446
Energy & Fuels 1988,2,446-453
Mechanistic Investigation of Soot Precursorst S. H.Bauer* and P.M. Jefferst Baker Laboratory, Department of Chemistry, Cornel1 University, Ithaca, New York 14853-1301 Received September 11, 1987. Revised Manuscript Received January 20, 1988
This report focuses on ascertaining which types of molecular species, both radicals and ring structures, are essential precursors to soot production. We have developed a minimal mechanism to account for the delay times for the synthesis of precursors when shock-heated samples of benzene and toluene were rapidly quenched by an expansion wave (test times in the back-reflected region of =500-700 ps) and have formulated a nucleation-association scheme for the onset of aromatic ring growth. For comparison, data are presented for benzene and toluene that were pyrolyzed in a singlepulse shock tube (T5= 1400-2200 K). Time-dependent absorption spectra were recorded over the range 400-800 nm, with 20-ps time resolution, throughout the duration of the reflected shock and expansion regimes. From the analysis of these histograms and their dependence on fuel composition and shock temperature, the kinetics of production and destruction of soot precursors were inferred.
Introduction Fuels with low H/C ratios are particularly prone to soot production during combustion; aromatic species generally (but not universally) do so more readily than aliphatics. It is now well recognized that "soot" is not a singular material even though during the past decade the jargon of scientista working on the soot problem implied that soot consists of specific substances for which one could write representative molecular structures and make estimates of their thermochemical parameters. Extensive efforts have been devoted to determining conditions that promote the development of soot in flames and in internal combustion engines, besides characterizing the variety of soots with respect to composition, volatile content, and structure. Many studies have been reported on the effects of inhibitors; reviews abound.' During the past decade chemical kinetics investigations have proliferated with the hope of unraveling the mechanisms for its generation, ultimately to permit control of the types and magnitudes of soot emissions. There is general agreement on species types that initiate condensed aromatic ring growth,2 still disagreement as to whether ions play a major role: and overall agreement on the spacial distribution of PAH in flames, as measured mass spectrometrically.4 Composition constraints on fuel/oxidizer ratios for the inception of sooting and the temperature range in flames wherein soot appears are sufficiently well-defined.5 In this report our objectives are as follows: ( A ) to call attention to the differences and the conceptual similarities between the onset of sooting and a kinetic model for nucleation-condensation; (B) to list the types of precursors required for sooting and the underlying experimental basis; (C) to present a minimal set of reactions, with rate constants, that model the observed time evolution of condensed molecular structures (soot precursors), where this list must incorporate a repetitive growth cycle for continued condensation and, as a minimum, the model must semiquantitatively reproduce observed delay times for the Presented at the Symposium on Advances in Soot Chemistry, 194th National Meeting of the American Chemical Society, New Orleans, LA, August 31-September 4, 1987. *Presentaddress: Department of Chemistry, SUNY at Cortland, Cortland, NY 13045. 0887-0624/88/ 2502-0446$01.50/0
onset of condensation; (D)to present qualitative spectral data that support point C. A. The characteristic kinetic features of a typical nucleation-ondensation mechanism6are as follows. The f i i t feature (stage a) is initial binary association sequence that reaches steady state at some small number (n):'
