photosensitizer which acts as a heat transfer agent
due to the limited absorption of carbon precur-
sors. Sulphur hexafluoride (SF6) absorbs strongly
the CO2-laser radiation and was used as transfer
* Corresponding author. Fax: +33-1-69-08-87-07.
E-mail address: creynaud@drecam.cea.fr (C. Reynaud).
1
^
Present address: CEA-Saclay, DRT-IdF, DECS/SE2M, bat.
117, 91191 Gif Sur Yvette Cedex, France.
0009-2614/$ - see front matter Ó 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.cplett.2003.08.010

F. Tꢀenꢀegal et al. / Chemical Physics Letters 379 (2003) 40–46
41
2. Experimental
agent. Although SF6 is characterized by a high
thermal and chemical stability, it was found to be
not chemically inert in the specific conditions of
laser pyrolysis [9], and could form radicals influ-
encing the reaction pattern [10].
The experimental set-up has been described
elsewhere [11]. The CO2-laser beam is focused by a
ZnSe lens at its intersection with the reactant flow
where power density up to 2.105 W/cm2 are
reached. Reactant gases are introduced through a
4 mm diameter nozzle and are mass-flow con-
trolled. The pressure is kept constant by a down-
stream control system. The temperature is
monitored with a monochromatic pyrometer. The
reactant flow was a hydrocarbon/oxidizer mixture.
Butadiene (C4H6) was chosen as carbon source,
since it absorbs efficiently the 10P26 laser line at
938.69 cmꢀ1, due to its CH2 wag mode at 907.8
cmꢀ1 [16], and led to a reaction flame under oxi-
dizing environment. N2O has been maintained as
oxidizing agent, since it gives better fullerene
production than molecular oxygen [9,17].
The effect of three parameters was studied:
residence time s, work pressure P, and C/O ratio
(Table 1). In the C253–C256 experiments, s was
varied by changing the total reactant flow without
changing the other parameters (P ¼ 400 Torr,
T ꢁ 1580 K). The C/O ratio was fixed to 1.2 since
this value appeared optimal previously [11]. To
evaluate s, the volume V of the reaction hot zone
was appreciated with the optical pyrometer, as the
flame volume having uniform brightness and so,
almost uniform temperature. The expression for s
is
In this Letter, fullerenes are synthesized by laser
pyrolysis of
a hydrocarbon/oxidizer mixture
without SF6. The objective is to study the fullerene
formation as a function of the main synthesis pa-
rameters. In a previous Letter [11], the C/O atomic
ratio in the precursor mixture was shown to in-
fluence strongly the fullerene yield. The highest
yields were obtained for a C/O ratio close to 1.2, as
in combustion experiments. Also, the C60/C70 ratio
was found to increase significantly for C/O ratio
decreasing down to 1.2. These effects have been
tentatively explained by a kinetic model [12] based
on a molecular growth process where the C/O ratio
is expected to influence the concentration of the
radicals involved in the edification of closed-cage
structures. This explanation will be checked here,
including C/O value under 1.2, to verify the
agreement with combustion data.
According to this explanation, residence time
in the flame is a sensitive parameter which could
lead to appreciable variations of higher fullerenes
yield (beyond C70). This parameter is valuable to
study both heterogeneous and homogeneous
nucleation [13]. In connection with the time re-
quired for fullerene formation, it could be
modulated experimentally to increase the fuller-
ene yield and particularly higher fullerenes. In
this Letter, this effect was studied by means of
total flow rate variation. The pressure is also a
crucial parameter. Fullerenes appear in a bell-
shaped window of thermodynamic stability,
limited by polycyclic aromatic hydrocarbons
(PAH) at low temperatures (<1500 K) and
mainly by acetylene (C2H2) at high temperatures
(>2500 K). The pressure reduction enlarges this
stability area and improves the conversion rates
[14,15]. Laser pyrolysis performed at lower
pressures could lead to larger fullerene yields,
too. Pressure effects will be discussed in relation
with residence time effects. Results obtained ei-
ther by total flow or pressure variations will be
compared to distinguish the respective influence
of these parameters.
V T0 P
D0 T P0
s ¼
;
ð1Þ
where D0 is the total flow delivered by the mass
flow-meters at standard pressure and temperature.
