Energy & Fuels 1988,2,487-493
487
Oxidation/Pyrolysis Chemistry As Related to Fuel Sooting Tendencies+ Kenneth Brezinsky,* Harjit S. Hura, and Irvin Glassman Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey 08544 Received October 12, 1987. Revised Manuscript Received March 26, 1988
Relationships between flow-reactor-derived chemical mechanisms and the macroscopically observed sooting tendencies in premixed and diffusion flames are developed. In particular, the impact of elements of the mechanism for the oxidation of benzenelphenyl radical on the inhibition of soot formation through the removal of a critical precursor is explored. Pyrolysis chemistry, especially those aspects altered by small amounts of oxygen, is related to the increased precursor concentration that is responsible for the augmented soot formation in ethene diffusion flames with oxygen added to the fuel stream. Flow-reactor data from ethene pyrolysis studies (with and without small amounts of added oxygen), demonstrating the enhanced production of acetylene and butadiene, are presented in support of some of the developed relationships between sooting phenomenology and chemical mechanisms.
Introduction The practical importance of understanding soot formation processes has motivated a series of studies examining both the macroscopic, phenomenological parameters that affect soot formation such as flame type and temperature and the microscopic, chemical processes that may be responsible for the rate-controlling soot initiation steps. At Princeton University, premixed- and diffusion-flame experiments have been conducted that have revealed the relationship between the sooting tendency of hydrocarbon fuels and temperat~re.'-~The relative sooting tendency between fuels was found to be different in the two types of flames. Under premixed conditions, an increased flame temperature was found to decrease the formation of soot. Moreover, the details of the initial fuel structure, such as the isomeric distribution of side chains, degree of conjugation and aromaticity were found to be unimportant in determining a fuel's sooting tendency except insofar as these details contribute to the total number of C-C bonds. The general conclusions are that soot most likely forms in the postflame region from an essential precursor and the sooting tendency of a particular fuel depends primarily on the balance between the amount of soot precursor it forms (a function only of the number of C-C bonds) and the amount of precursor consuming OH radicals the fuel produces. Since the OH attack increases faster with temperature than does the soot precursor formation,l sooting tendency decreases with increasing temperature. The OH concentration also depends on the C/H ratio, which, too, is a function of the number of C-C bonds.' Diffusion-flame experiments have led to the observation of almost completely opposite patterns. An increase in the diffusion-flame temperature has been shown to increase the sooting tendency of a hydrocarbon fuel. The structure of the fuel affects the degree of soot formation through the mechanism by which the fuel pyrolytically decays. Since different fuels have different pyrolysis mechanisms, the nature of the initial fuel structure becomes a significant determinant in the degree of soot formation. Furthermore, 'Presented a t the Symposium on Advances in Soot Chemistry, 194th National Meeting of the American Chemical Society, New Orleans, LA, August 30-September 4, 1987.
0887-0624/88/2502-0487$01.50/0
soot is formed on the fuel side close to the flame front. Consequently, oxidation at the flame front consumes fuel, fuel fragments, and soot particles. Particles break through the flame front and become observed as soot only when temperatures are lowered by radiation in a localized area and there is insufficient oxygen to consume both the molecular hydrocarbons and the soot particles. The details of the chemical mechanisms of the pyrolysis and oxidation processes that so strongly impact the sooting tendencies of various hydrocarbon fuels could not be revealed by the premixed- and diffusion-flame experiments that were performed. Insight into chemical mechanisms, however, has been obtained from chemical kinetic, flowreactor studies conducted a t Princeton. Studies of the oxidation of benzene: t ~ l u e n e ethylbenzene,6 ,~ propylb e n ~ e n e ,b~tylbenzene,~ ~?