Note that at this stage the rates of association-dissociation are nearly balanced for each step, with k,,[A,] = k-+l). No activation energies are involved for the associations, which are driven by decreases in enthalpy; dissociations are favored by almost equal TAS terms. This stage is followed by stage b
wherein steady state is achieved by unidirectional flow, (1)Sections on soot and combustion-generated particulates can be found in the 15th-21st issues of: Symposium (International) on Combustion; The Combustion Institute Pittsburgh, PA starting with the 15th (1974) through the 21st (1986). See also: Wagner, H. G. 17th Symposium (International) on Combustion; the Combustion Institute: Pittsburgh, PA, 1978; p 3. Hayes, B. S.; Wagner, H. G. Prog. Energy Combust. Sci. 1981, 7 , 229. (2) (a) Frenklach, M.; Clary, D. C.; Gardiner, W. C.; Stein, S. E. 20th Symposium (International)on Combustion; The Combustion Institute: Pittsburgh, PA, 1984, p 887. (b) Bockhorn, H.; Fetting, F.; Wenz, H. Ber. Bunsen-Ges. Phys. Chem. 1983,87,1067. (c) Homann, K. H. 20th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1984; p 857. (3) Keil et al. (Keil, D. G.; Gill, R. J.; Olson, D. B.; Calcote, H. F. 20th Symposium (International)on Combustion; The Combustion Institute: Pittsburgh, PA, 1984; p 1129and the symposium presented in this issue) favor ionic mechanisms. Bertrand (Bertrand, C.; Delfan, J-L. Combust. Sci. Technol. 1985,44,25) discount significant contributions from ions. (4) Longwell, J. P. 19th Symposium (International) on Combustion; The Combustion Institute Pittsburgh, PA, 1982, p 1339. (5) Smith, K. C.; et al. Combust. Flame 1985,62, 157. Frenklach, M.; Hus, J. P.; Miller, D. L.; Matula, R. A. Combust. Flame 1986, 64, 141. (6) Bauer, S. H.; Frurip, D. J. J. Phys. Chem. 1977,81, 1015. (7) Bauer, S. H.; Wilcox, C. F., Jr.; Russo, S. J.Phys. Chem. 1978,82, 59.
0 1988 American Chemical Society
Investigation of Soot Precursors
no.
Energy & Fuels, Vol. 2, No. 4,1988 447 Table I. Minimal Mechanism log A 15.7 13.3 12.3 14.8 10.76 16.0 13.7
C6Hs
1
2 3 4 5 6 7 8 9
13.0 13.0 13.0 13.0 11.0 12.74 13.0 13.0 12.74 (12.08) 13.3
10 11 12
13 14 15 16 17 18
E,,, cal mol-’ 108000 0.00 65000 82800 52500 45000“ 2000.0 0.00 2000.0 0.00 2000.0
ref a b
11000 10000 2000.0 0.00
f
10000 92600 6600h
g
c
d e b
e b e b e
g
e b
b
a H ~ sD. , S. Y.; Lin, C. Y.; Lin, M. C. 20th Symposium (International) on Combustion;The Combustion Institute: Pittsburgh, PA, 1984. *Frenklach, M.; Clary, D. W.; Gardiner, W. C.; Stein, S. E. 20th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1984. Warnatz, J. Ber. Bunsen-Ges. Phys. Chem. 1983,87, 1008. However, Eo = AHr,,,, = 67.4 kcal mol-’ (rather than 54 kcal mol-’). dAsaba, T.; Fujii, N. 13th Symposium (International) on Combustion;The Combustion Institute: Pittsburgh, PA, 1971; p 155. eValues for the reverse reaction, from footnote b, reflected through the equilibrium constant. fA slightly adjusted value based on: Brooks, C. T.; Peacock, S. J.; Renber, B. G. J. Chem. Soc., Faraday Trans. 1979, 75, 652 and footnote d. #Based on the reaction:
Q0
+
C2H2
- 6JH+
H
k.10’~
For (13) and (16) we assumed a somewhat lower A value and inserted Eo = 10 kcal mol-’, because in these the H atom is lost from the ring rather than from the added moiety (C2H2).hE,,’s were set equal to AHr,,,, derived from Table 111.
Table 11. Initiation Steps for Toluene CnHsCHn log A 12.9 C7H8 = C7H7 + H 11.6 C7H8 = C6H5 + CH3 H + C7Hs = C6H5 + CH4 1 + 4(log 7‘) H + C7Hs = C7H7 + H2 1.6 + 4(log 2‘) C7H7 = C3H3 + 2CzH2 14.0 C7H7 + H = C3H3 + C4H3 + Hz 14.65 CzH6 + Ar = 2CH3 + Ar 14.6 C3H3 + C3H3 = C& 13.0
no. TI T2
T3 T4 T5 T6 T7 T8 a
Estimated-refer
E,. cal mol-’ 72600 90000 2100 2100 84800 80000 88400
~
ref 12 12 a
12 12 12 a a
to text.
