J. Phys. Chem. 1995,99, 13509-13512 C2
Formation during High-Temperature Pyrolysis of Fullerene
13509 c 6 0
in Shock Waves
Thorsten Sommer, Thomas Kruse, and Paul Roth* Institut f i r Verbrennung und Gasdynamik, Universitat Duisburg, 47048 Duisburg, Germany Received: March 15, 1995; In Final Form: June 9, 1995@
The formation of CZduring thermal decomposition of fullerene c 6 0 was studied behind shock waves by time dependent absorption and emission measurements. The experiments were performed in the temperature range 2450 5 T 5 3450 K at pressures in the range 1.1 5 p 5 2.0 bar. C2 concentration profiles were measured with a ring dye laser spectrometer at Y = 21393.12 cm-I. Simultaneously, the time behavior of spectrally resolved light emitted from the shock-heated aerosol was recorded by an intensified CCD camera. A simplified reaction mechanism with both CZformation and consumption reactions was proposed, and the experimental results were discussed in terms of kinetic parameters. A rate coefficient of the initial reaction c 6 0 Ar CZ CSS Ar,kl = 2.7 x 10” exp(-495 kJlmoYRT) cm3 mol-’ s-’, was determined. Two different models for the further decomposition of fullerene fragments were discussed and compared with computer simulations.
+
+ +
-
Introduction The existence of fullerenes as a new carbon modification has been established during the last few years. The Cm molecule discovered to be stable by Kroto et al. in 1985’ was proposed to be a soccer-ball-shaped cage where the carbon atoms are placed at the edges of 12 pentagons and 20 hexagons. The proposed structure was confirmed in an immense number of experimental studies. In contrast to many investigations concerning the stability of c 6 0 molecules, only a few papers deal with its chemical kinetics during decomposition. O’Brien et al? found that the decisive step in fragmentation during photolysis of fullerene is the loss of CZ followed by a rearrangement of the remaining fragments. The aim of the present work was to study the formation of C2 during pyrolysis of fullerene Cm at high temperatures behind reflected shock waves and to describe the experimental results by a kinetic model. Two different diagnostic tools were applied to follow the shock-induced reactions in c6d.k mixtures. A ring dye laser spectrometer was used to detect time dependent C2 formation by absorption measurements in the (1-0) band of the d311,-a311, transition. The line strength data are available from the l i t e r a t ~ r e . ~Simultaneously .~ the light emitted from the shock-heated c6d.k mixture was dispersed by a spectrograph and the time behavior of the spectrally resolved emission was recorded by an intensified CCD camera.
Experimental Section The experimental setup used to study fullerene behavior at high temperatures is shown schematically in Figure 1. It consists of an aerosol generator for dispersing the fullerene powder, the main shock tube for heating the C6o-containing aerosol to high temperatures, a ring dye laser spectrometer for the measurement of C2 absorption signals, a spectrograph with an intensified CCD camera for the detection of spectrally resolved and time-resolved emission (see upper part of Figure l), and an IR diode laser system. The Cm powder was dispersed in argon by an expansion wave driven aerosol g e n e r a t ~ r .It~ consists of a glass vessel of V = 3 x lo4 cm3 connected to a glass tube of 5 cm inner diameter, which was separated by a thin diaphragm from a high-pressure tube of the same diameter. A thin dispersion plate was @
Abstract published in Advance ACS Abstracts, August 15, 1995.
0022-365419512099-13509$09.OO/O
l l
n
I
Figure 1. Schematics of the experimental setup including shock tube and aerosol generator and different optical diagnostics.
positioned near the diaphragm, which carried for every experiment about 100 mg of the fullerene powder. Before the powder dispersion was begun, the low-pressure section of the aerosol generator was evacuated to a pressure of about 5 x mbar. Subsequently, argon was filled into the high-pressure tube. After bursting of the diaphragm, the high-pressure gas argon expanded in the form of a supersonic expansion wave into the evacuated vessel, thus dispersing the fullerene powder and bringing it into aerosol form. The fullerene Cm powder used during the present experiments was supplied by Hoechst AG, Germany; it contained less than 1% of C ~ O .The generated aerosol was subsequently filled into the running section of the main shock tube, which was used as an isothermal-isobaric reactor. The stainless steel shock tube used has an internal diameter of 8 cm, a total length of 11 m, and a running section of 7.2 m in length. The initial pressure of the aerosol measured by calibrated diaphragm-type pressure transducers was in the range of 17 to 30 mbar. The shock velocity was measured by piezoelectric pressure transducers positioned at known distances 0 1995 American Chemical Society
Sommer et al.