A strong correlation between V and D0 was ob-
served (with a quadratic-like behaviour), by ex-
tension of the hot zone along the direction of the
reactant flow. Calculation of s accounts for these
variations. V was the largest for the largest flow
(C253). For higher flow, strong flame instabilities
were observed. Then, it was not possible by flow
variation to reduce s below 7 ms. Contrary to the
case of the photosensitised system [11], the flame
temperature was not under laser power control but
was mainly influenced by pressure and C/O ratio
variations.

Table 1
Synthesis parameters for the samples obtained from C4H6/N2O mixtures and yields determined by HPLC for C60 + C70 and higher fullerenes
Sample Studied
parameter
Flow rate
(cm3/min)
Velocity C/O
ratio
T
(K)
P
(Torr) (ms)
s
Soot
(g/h)
C60 + C70
(%)
C60/C70 C70 þ 2n (%)a in
(cm/s)
ratio
total fullerenes
C4H6 N2O
of total in soot in total
carbon
fullerenes
C253
C254
C255
C256
Residence
time s
179
138
91
570
460
295
188
99
79
51
32
1.25
1.2
1583
1573
1588
1578
400
400
400
400
7
7
1.2
0.5
0.2
0.5
0.04
0.06
0.8
2.1
0.2
0.03
88.2
88.7
86.4
99.9
1.3
1.5
1
6.9
6.8
3.1
–
1.23
1.17
8
11
0.01
<0.01
55
0.8
C257
C258
P at
fixed s
89
179
289
588
50
101
1.23
1.22
1573
1603
280
550
5
9
0.7
0.8
0.04
0.02
0.7
0.6
88.3
89.9
0.7
1.3
6.5
4.6
C259
C260
P at
fixed flow
138
138
470
470
81
81
1.17
1.17
1388
1618
280
550
3
11
0.3
0.7
0.07
0.03
4.4
0.7
92.1
89.7
2.1
1.3
3.6
4.3
C261
C262
C/O ratio
(with C254)
110
165
487
425
79
78
0.9
1.55
1378
1663
400
400
3
7
0.2
3
0.03
<0.01
2.1
0.01
92.9
58.6
1.6
0.6
3.3
16.4
a The sum of the C70 + C60 and C70 þ 2n percentages is not equal to 100 due to the presence of extra peaks attributed to fullerene-oxide structures.

F. Tꢀenꢀegal et al. / Chemical Physics Letters 379 (2003) 40–46
43
3.1. Effect of the C/O ratio
The aim of the C257–C260 experiments was to
check the pressure influence by comparison with
C254. In the first set (C257, C258), P was changed
trying to keep s constant by flow adjustment, while
in C259 and C260 only P was varied, resulting in s
variations. Unfortunately, the temperature could
not be maintained constant. Finally, the experi-
ments C261–C262 had the goal to check C/O ratio
effects.
Fullerenes were extracted from soot with tolu-
ene by the Soxhlet continuous extraction proce-
dure. Separation and quantification of C60, C70
and higher fullerenes were performed by High
Performance Liquid Chromatography (HPLC).
Calibration of C60 and C70 contributions was made
with solutions of known concentration of pure
fullerenes. The three-dimensional analysis of the
chromatograms (wavelength, elution time and
absorbency) was obtained by a UV–Visible diode
array detector, coupled with HPLC. This allowed
the identification and comparison of the respective
amount of the larger fullerenes (above C70) in the
soot through their normalised absorbency spectra
in the UV–Visible range and areas of elution peaks
normalised to the same mass (1 g) of soot. Ex-
tracted samples were analysed by Fourier Trans-
form Infra-Red (FTIR) spectrometry.
The results obtained from increasing C/O ratio
(runs C261, C254 and C262, Table 1 and Fig. 1)
confirm that this ratio has a strong effect on the
fullerene formation and that C/O ¼ 1.2 seems ac-
tually an optimum value. When C/O is reduced
down to 0.9 (run C261), similar fullerene weight in
soot (2.1%) is obtained than for C/O ¼ 1.2 (run
C254), but the soot efficiency is smaller (0.2 versus
0.5 g/h) and the C60 + C70 mass yield as well. In
Fig. 1, the results obtained with the C6H6/N2O/SF6
system [11] are also reported. For C/O ¼ 1.2, both
fullerene weight in soot and C60/C70 ratio are
similar, which suggests a negligible participation of
3. Results and discussion
The data obtained from HPLC measurements
(Table 1) show firstly that fullerenes are present in
soot produced by laser pyrolysis of butadiene-
based mixtures. According to the experimental
conditions, the C60 + C70 yield varies from 58.6%
to 99.9% of the total fullerene yield. Higher fulle-
renes (beyond C70) are also observed, up to
16.45%. Signatures attributed to fullerene-oxide
adducts were also detected on chromatograms.