~ a-methylnaphthalene,l" butadiene,ll ethylene,12propane,13 butane,14and other hydrocarbons have revealed many of the mechanistic steps by which these fuels decompose. A limited number of pyrolysis studies have also yielded mechanistic information for oxygen-free conditions. Despite the information obtained (1) Takahashi, F.; Glassman, I. Combust. Sci. Technol. 1984, 37, 1. (2) Glassman, I.; Yaccarino, P. Symp. (Int.) Combust. [Proc.] 1981, 18th, 1175. (3) Gomez, A,; Sidebotham, G.; Glassman, I. Combust. Flame 1984,58, 45. (4) Venkat, C.; Brezinsky, K.; Glassman, I. Symp. (Int.) Combust. *IProc.1 1982. 19th. 143. (5) Brezinsky, Ki Litzinger, T. A.; Glassman, I. Int. J. Chem. Kinet. 1984,16, 1053. (6) Litzinger, T. A.; Brezinsky, K.; Glassman, I. Combust. Flame 1986,
- -- -. fi.? -, 9.51
(7) Litzinger, T. A.; Brezinsky, K.; Glassman, I. J.Phys. Chem. 1986, 90. 508. (8) Litzinger, T. A.; Brezinsky, K.; Glassman, I. Combust. Sci. Technol. 1986, 50, 117. (9) Brezinsky, K.; Linteris, G. T.; Litzinger, T. A.; Glassman, I. Int. Symp. (Int.)Combust. [Proc.]in press. (10) Litzinger, T. A.; Brezinsky, K.; Glassman, I. Presented at the Eastern States Section/The Combustion Institute 1985; Paper 68. (11)Brezinsky, K.; Burke, E. J.; Glassman, I. Symp. (Int.) Combust. [Proc.] 1985, 20th 613. (12) Westbrook, C. K.; Dryer, F. L.; Schug, K. P. Symp. (Int.) Combust. [Proc.] 1984, 19th 153. (13) Hautman, D. J.; Schug, K. P.; Dryer, F. L.; Glassman, I. Combust. Sci. Technol. 1981, 25, 219. (14) Pitz, W. J.; Westbrook, C. K.; Proscia, W. M.; Dryer, F. L. Symp. (Int.) Combust. [Proc.] 1985, 20th, 831.
0 1988 American Chemical Society
488 Energy & Fuels, Vol. 2, No. 4, 1988
Brezinsky et al.
Scheme I. Acetylene Soot Formation Mechanisma H
n
H
In this article, the relationships between the elements of the soot formation mechanism described in Scheme I that are relevant to the premixed- and diffusion-flame experiments and the mechanisms derived from the flowreactor oxidation and pyrolysis of hydrocarbons are developed. From the relationships, the particular contribution that the oxidation and pyrolysis of intermediates makes to either inhibiting or enhancing soot formation is clarified. In order to logically develop the relationships, the set of Princeton flame experiments will first be described; the results obtained from them will be succinctly stated and then the proposed chemical relationships discussed.
Experimental Section In both the premixed and diffusion flames the effects of temh‘ h H h perature and fuel structure on the propensity to soot have been examined by changing the amount of diluent, usually nitrogen. fragmentation fragmentation Fuel structure effects were examined by the selection of a wide variety of hydrocarbon compounds. However, the two types of flames are different in the manner in which the fuel and oxidizer come together at the flame front. As a consequence, there are profound effects on the relationships between temperature, fuel structure, and sooting tendency. In a premixed flame, fuel, oxidizer, and diluent are mixed CEC-M upstream of the flame front and arrive at the flame front as Fused polycyclic aromatics ccomponents of a homogeneous gas. In the experiments,l preset quantities of nitrogen and oxygen were mixed with a variable h’ \ amount of fuel and fed to a Bunsen type tubular burner. The fuel flow rate was increased while the oxygen and nitrogen flow “Reprinted with permission from ref 24, adapted from ref 15. rates were kept constant until luminous continuum radiation was Copyright 1987 Academic. detected at the sides of the conical flame. Then the fuel flow rate was decreased just enough to cause the radiation to disappear. from the flow-reactor studies, the absence of a chemical The average fuel flow rate associated with both the appearance and disappearance of luminosity was used to calculate the critical soot formation mechanism has prevented the linking of equivalence ratio for the onset of soot. The critical effective the flow-reactor-derived mechanisms with the pyrolysis equivalence ratio, qC,is defined as the ratio of the stoichiometric and oxidation chemistry that must be responsible for the oxygen necessary to convert all the fuel to CO and HzO,to the macroscopic, phenomenological observations of the flame experimental amount. To establish the critical equivalence ratio