13510 J. Phys. Chem., Vol. 99, No. 36, 1995 along the shock tube axis. The postshock temperature and pressure were calculated from the incident shock speed by applying one-dimensional conservation equations. The shock wave running into the C6dAr aerosol caused a sudden temperature and pressure increase. First the evaporation of the solid fullerene particles was initiated in less than 30 ps. Subsequently, the fullerene vapor started to decompose. The optical setup for measuring time dependent C2 absorption consists of a ring dye laser spectrometer, which was connected to the measurement section of the main shock tube by an optical fiber. The time dependent C2 absorption during high-temperature pyrolysis of Cm was determined by measuring the laser intensity with a fast Si detector after the beam passed through the shock tube. A second Si detector was used to record the incident laser light intensity. The laser was fixed to the line center wavenumber at Y = 21 393.12 cm-I, corresponding to an overlap of the R1(37), R2(36), and R3(35) rotational lines in the (1-0) band of the d 311,-a 311utransition of C2. The absorption profiles were stored in a transient recorder and transmitted to a personal computer for storage and data processing. To achieve a low detection limit, it was necessary to stabilize both the frequency and the intensity of the dye laser. A complete description of technical details of the spectrometer is given in ref 6. The light emitted from the shock-heated C6o-containing aerosol was focused via a light fiber on the entrance slit of a l / ~ m spectrograph which has a spectral range of M = 250 nm and a resolution of 1.3 nm. The spectrally resolved light was detected by an intensified CCD camera, allowing a fast time shift of the optical information. For this purpose, only a few lines at the top of the CCD sensor chip were illuminated for a certain time interval. The stored spectral emission was stepwise transferred line by line into the dark zone of the CCD chip, which served as a memory. In this way, the spectral as well as the temporal behavior of the emitted light could be recorded. Spectral characteristics of the quantum efficiency of the intensifier and of the diffraction grid were taken into account. For a quantitative data interpretation in terms of kinetic parameters, the initial concentration of c 6 0 was required. For this purpose fullerene C a powder was dispersed in Ar containing about 6% of 0 2 . The aerosol mixtures were heated by the shock wave to conditions similar to the pyrolysis experiments and resulted in a complete oxidation of c 6 0 by 0 2 . The gas phase oxidation products CO and C02 were quantitatively measured by the tunable infrared diode laser. From these measurements, the initial c60 gas phase concentration could be deduced.
Results The measurements of both the spectral emission of hightemperature C d A r mixtures and the absorption of C2 were performed behind reflected shock waves at temperatures between 2450 and 3450 K and at pressures between 1.1 and 2.0 bar. Detectable C2 absorption signals were recorded only for temperatures above 2650 K. Representative time-resolved absorption profiles converted to Cz concentrations via LambertBeer's law are shown in Figure 2. It is obvious that both the formation rate of C2 after shock arrival and the C2 peak concentration depend on temperature. Higher temperatures lead to a higher yield of C2, and also the maximum concentration is reached at shorter times. In the temperature range of detectable C2 concentration, this time of increase and the peak concentration varied from 600 ps and 1 ppm of C2 at 2650 K up to 10 ps and 20 ppm of C2 at 3450 K. At temperatures T > 2800 K, the measured C2 absorption profiles show a characteristic decrease after reaching maximum concentration.
10,
'
,
F i1
'
'
I
I
'
1
T = 2765 K
?'
O k - , # 0
1
-
,
I
\
200
400
1
600
1
800
t lys Figure 2. Examples of CZconcentrations measured during the pyrolysis of C60 at different postshock temperatures.
350
400
450
500
550
600
wavelength I nm Figure 3. Examples of spectral emission of shock-heated C d A r mixtures at two postshock temperatures. t = 150 p s after shock anival.
The emission spectra of high-temperature C d A r mixtures were recorded in the wavelength range between 360 and 600 nm. Typical examples of spectrally resolved emission obtained from two different shock tube experiments at time t = 150 ps after shock arrival are illustrated in Figure 3. The characteristic band structures with peaks at 1 467, 516, and 560 nm were identified as different progressions (Au = -1, 0, f l ) of the d 311,-a 311u transition of C2. The measured signals in the range between 360 and 420 nm correspond to the A 'n,-X 'Es+ transition of C3.' Similar spectra were obtained for all experiments with temperatures T > 2650 K, whereas no characteristic emission of C2 or C3 was observed at lower temperatures. It was found that the time-resolved emission signals of the three different progressions of the d 311,-a 311utransition were in excellent agreement with the C2 absorption measurements. The infrared absorption measurements of CO and C02 performed during high-temperature oxidation of fullerene c 6 0 yielded an initial concentration of about 40 ppm c60 for all experiments. Shock tube experiments with either pure argon or 02/Ar mixtures without any fullerenes showed no absorption of C2, CO, and C02. Also, in Cm pyrolysis experiments no absorption of CO and C02 could be detected.