The best C60 + C70 yield (0.06% of total consumed
carbon) is obtained with a residence time around
7 ms (C254) at T ꢁ 1600 K, P ¼ 400 Torr and
C/O ¼ 1.2. For the C6H6/N2O/SF6 system [11],
higher values (up to 0.23%) were obtained in
similar conditions but with a lower residence time
(1 ms). It is a first indication that residence time is
a determinant parameter.
Fig. 1. Comparison of the effect of C/O ratio between laser
pyrolysis and combustion (from Richter et al. [5]). (a) Fullerene
(C60 + C70) yield in soot and (b) C60/C70 ratio. Open triangles
are for combustion results, open diamonds for benzene-based
mixtures for laser pyrolysis [11] and filled squares for present
results (laser pyrolysis of butadiene-based mixtures).

44
F. Tꢀenꢀegal et al. / Chemical Physics Letters 379 (2003) 40–46
radicals coming from SF6 dissociation on fullerene
growth. Values recorded by Richter et al. [5] for
combustion flames are also reported for compari-
son in Fig. 1. The good agreement of the pyrolysis
results with combustion data is confirmed.
The typical features of C60 (527, 577, 1183, 1428
cmꢀ1) and C70 (535, 565, 578, 641, 674, 795, 1134,
1430 cmꢀ1) are identified in FTIR spectra of tol-
uene extracts (Fig. 2). While for runs C254 and
C261 these features are evident, the absorption
bands assigned to PAH (aromatic C–H deforma-
tions in the 700–900 cmꢀ1 range and CC stretch
mode at 1640 cmꢀ1) are more underlined for run
C262 where the C/O value is the highest.
The present results are consistent with the ki-
netic interpretation suggested previously [11]. C/O
variation changes the concentrations of radicals
involved in the nucleation and growth of fullerenes
(H, H2, C2H2) leading to a modification of the
growth kinetic in the flame. Nevertheless, both
C60 + C70 yield and C60/C70 ratio seem to be more
sensitive to C/O variation in the case of butadiene
pyrolysis compared to benzene-based mixture py-
rolysis (Fig. 1). The C/O variation from 1.2 (C254)
to 1.55 (C262) provokes a steeper decrease of
C60 + C70 weight in soot (2.1–0.01%) and the C60/
C70 ratio falls much more rapidly (1.5–0.6). This
could be related to the higher H content in the
butadiene case, and possibly reflects an influence
of the C/O ratio through its effect on hydrogen
concentration in the reaction volume.
3.2. Effect of the residence time
In the experiments C253–C256 (C/Oꢁ1.2), the
flow rate decrease leads to a s increase from 7 to 11
ms. Since the hydrocarbon flow is not constant, the
analysis of fullerene production will be done only in
term of fullerene weight in soot or toluene extract.
The striking changes in the FTIR spectra (Fig. 3)
show that s influences strongly the fullerene for-
mation. C60 and C70 yields are reported in Fig. 4a as
a function of s. The results display similarities with
Fig. 3. FTIR absorption spectra for four extract samples
for increasing residence time (C253, C254, C255, C256) at
C/O ¼ 1.2. Vertical arrows on the bottom axis indicate the
positions of IR bands for C60 (open triangle) and C70 (filled
triangle), respectively.
Fig. 2. FTIR absorption spectra for three extract samples with
C/O ratio of 1.55 (C262), 1.2 (C254) and 0.9 (C261). Vertical
arrows on the bottom axis indicate the positions of IR bands
for C60 (open triangle) and C70 (filled triangle), respectively.