studies. at another temperature, the same procedure was repeated with A recently developed soot formation mechanism by a different preset quantity of nitrogen. Frenklach, Clary, Gardiner, and Stein,15 which evolved A diffusion flame is unlike a premixed flame in that the fuel from a sequence of shock-tube experiments on acetylene and oxidizer meet in a reaction zone as a result of molecular pyrolysis, can serve as a conceptual framework for condiffusion and convective transport. The diffusion-flame studies2” necting known chemical mechanisms with macroscopically were conducted with a burner in which a central tube delivered fuel into an outer tube containing flowing air. An excess of oxidizer determined sooting tendencies. The ring-formation part led to the elongated shape characteristic of an overventilated of the mechanism is similar to others recently developed from flame,16J7 flow-tube,ls and s h o c k - t ~ b eexperi, ~ ~ ~ ~ ~ flame. The sooting tendency of a particular fuel was established by varying the volumetric fuel plus diluent flow rate for a given ments. All the ring-formation mechanisms emphasize amount of air. When visible soot particles exited from an annulus similar types of radical reactions generally involving two around the top of the flame, the soot height, i.e. the length of the and four carbon species. Therefore, the mechanism of luminous zone from the burner lip to the flame apex, was meaFrenklach et al. draws attention to the types of critical sured. Sooting heights for different fuels with different amounts species and the types of chemical sequences that appear of added nitrogen or argon diluent were evaluated in order to to play a role in all the postulated mechanisms. In parestablish the sooting propensity at different temperatures. For the chemical mechanism studies,the Princeton flow reactor ticular, the mechanism proposes a sequence of events was used.21 The flow reactor is a tubular, high-temperature, starting from acetylene, proceeding through butadienyl and turbulent reactor that is designed to permit the examination of vinylacetylenyl radicals to the formation of a phenyl radical oxidation and pyrolysis processes without complications due to (Scheme I). From the phenyl radical, the growth of large the diffusion of heat and mass. Fuel is introduced into the reactor polycyclic aromatics leading to soot would proceed relathrough four injectors at a throat section of the reactor tube. The tively easily. These initial reactions in the soot formation gaseous fuel mixes rapidly with the high-velocity nitrogen carrier mechanism, or ones like them, would be appropriate for flow. The oxygen content of the carrier flow is independently the postflame region of a sooting premixed flame and the variable so that a range of conditions from fuel lean (excess fuel stream of a diffusion flame. oxygen) to pyrolysis (absence of oxygen) can be achieved. Since there is always a very small amount of oxygen contained in the nitrogen boil off used as the carrier flow, an “oxidative” pyrolysis (15)Frenklach, M.;Clary, D. W.; Gardiner, W. C., Jr.; Stein, S. E. rather than a true pyrolysis is achieved. Species concentrations Symp. (Int.)Combust. [Proc.] 1’ with respect to time for the dilute reacting oxidation or pyrolysis (16) Smyth, K. C.; M experiments are obtained by withdrawing samples with a 1540. water-cooled probe at discrete locations within the tube. Analysis (17)Cole, J. A.;Bittner, J. D.; Longwell, J. P.; Howard, J. B. Combust. of the chemical content of each sample is performed with either Flame 1984, 56,51. (18)Weissman, M.; Benson, S. W. Int. J. Chem. Kinet. 1984,16, 307. (19)Colket, M.B. Symp. (Int.) Combust. [Proc.], in press. (20) Wu,C. H.;Kern, R. D. J. Phys. Chem. 1987, 91, 6291. (21) Brezinsky, K.Prog. Energy Combust. Sci. 1986,12, 1.
F
Energy & Fuels, Vol. 2, No. 4, 1988 489
OxidationlPyrolysis Chemistry 1.5
1.5
+ 1.4
t
1.3
I
1.2
I ETHANE
P P I
WTH "AIR"
1,
-
2200 K
/ J l PROPANE,
-
#
1.1 1-HMENE
*o
1.o .9 OCTANF
t
t
.7
BENZENE' T O L U
E
g
*
I
.8
-
.7
-
.6
-
I
_
L-.2529 CUMENE
1.2.3,4-TETRAHTDRONAPl-THA!LNE
1.2.3.4-TETRAMDWMNE
I
1-METhYLNAPHlHALENE
. 5 ' < 0 2
.6 1.7
1.8
1.9
2.0
2.1
Tf X
2.2
2.3
2.4
2.5
(K)
Figure 1. Critical effective equivalence ratio at sooting as a function of adiabatic flame temperature. Reprinted with permission from ref 1. Copyright 1984 Gordon & Breach.