such that k,,, = lz,, with insignificant reversibility. Here the enthalpy factor completely dominates. The magnitude of n at which tis “switchover” occurs characterizes the critical size nucleus. The contrast with soot production from C/H fragments the rate of pyrolysis is striking. The initial lag is due to (a1) of the fuel, to generate small reactive fragments, generally referred to as “acetylenic species”, and (a2)their partial recombination to (in some cases the direct production of) aromatic ring radicals; activation energies control these steps. Hence, minimal temperatures of ~ 1 5 0 K 0 are required. It is likely that a dynamic local equilibrium, similar to the steady state (a), develops between these small highly reactive radicals. When adequate levels of both types of species are attained, stage /3 follows, i.e., an essentially unidirectional growth sequence, wherein the acetylenic species add onto the aromatic radicals, in analogy with stage b. Thus, there occurs a “switchover” that has the appearance of a catastrophic onset of sooting. Since at all times in the /3 stage the driving enthalpy for growth is countered by an opposing TAS term, at some higher tem0 This accounts perature TAS quenches sooting ( ~ 2 1 0 K). for the bell-shaped generation profile [soot yield vs temperature] reported by many observers. B. What are the essential precursors which operate in regions a1and a*? Observations, previously reported for
shock tube pyrolysis studies of 10 polycyclic aromatics,s guided our choice of the smallest number of species that have to be incorporated in a minimal mechanism. For shock durations of G O O FS, over the temperature range 1500-2200 K, acetylene, tetramethylpentane, acenaphthene, or acenapthalene, when individually pyrolyzed, yielded insignificant amounts of soot. However, any aliphatic/ aromatic combination of the above under the same shock conditions produced copious amounts of soot. Clearly, two types of molecular species are required for the onset of sooting. It follows that one should anticipate longer delays when a single type is initially present because of the time required to generate the other type. Schematically
& k
initiation (pyrolysis)
s m a l l fragments hi
aromatic ring radicals
+
dl
h-i
acetylenic species
1 soot precursors
(8)Bauer, S. H.; Zhang, L.-M. 14th International Symposium on Shock Tubes and Waves; Archer, R. D. Melton, B. E.,Eds.; New South Wales University Press: Kensington, NSW, Australia, 1983; p 654.
Bauer and Jeffers
448 Energy & Fuels, i Vol. 2, No. 4, 1988 Table 111. Species Considered Afff030300?
no. 1 2
source
3
19.8
4 5 6 7 8 9 10 11
52.1 0 122.0 54.2 105.1 123.7 67.0 132.3
a a a a
12 13
73.6 a
14 15
species
kcal/mol 0 78.5
0’ 0 (yzH
iwP030300,
no.
source
kcal/mol
22
a
113.9
23
species
55.2
24
a
167.7
25
a
125.7
26
a
72.2
27
a
103.8
28
a
79.4
UC*”
94.7 36.0
a
148.5
16
89.8
17
50.0
18
a
42.9
19
a
36.5
20
a
108.7
21
a
50.0
29
109.0
30
67.0
aValues for hHfo,ASo, and C,(T) were estimated from Benson’s group additivity tables.
The relative magnitudes of kf and k, are determined by the structures of the fuel. C. The mechanisms proposed below show that initial fragmentation can lead to subsequent aggregation to condensed aromatic structures and account semiquantitatively for our experimentally observed delay times. It is similar to the sequences suggested by Bockhorn et al.2b and by Frenklach et ala2*Two sets of reactions with appropriate rate constants are listed in Tables I and I1 and for benzene and toluene, respectively. We attempted to identify the smallest number of essential steps, not to list all reactions that plausibly occur concurrent with, prior to, and during soot production. In the “filtering process”, we considered more than 30 species (Table 111).