F. Tꢀenꢀegal et al. / Chemical Physics Letters 379 (2003) 40–46
45
runs, we have tried to keep s constant by total flow
adjustment. Unfortunately, as discussed above, the
correlation between P, the hot zone volume V and
s, is such that s changes each time P is changed. In
C259 and C260 runs, P was changed but with
constant total flow, thus s variations are even more
important. Fig. 4a reports the dependence of the
C60 and C70 yields with variation of s induced by
flow variations at different pressures. The strongest
yields are obtained for the lowest s. However, at
550 Torr, the yields are quite constant between 9
and 11 ms but stronger than the yields at 400 Torr.
This could be explained by a higher flame tem-
perature at 550 Torr.
In Fig. 4b, one can see that the C60 + C70 yield
evolves differently with s depending on whether s
is varied by changing the pressure (at constant
reactant flow) or the reactant flow (at constant
pressure). The yields obtained by pressure varia-
tion are higher than those obtained by flow vari-
ation. Lower s are better for fullerene formation as
well as low pressures. Unfortunately, it was not
possible to investigate s shorter than 7 ms by flow
variation due to flame instabilities. The pressure
reduction is more efficient to reduce s down to 3 ms
where the highest yield is observed. A similar in-
crease of the fullerene yield was observed previ-
ously for benzene-based pyrolysis [11] when s and
T were reduced respectively to about 1 ms and
1650 K. This result was interpreted by an inhibit-
ing effect of SF6 at high temperature. The present
results suggest that this inhibiting effect is weaker
than supposed.
According to global equilibrium prediction for
C60 and C70 [15], when P decreases, both threshold
and maximum temperatures of the bell-shaped
window of thermodynamic stability are shifted
towards lower values. At the same time, there is an
abrupt change (over a range less than 100 K) from
PAH to fullerenes when temperature increases
[15]. These thermodynamic considerations are in
agreement with our results showing a large yield at
low pressure for the lowest measured temperature
(sample C259, Fig. 4b). In that case, C60 + C70
represent 4.4% of the soot which is close to the
maximum value reported in combustion experi-
ments [5] at much lower pressures (around 50
Torr, see Fig. 1a). This value is higher than those
Fig. 4. Effect of the residence time on fullerene yield. (a) C60
(open symbols) and C70 (filled symbols) yields from flow vari-
ations for increasing pressures: 280 Torr (circles), 400 Torr
(triangles) and 550 Torr (diamonds); (b) (C60 + C70) yield from
pressure variations (filled diamonds) and flow variations (open
squares).
the C/O effect. As s increases, the yield decreases and
the C60/C70 ratio also decreases. From the kinetic
model [12], large residence times are more favour-
able to fullerene formation and particularly C70. It is
consistent with the present decrease of the C60/C70
ratio when s increases, but the expected increase of
the total yield is not observed. On the contrary, a
strong decrease is observed by HPLC measurements
and confirmed by the FTIR spectra (Fig. 3). This
yield decrease is not well understood. An effect
coming from oxidation of the closed cage structure
could be invoked.
3.3. Effect of the pressure
C257–C260 runs were performed to evaluate
the pressure influence (Table 1). In C257 and C258

46
F. Tꢀenꢀegal et al. / Chemical Physics Letters 379 (2003) 40–46
obtained by laser pyrolysis of benzene-based
mixtures [11].
cess is weaker than suggested previously. The
yields (% of the soot) are constant through varia-
tion of C/O from 1.2 to 0.9 but the production is
reduced due to a decrease of the soot production at
C/O ¼ 0.9. Then, the expected effect of the C/O
ratio on kinetic growth is confirmed by study of
the residence time which appears definitely as a
very sensitive parameter for fullerene formation.
The lowest residence times are more favourable to
C60 and C70 formation as suggested in the benzene-
based system study. Low pressures are also found
to favour the C60 and C70 formation in agreement
with thermodynamic predictions.
Another interesting effect of the pressure reduc-
tion is that the C60 yield exceeds the C70 yield at a
lower residence time (Fig. 4a). At 550 Torr, it is lo-
cated beyond 12 ms while at 400 Torr, the value is
close to 8 ms and near 4 ms at 280 Torr. Also, the C60
yield variations with s are more rapid at low pres-
sure compared to C70 yields. Therefore, the fullerene
yield seems to be much more sensitive to residence
time variations at low than at high pressure.