'
I
4
I
I
"
6
8
I
I
10
"NUMBER OF C-C
12
a
'
14
I
I
16
18
BONDS"
Figure 2. Correlation between the critical equivalence ratio at sooting and the number of C-C bonds. Reprinted with permission from ref 1. Copyright 1984 Gordon & Breach.
gas chromatography or gas chromatography/mass spectrometry. Concentrations at the various locations in the reactor are related to the extent of reaction by taking into account the flow velocities within the reactor. Experiments examining the oxidation and pyrolysis of various hydrocarbon fuels have all been conducted at 1atm pressure and in a 900-1200 K temperature range. The temperature range of the flow reactor, though lower than the range of hydrocarbon adiabatic flame temperatures (2000-2500 K), is nevertheless quite relevant to the chemistry of soot formation processes. The 900-1200 K range corresponds to both the temperature in a flame where the initial fuel decomposition occurs and also the temperature of the zone of a diffusion flame where soot particles are first observed.22
Results The critical equivalence ratio, a measure of the sooting tendency, is plotted as a function of the premixed, adiabatic flame temperature for a wide variety of hydrocarbons in Figure 1.' The larger is the critical equivalence ratio, the smaller is the tendency of fuel to soot. Therefore, from Figure 1it can be seen that ethane a t all temperatures has a smaller tendency to soot than does methylnaphthalene. The most obvious trend in Figure 1 is that for all fuels, the sooting tendency decreases as the flame temperature is increased. Furthermore, the change in sooting tendency with temperature is roughly the same for all fuels regardless of fuel type. Consequently, a vertical slice through Figure 1a t any one temperature should permit an ordering of sooting tendencies that is representative of those at all temperatures. Such an ordering is presented in Figure 2 for a flame temperature of 2200 K.' The abscissa, "number of C-C bonds", represents the total number of carbon to carbon bonds in the parent fuel molecule when each single carbon-carbon bond is considered to contribute one, each double bond contributes two and each triple bond contributes three to the total number of bonds. The number of C-C bonds counted in this manner is, in effect, a surrogate for both the C/H ratio and molecular size.' The predictive correlation of Figure 2, which is independent of a detailed knowledge of isomeric structures, conjugation, (22) Kent, J.; Wagner, H.G. Symp. (Znt.)Combust. [Proc.] 1984, ZOth,
1007.
4.1
4.3
4.5
4.7
4.9
5.1
5.3
5.5
l/r (l/DEG)*10000
Figure 3. Sooting tendency of some hydrocarbon fuels as a function of the reciprocal of the calculated adiabatic flame temperature. Reprinted with permission from ref 24. Copyright 1987 Academic.
and aromaticity, implies that fuel structure is an inconsequential factor in soot formation in premixed flames. The results of these premixed-flame experiments along with those that have examined the postflame region as a function of initial fuel suggest that soot formation occurs in the postflame region from a universal soot precursor whose concentration, but not nature, is affected by the structure of the initial fuel. The results of a diffusion-flame study of the sooting tendency of hydrocarbons as a function of temperature are shown in Figure 3.24 The sooting tendency is best measured by the inverse of the fuel flow mass rate (FFM) a t the smoke height in order to normalize the molecular (23) Harris, S. J.; Weiner, A. M.Combust. Sci. Technol. 1984,38, 75. (24) Glassman, I. COMBUSTZON; Academic: N e w York, 1987.
Brezinsky et al.