Species of Mechanisms
C6H6.For the benzene pyrolysis our final set consists of 18 reactions with 18 H/C species (plus Ar), although at least twice that number of reactions and species were considered during the preliminary calculations. Not in-
eluded in Table I are intermediate radical stabilization steps. We treated the addition of acetylene as a single step
C8H5 + H2 k7 = 1013.7 exp(-2000/RT)
C&
+ C2H2
-
-+
rather than a two-reaction sequence
C6H5 + CzH2
C8H6 + H
-
k7’
= loi3
C8H6 + H C8H5 + H2 k7“ = 2 X 1013exp(-6800/RT) The typical growth cycle is illustrated by
Investigation of Soot Precursors
Energy & Fuels, Vol. 2, No. 4,1988 449 -2.0
-4.0
-6.0
g
”
u Y
“(
9
-8.0
B
1y 0
e2
4
-10.0
-12.0
-5 0
-4.5
-4 0 l o g l o t (s)
-3 5
-3 0
-2.5
-5.0
Figure 1. Calculated dependence of mole fraction on time, for various species in shock-heated 1%C6H6in Ar,with Ts(initial) = 2120 K and Ts(final)= 1907 K. Reactions and rate constants are listed in Table I.
During each cycle one ring is added and a new radical is generated, which repeats the cycle. As ring condensation proceeds there are possibilities for alternate routes to yield other observed products, for example R12; it was included because biphenyl is a commonly found product, although no further reactions of biphenyl occur other than the reverse of its formation. The other stable molecular products incorporated in this mechanism are C2H2,C4H2, H10C18, and C22H12.It is necessary to impose an upper limit, i.e. the largest species incorporated in the computer code; we chose C22H12,which serves as a “sink”. Tests show that the concentration/ time patterns for the last three species, for any selected terminus, remain essentially unchanged when the largest assumed unit was varied (Cls Czo7 C2J. Clearly, the two-step growth sequence continues until the system is quenched. Kinetic calculations to model rates of production of soot precursors by pyrolysis were performed with the Mitchell-Keeg shock-kinetics program. All reactions were considered reversible, with their reverse rate constants calculated within the program by reflection through their equilibrium constants. Most of the unavailable thermodynamic parameters were estimated by Benson’s group contribution recipe. Figure 1 is a plot of the computed concentration-time profiles for 1% C6H6in argon processed by a reflected shock, heated to 2120 K (initial). I t shows the expected general features. During early times the mole fractions of H, C2H2,C4H2, and H2 rise, that of C6H6declines slowly, and nearly steady state concentrations of C6H5 and C8H5 develop. Then the higher molecular weight products slowly begin to grow, in a sequence of increasing carbon content. The time dependence of the imposed cutoff at CZ provides a measure of the delay time for the onset of avalanche soot growth. Note that after a gradual decline, C6H6drops sharply, as do all the heavier species even though all reactions were treated reversibly. C22H12(the terminal species) increases (designated as the soot initiator). Eventually, after the lower carbon content species had
-
(9)Mitchell, R. E.;Kee, R. J. Sandia Natl. Lab. [Tech. Rep.] SAND 1982, SAND-82-8205.
-14.0
-4.5
-3.5
-L.O
loglot
-3.0
-2.5
(s)
Figure 2. Same as Figure 1,but with a lower rate constant for R1 (see text). -2.0
-4.0
g p
u
c
-6.0
u
I
3
P
\I
/ /
-1
“ 0i
-8.0
2
-10.0
I
-5.0
-4.5
-4.0
loglo t
-3.5
I
-3.0
-2.5
(s)
Figure 3. Same as Figure 1, but with added (initial) C2H2.
passed through maximum levels, C2H2,Cis, and H2 dominate. A measure of the sensitivity of this mechanism to the rate constant for initiation (Rl) was obtained by using a value derived by reflection of Frenklach‘s2estimate for the reverse of R 1 , l X 1013mol/(cm3 s) (Figure 2), whereas the curves in Figure 1were calculated with kl about a factor of 10 larger, as suggested by Lin;’O it is very close to the value reported by Fujii and Asaba.” The faster initiation rate results in a much earlier appearance of heavy products and considerably more extensive (eventual) destruction of benzene. When the one-step phenyl radical decomposition (R5) was replaced by the sequence C6H5 C6H5(l)(linear) k4‘ = 2.4 X 1OI2 exp(-58800/RT)
-
-
C6H5(1) C4H3 + C2H2 k / = 1.8 X l O I 5 exp(-40000/RT) (10)Hsu, D.5. Y.; Lin, C. Y.; Lin, M. C. 20th Symposium (Znternational) on Combustion; The Combustion Institute: Pittsburgh, PA, 1984; p 623. See also: Kieffer, J. H.; et al. J. Phys. Chem. 1985,89, 2013. (11)Fujii, N.;Asaba, T. 14th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1973;p 433.