3.4. Higher fullerenes
The yield of the higher fullerenes (C74, C76, . . .)
is the largest (16.4%) for C262 run (Table 1), and
goes with very low C60 + C70 yields, suggesting
quite different stability domains for higher fulle-
renes and C60 + C70. This effect could be due to
different C/O ratio (comparison of C254 and
C262). The edification of large closed-cage needs
the addition of large flat carbonaceous fragments
to cage nuclei. When C/O is too low, these species
are more oxidised and their proportion decreases.
By contrast, low C/O values are favourable to
radicals involved in the edification of C60 and C70
nuclei (carbon sheet including 5 member rings).
Also, high C/O value leads to high soot efficiency
(3 g/h), suggesting a prevalent heterogeneous
growth mechanism. Soot particles could act as
heat carriers that prolong the growing time and
favour the higher fullerenes formation [18]. Thus,
less oxidizing environments favour higher molec-
ular weight. Flame temperature can also be in-
voked since the highest yield corresponds to the
highest temperature. This parameter could be rel-
evant for higher fullerene formation.
References
[1] H.W. Kroto, J.R. Heath, S.C. OÕBrien, R.F. Curl, R.E.
Smalley, Nature 318 (1985) 162.
[2] W. Kraetschmer, L.D. Lamb, K. Fostiropoulos, D.R.
Huffman, Nature 347 (1990) 354.
[3] J.B. Howard, J.T. McKinnon, Y. Makarovsky, A.L.
Lafleur, E. Johnson, Nature 352 (1991) 139.
[4] H. Richter, A. Fonseca, S.C. Emberson, J.M. Gilles, J.B.
Nagy, P.A. Thiry, R. Caudano, A.A. Lucas, J. Chim. Phys.
92 (1995) 1272.
[5] H. Richter, A. Fonseca, S.C. Emberson, J.M. Gilles, J.B.
Nagy, A.A. Lucas, Mat. Res. Symp. Proc. 359 (1995) 17.
[6] M. Ehbrecht, M. Faerber, F. Rohmund, V.V. Smirnov, O.
Stelmach, F. Huisken, Chem. Phys. Lett. 214 (1993) 34.
[7] I. Voicu, X. Armand, M. Cauchetier, N. Herlin, S.
Bourcier, Chem. Phys. Lett. 256 (1996) 261.
[8] W.R. Cannon, S.C. Danforth, J.H. Flint, J.S. Haggerty,
R.M. Marra, J. Am. Ceram. Soc. 65 (1982) 324.
[9] R. Alexandrescu, X. Armand, M. Cauchetier, N. Herlin, S.
Petcu, I. Voicu, Carbon 36 (1998) 1285.
[10] R.J. Lagow, J.J. Kompa, H.C. Wei, S.L. Battle, J.W.
Genge, D.A. Laude, C.J. Harper, R. Bau, R.C. Stevens,
J.F. Haw, E. Munson, Science 267 (1995) 362.
[11] F. Tenegal, S. Petcu, N. Herlin-Boime, X. Armand, M.
Mayne, C. Reynaud, Chem. Phys. Lett. 335 (2001) 155.
[12] C.J. Pope, J.A. Marr, J.B. Howard, J. Phys. Chem. 97
(1993) 11001.
[13] W. Benzinger, A. Becker, K.J. Huttinger, Carbon 34 (1996)
957.
4. Conclusion
[14] C. Vahlas, A. Kacheva, M.L. Hitchman, Ph. Rocabois,
J. Electrochem. Soc. 146 (1999) 2752.
Experiments using butadiene-based mixtures
for fullerene synthesis by laser driven-gas phase
pyrolysis were successful. The study of the C/O
ratio effect confirms previous results obtained in
SF6-photosensitised benzene-based mixtures, with
an optimum value of C/O around 1.2 for C60 and
C70 production. Thus, the role of SF6 in this pro-
[15] C.J. Pope, J.B. Howard, Tetrahedron 52 (1996) 5161.
[16] T. Shimanouchi, in: Tables of Molecular Vibrational Fre-
quencies, vol. I, Lawrence M. Kushner, Acting Director.
[17] S. Petcu, M. Cauchetier, X. Armand, I. Voicu, R.
Alexandrescu, Comb. Flame 122 (2000) 500.
[18] Z.J. Shi, X.H. Zhou, Z.X. Jin, Z.N. Gu, J. Wang, S.Q.
Feng, X.L. Xu, Z.Q. Liu, Solid State Commun. 97 (1996)
371.