490 Energy & Fuels, Vol. 2, No. 4, 1988
weight differences of the fuels.2 The smaller the inverse flow rate the smaller is the tendency to soot. The sooting tendency of different classes of species varies widely from class to class but is relatively constant within a class. For example, aromatic compounds such as benzene, ethylbenzene, and methylnaphthalene soot more easily that do the small alkenes such as propene, butene, and even cyclohexene. The classes of compounds also have different sooting temperature dependencies as revealed by the slopes of each line in Figure 3. However, in contrast to the behavior of hydrocarbon fuels in premixed flames, fuels show an increasing sooting tendency with increasing flame temperature. A structure-independent correlation of the type available for premixed flames cannot be developed from the diffusion-flame results. It appears that initial fuel morphology does play an important part in the tendency of a fuel to soot as is evidenced in Figure 3 by the groupings of sooting tendency according to class of hydrocarbon. Figure 3 also demonstrates linear relationships between the log of the inverse fuel flow rate at the sooting height and the inverse temperatures that are reminiscent of Arrhenius kinetics plots for pyrolysis reactions. Since hydrocarbon fuels will pyrolytically decompose a t the temperatures present in the oxygen-free fuel stream a t a distance far from the flame front, it appears that the effect of initial fuel structure on sooting tendency is manifested through kinetically controlled pyrolysis processes.2 Flow-reactor studies of hdyrocarbon fuels cannot be so concisely summarized in three plots as were the flame results. Generally, each oxidation study is conducted a t rich, stoichiometric and lean equivalence ratios at one or more temperatures. Figure 4 gives examples of the types of species profiles that are obtained from a study of the oxidation of benzene. In the figure, the benzene fuel decay profile is observed as well as the growth and decay of intermediates such as cyclopentadiene and vinylacetylene. Also obvious is the formation of the final products CO and C 0 2 along with the attendant temperature rise. These species profiles are characteristic not only of the oxidation of benzene at all lean, stoichiometric, and rich equivalence ratios but are also characteristic of the species profiles obtained during oxidation studies of other hydrocarbons. Flow-reactor studies are always conducted so that the profiles are obtained for a major portion of the fuel decay, intermediates production and decay, and some final product formation. Other examples of flow-reador species profiles relevant to the subsequent discussions in this article are available in Figure 5a,b, the “oxidative”pyrolysis and very rich oxidation of ethylene. From the species concentration profiles with respect to time, mechanistic information can be deduced. The mechanism of the high-temperature oxidation of benzene/phenyl radical that was developed from a series of flow-reactor oxidation studies is displayed in Figure 6.21 This mechanism is chosen for display because of its relevance to important elements of the soot formation mechanism to be discussed later.
Discussion Before discussing the relationships between flame and flow reactor experimental results, the assumption must be made that the diffusion of species in flames affects the rates of reactions but not the basic pathways of a chemical mechanism that would be observed in the reduced diffusion environment of the flow reactor. Some preliminary, direct comparisons of flame species obtained from the probe sampling of an ethene diffusion flame and flow-reactor pyrolysis experiments appear to support this con-
1250
1200
m -4 11505
CO2/10
__I
-
m
5
1100:
m
-
C6H5OH
0
20
1050
60
40
a
80
100
1000 120
TIME MSEC
+. .4
I-
0
20
40
b
60
80
100
120
80
100
120
TIME MSEC
2H2 \
c o
20
40
60
TIME MSEC
Figure 4. (a) Selected species profiles from the stoichiometric oxidation of benzene: C&, benzene; C,H50H,phenol; CO, carbon monoxide; COz, carbon dioxide. Initial temperature = 1115 K. (b) Selected species profiles from the stoichiometric oxidation of benzene: C5HB,cyclopentadiene; C4H4,vinylacetylene; C3’s, propane and propene. (c) Selected species profile from the stoichiometric oxidation of benzene: C2H2, acetylene; C2H4, ethylene; CH,, methane. Reprinted with permission from ref 21. Copyright 1986 Pergamon.
c l ~ s i o n . ~Presumably, ~ diffusion of radicals would also leave the basic mechanisms of oxidation and pyrolysis in premixed flame essentially unaffected but, as in diffusion flames, alter the relative impact of different reactions. The results of the studies of sooting tendencies in premixed flames suggest that in these types of flames, at or just past the critical equivalence ratio, the different fuels rapidly break down in the reaction zone, forming, in part, (25) Sidebotham, G. Ph.D. Thesis, Mechanical and Aerospace Engineering Department, Princeton University, 1987.
OxidationlPyrolysis Chemistry
Energy & Fuels, Vol. 2, No. 4, 1988 491
10-1
INITIATION PROCESSES
-
RADICAL POOL ( H, O H , 0 , HOn )
4H+OH
$HE+/