Bauer and Jeffers
450 Energy & Fuels, Vol. 2, No. 4,1988
a dramatically slower depletion of all the low molecular species was indicated. This merits further investigation, possibly by direct assay of the time dependence for appearance of C4H3 The proposed mechanism is sensitive to one more feature-the initial concentration of CzHz. If one starts a calculation, as for Figure 1, but with added CzHz(mole fraction of 0.005), the production of all the heavy products is strongly accelerated; see Figure 3. This agrees with our observations that soot formation from acenaphthene or acenaphthalene is significantly accelerated when CzHz(or tetramethylpentane, which at shock temperatures readily produces C2H2)was included in the initial mixture. Two variations of the mechanism were found to exert moderate effects on the rate of heavy product formation. The C2H2addition sequence via one step rather than two steps delays somewhat the appearance of the final product. Reducing the value of k l z by a factor of 100 makes (at most) a 15-20% change in heavy species mole fractions at 5 ps, with slightly larger differences in ClzHl0and CsH6 concentrations at earlier times. Several factors were found to have slight or negligible effects. These noncritical factors include deletion of radical stabilization steps, incorporation of an initial (artificially) high H atom concentration, the value for k, (acetylene addition to phenyl), and the inclusion of an alternate initiation process
H
CeHe
-
C&
kl’ = 4.5 X1Oi4 eXp(-4300/RT)
-
C4H5 + CzHz kl” = 6 X 1014exp(-92000/RT) C6H7
Reduction of E, from 92 to 80 kcal/mol for the second step had no effect. Substituting the bimolecular reaction
-
C6Hs + C& C&, + C&7 kl” = (5 X 10l2 to 5 X 1 0 9 exp(-85000/RT) for R1 failed to provide sufficient radical species for the pyrolysis sequence to proceed. Finally, we investigated the interesting possibility that a significant (but not a substantial) fraction of aromatic ring condensation occurs via Diels-Alder additions: CsHg
+
C&t6
-
-
[CI~HIO]
+
H2
CioHs
I
2
I1
Glob
-I- C6H6
-
+
C2H2
.
IV
I11 c c i ~ b z ] + H2
Ci4Hio
+
C2H2
The magnitudes of the enthalpy increment and activation energies were estimated by Wilcox and Carpenter12 11-1:AHD2s27k c a l / m l , w i t h E o I I , I ~ 4 kcal/mol 1 111-11:E ’ I I I , I I % ~ kcal ~
)E0111,1z76
kcal
Overall, such a condensation sequence would be driven by the increase in entropy but countered by the high enthalpy of the ejected acetylene. Estimates indicate that the large (12) Wilcox, C. F., Jr.; Carpenter, B. K. J.Am. Chem. SOC.1979,101, 3897.