Discussion It can be assumed that several subsequent processes occur in the shock-heated C d A r aerosol. Due to the relatively high vapor pressure of c60, the solid agglomerated carbonaceous particles evaporate rapidly at high temperatures. This process starts behind the incident shock wave, where temperatures are in the range 1300 5 T 5 1650 K, and continues behind the reflected wave much more rapidly because of the higher temperature level. An upper limit for the fullerene particle evaporation of about 30 ps for conditions used during the present study can be calculated on the basis of thermodynamic data for
C2 Formation during High-Temperature Pyrolysis of C a
J. Phys. Chem., Vol. 99, No. 36, 1995 13511
-
. Y
1
E
k%
T
k, Simulation
2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 Figure 4. Arrhenius diagram of the apparent rate coefficient kappand of rate coefficient kl determined from computer simulations based on the decomposition model I.
the fullerene This value was experimentally confirmed by laser light scattering and extinction measurements behind shock waves.I0 We therefore can assume for all our experiments that the initial reactant C a is in the vapor phase for nearly the complete observation time and that the C60 pyrolysis is dominated by homogeneous gas phase kinetics. A first step toward an interpretation of the absorption measurements is to determine the C2 slope at early reaction time and to relate it to the concentration of the initial reactants. In this way an apparent rate coefficient for the appearance of C2 during high-temperature pyrolysis of C a can be defined: (1)
All individual values for apparent rate coefficients obtained in this way are summarized in the Arrhenius diagram of Figure 4 (see open squares). A least-squares fit to the data points represented by the line in this diagram leads to an activation energy of about 495 kJ mol-' and can be expressed by
kapp= 2.7 x 10" exp(-495 kJ/moYRT) cm3 mol-' s-'
(2)
The scattering of the data points around the given mean value is comparatively low. It is clear from the C2 concentration measurements of Figure 2 that for a more detailed kinetic interpretation at least one C2 formation and one C2 consumption reaction must be considered. It is near at hand to assume a direct C2 abstraction from c 6 0 and a removal of the C2 concentration by a bimolecular CZ C2 reaction, which was recently observed during high-temperature pyrolysis of acetylene highly diluted in argon.'
+
'
+ Ar
+ C,, + Ar c, + c, =c3 + c C,,
C,
,
(R1) (R2)
The remaining question is what happens to the fullerene fragments. With respect to our measurements, we can consider two extreme cases: a further fragmentation without any C2 formation (decomposition model I), C,,
+ Ar -C, + xt2
c58-,
+ Ar
or a further breakdown of the fullerene fragments with CZ formation during any partial reaction step (decomposition
,,*,,''L
Simulation based onmodelll
I
0
io4K / T
d[C,]/dt = [c,,] [Ar]I w o
4
8
200
400
600
800
ZIpS Figure 5. Comparison between measured (noisy lines) and calculated Cz concentrations: (-) decomposition model I; (- - -) decomposition model 11.
model 11)
Computer simulations based on the two proposed decomposition models were performed using the SENIUN program package.I2 Due to the fact that the thermodynamic equilibrium between C, C2, and C3 is not reached within the detection time of 800 ,us at temperatures lower than 3500 K, reaction R2 is assumed to be quasi-equilibrated. The rate coefficient of the reverse reaction was calculated on the basis of the equilibrium constant with a slightly modified value of the heat of formation of C2 (see refs 13 and 14). This variation of the thermodynamic data was confirmed by recent measurements in the different reaction systems.I The rate coefficient k~ of reaction R2 was taken from ref 11 to be
Rate coefficient kl was varied during computer simulations to achieve a good fitting of the measured Cz profiles. In case of decomposition model 11, as a lower limit, all rate coefficients of subsequent C;?-formingreactions were assumed to be equal to kl, although fullerene fragments are expected to be less stable than c60. Typical examples of calculated Cz profiles in comparison with experimental traces are illustrated in Figure 5. In the case of decomposition model 11, it was not possible to fit the measured C2 profiles by varying the rate coefficient kl (see dotted line). If more realistic, higher rate coefficients for the decomposition of fragments were considered compared to kl , the deviation between measured concentration profiles and simulationsbased on decomposition model Il would be even greater. For the decomposition model I, nearly complete agreement between calculated and measured C2 profiles for the two examples in Figure 5 as well as for all other experiments was obtained. The individual rate coefficients kl used as bestfit parameters are summarized in the Arrhenius plot of Figure 4 (solid triangles). The deviations from the early proposed apparent rate coefficient k,, are comparatively small. Although this fact cannot be regarded as a verification of the proposed decomposition model I, it nevertheless must be interpreted as a hint for the correctness of the mechanism. The differences in the kl and kappdata values of Figure 4 at the low temperature end are clearly caused by uncertainties in the determination of d[C2]/dtlt=odue to the relatively low signal-to-noise ratio. The influence of variations of kl on the calculated CZ concentration profiles is demonstrated in Figure 6. The example clearly demonstrates that the rate coefficient kl determined in
13512 J. Phys. Chem., Vol. 99, No. 36, 1995
Sommer et al.