-3.0
-5.0
-7.0
g i l
m 0
1c(
2
-9.0
P w M
--11.0
I
//’22”2
-132’
I
-4.5
-4.0
loglot
-3.5
I
-3.0
(s)
Figure 4. Calculated dependence of mole fraction on time, for various species in shock-heated2% C6H&H3in Ar,with ?“,(initial) = 1817 K and T,(final) = 1726 K (incidentshock speed = 1.02
mmlps). activation energies for the first step decrease for the higher ring systems. Nonetheless, the estimated rate constants are too low to compete with the mechanism outlined in Table I. C6H5CH3. The adequacy of our minimalist approach to develop a mechanism for the precursor stage of sooting is illustrated by an analysis of the pyrolysis of toluene. The initiation steps (Table 11) differ from those for benzene (steps 1-6, Table I) but thereafter the growth sequence is the same; see Figure 4. Here also the calculated and observed delay times are in acceptable agreement. Six additional species must be included (C7H8;C7H7;C3H3; CH,; CHI and C&). Reactions T1, T2, T4, T5, and T6 were taken from Mizerka and Kiefer.13 The combined effect of reactions T2, T3, and T8 rapidly generates appreciable concentrations of C6H5 and C , & . No effort was made to determine which reaction path is most effective. At temperatures around 1800 K and higher, T2 is a significant pathway for the initial breakup of toluene. The rate constant for T3 was assumed to be one-fourth that for T4; both are exothermic. T7 is a reasonable sink for CH3radicals, while the recombination of C3H3)s(T8)keeps the reaction sequence alive in a simple fashion. All the initially estimated rate constants were used in the calculations without subsequent adjustments. The possibility that there are steps in the sequence T1- T8 that are not essential for the “minimal”mechanism was not fully tested; it appears that T7 could be dropped. The reaction C6HB+ C7H8 C6H6+ C7H7could be a major depletion step for C6H5. If this reaction were included in the scheme (11),its effect would be to decrease the toluene concentration more rapidly and provide an earlier access to the benzene mechanism (I). However, no new species are thereby introduced and no major bottlenecks are bypassed by this route. Our conclusion is that a few additional steps added to the simplified mechanism proposed for benzene can ac-
-
(13) Mizerka, L. J.; Kiefer, J. H. Int. J. Chem. Kinet. 1986, 18, 363.
Energy & Fuels, Vol. 2, No. 4, 1988 451
Investigation of Soot Precursors
t'
I
7b
70
!5i
,7c
'9
I
.6-@
,/
2
3 Plale
,
H e a t e d lo
1
150'c
cx=w , (Vi
o
~
~
+
~ 3
I
l8
*
H e o l e d to
t50'C
4
,
,
x 3-0
Figure- 6. Wave~ diagram. ~ was replaced by a Xe lamp and the phototube by a monochroof diodes for scanning other regions of the spectrum (Figure 5b).14 The signals from the piezo gauges allow evaluation of the shock speed b d dwell time. Since the gas at different initial positions along the tube is heated for different lengths of time, the extent of light absorption (or scattering) must be integrated along the entire tube length. (The corresponding wave diagram is illustrated in Figure 6.) Let A,(x;t? be the instantaneous concentration of the nth absorbing species, where t' = tw- x/ur;t,is the laboratory time for shock reflection a t the quartz window, x is the return distance from the window at the location of the reflected shock, and u, is the reflected shock speed. Then, the recorded light loss is
array L-_--______ 5.._ _ J _ -_ -_ _ _ _ _ _ _mator/linear ..-----
Figure 5. Schematic of shock tube and optical configuration. (a) Shock tube (1in. i.d.; 60 in. long): (1) helium-neon laser; (2) mirror; (3) plastic back-plate; (4) diaphragm location; (5) quench tank; (6) lead to vacuum lines; (7a,b) piezo gauges; (7c) piezo crystalto trigger; (8)sampling port; (9) quartz window; (9a) optical filter; (10) phototube. (b) Optical configuration: (10) lens; (11) monochromator; (12) linear array detectors; (13) amplifiers and digitizers; (14) read and hold; (15)computer.
count in a semiquantitative way for many features of the pyrolysis of aromatics. F o r soot production, the major pathways are evident and inherently reasonable. D. What experimental evidence exists, or can be developed, to support the above proposals? Other than direct mass spectrometric detection of PAH, one must look for some in situ diagnostic technique. Absorption and fluorescence spectra5 are indicated, but there are obvious limitations. The shock-heated samples are complex brews, characterized by superposed broad spectral bands. Thus, there is little likelihood that one could identify specific species. But, we are concerned with molecular types; the saving feature is the absence of oxygen o r nitrogen chromophoric structures. The recorded spectra in the near-UV and visible regions, down to the near-infrared region, must arise from condensed polycyclic aromatics, either the stable species, as reported in the literature, or their radicals. To distinguish between general turibidty and characteristic absorptions, one should measure the temporal and wavelength dependence of light loss on passage through the reacting medium, as a function of temperature. A condition for the adequacy of our minimal mechanism is that it would correctly predict the temperature dependence of the time delays of the growth of condensed polycyclic aromatics.