6
Acknowledgment. This work is originated in the Sonderforschungsbereich 209 of the Universitat Duisburg. The financial support of the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
. -
References and Notes
10
z 5
si
0 0
200
400
600
800
t Ips Figure 6. Sensitivity of kl demonstrated by variations of f30%.
this study is very sensitive to the measured property, particularly at early reaction times. Because of the minor differences in the two sets of data points in Figure 4, we assume the rate coefficient kl for the ejection of C2 from Cw is equal to the apparent rate coefficient kappexpressed by eq 2. The activation energy of 495 kJ mol-’ or 5.1 eV determined in the present study is in between values of 4.0 and 7.8 eV reported by Eckhoff et al.,I5 Whetten et a1.,I6 and Kolodney et al.” Theoretical calculations based on Hartree-Fock methods often predict much higher values between 11 and 14 eV.
Conclusion Time-resolved C2 concentration measurements during hightemperature pyrolysis of the fullerene CW were performed behind shock waves under well-defined conditions using a ring dye laser spectrometer and an intensified CCD camera. c 6 0 seems to be stable at temperatures below 2650 K, and only at higher temperatures could an increasing amount of C2 be observed. A simplified reaction mechanism based on a C2 formation and consumption reaction was proposed. Computer simulations were compared with the experimental C2 traces, and rate coefficients for the abstraction of C2 from Cm were determined.
(1) Kroto, H. W.; Heath, J. R.; O’Brien, S. C.; Curl, R. F.; Smalley, R. E. Nature 1985, 318, 162. (2) O’Brien, S. C.; Heath, J. R.; Curl, R. F.; Smalley, R. E. J . Chem. Phys. 1988, 88 ( l ) , 220. (3) Naulin, C.; Costes, M.; Dorthe, G. Chem. Phys. Lett. 1988, 143 ( 5 ) , 496. (4) Phillips, J. G.; Davis, S. P. The Swan System of the C2 Molecule; University of California Press, 1968. ( 5 ) Rajathurai, A. M.; Roth, P.; Fissan, H. Aerosol Sci. Technol. 1990, 12, 613. (6) Markus, M. W.; Roth, P. Int. J . Chem. Kinet. 1992, 24, 433. (7) Kiess, N.; Broida, H. Can. J . Phys. 1956, 34, 1471. (8) Matsuo, T.; Suga, H. David, W. I. F.; Ibberson, M.; Bernier, R. P.; Zahab, A.; Fabre, C.; Rassat, A.; Dworkin, A. Solid State Commun. 1992, 83, 711. (9) Mathews, C. K.; Sai Baba, M.; Lakshmi Narasimhan, T. S.; Balasubramanian, R.; Sivaraman, N.; Srinivasan, T. G.; Vasudeva Rao, P. R. J . Phys. Chem. 1992, 96, 3566. (10) von Gersum, S.; Kruse, T.; Roth, P. Ber. Bunsen-Ges. Phys. Chem. 1994, 98, 979. (11) Kruse, T.; Roth, P. To be published. (12) Lutz, A. E.; Kee, R. J.; Miller, J. A. SANDIA REPORT, SAND878248.UC-4; Sandia National Laboratories: Albuquerque, NM; Livermore, CA, 1988. (13) Urdahl, R. S.; Bao, Y.; Jackson, W. M. Chem. Phys. Lett. 1991, 178 (4), 425. (14) Bauschlicher, C. W., Jr.; Langhoff, S. R.; Taylor, P. R. Astrophys. J. 1988, 332, 531. (15) Eckhoff, W. C.; Scuseria, G. E. Chem. Phys. Lett. 1993,216,399. (16) Whetten, R. L.; Yeretzian, C. Int. J . Mod. Phys. E 1992, 6, 3801. (17) Kolodney, E.; Tsipinyuk, B.; Budrevich, A. J . Chem. Phys. 1994, 100 ( l l ) , 8542. Jp950744C