Experimental Section Since a large diameter shock tube was not available to measure the time-dependent spectra transverse to the shock flow, we had to resort to recording integrated absorption spectra by passing the probing beam axially along the shock direction. The data were resolved on the bais of an additional integration step in the analysis. A 1-in. i.d. I.D. stainless-steel shock tube (Figure 5a) was fitted with a clear plastic end wall at the driver section; a quartz window and filter terminated the test section. In the first set of experiments (group a) a He/Ne laser beam (6328 A), directed along the axis of the tube, was aimed at a small aperture inserted between the quartz end plate and the narrow band-pass filter. The phototube output was recorded simultaneously with the output of two pressure transducers located 1 and 11 cm from the downstream end. In later experiments (group p) the laser
Here we assumed that no chemical processing and no absorption (at the probing wavelength) occur during the incident shock. Reflected shock temperatures ranged from 1400to 2200 K, and were controlled by varying both the initial pressure of the test gas and the diaphragm thickness. Residence times were generally about 700 ps, followed by a rapid quench due to expansion. Analysis by GC of the shock-heated samples (both gas-phase and condensable species) indicated that during the test time the products had not achieved their equilibrium concentrations at the reflected shock temperature. I. In the a group of experiments we dealt with molecular species that strongly absorb red light (He/Ne). These are shock generated after induction periods (7)that are temperature and reagent dependent. The induction time for the development of noticeable absorption is measured from t , = 0 (Figure 6). The time-dependent absorption of 632.8-nm light, I@), is illustrated in Figure 7a for biphenylene (1% in Ar); this type of sigmoidal shape is typical of histograms for other shock-pyrolyzed hydrocarbons. The simultaneously recorded pressure trace shows that subseqauent to an elapsed 7,there is a rapid rise in absorbers and some scatterers; these attained initial concentration levels such that the laser beam was never fully attenuated even though it traversed the entire sample. The initial rise occurs well before the expansion wave reaches the piezo detector 7a (Figure 5). For some species (acetylene and tetramethylpentane, for example) substantial light loss was observed even when no significant soot was generated over the 700-ps test period. For any reactant, both 7 and I ( t d are temperature and wavelength dependent (p group); as expected 7 decreases with increasing temperature. After correction for light emission by the hot sample, we found that a plot of temperature dependence of the initial saturation levels was similar in shape to that reported by many observer^'^ (14)Dunnam, C. R.; Chiu, N.-S.; Bauer, S. H. Rev. Sci. Instrum. 1986, 57, 384. (15) Graham, S. C. 16th Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1976; p 663.
Bauer and Jeffers
452 Energy & F u e l s , Vol. 2, No. 4, 1988
a
tw=O
5
4
t 3
i
2
-
I ( in units of 60 ps)
-
1
1iA;tl I,: 0
b
(A)--
Figure 8. (a) Absorption spectra in the phene series. (b) Absorption spectra of the acene series.
C
t (in units of 120
ps1
-
Figure 7. (a) Typical oscillogram for a experiments (1% biphenylene in Ar): upper trace, the transmission signal (He/Ne); lower trace, a record of concurrent shock pressure. The quench wave arrives at location 7a (Figure 5) at the tip of the lower arrow (+t). (b and c) Absorption spectra of shock-heated toluene (2% in Ar) in the reflected shock regime ( p experiments). All spectra were recorded with 20-gs time resolution; t, = 0 indicates the time when the shock was reflected from the quartz end plate. In illustration b the incident shock speed, ui, is 1.031 mm/ps. 0 radiation Compare the delay time and the decay rate for ~ 4 0 nm with those in Figure 7c. In illustration c the incident shock speed, ui, is 1.020 mm/gs; spectra were recorded at 4 0 0 nm. for total light scattering. They are approximately bell-shaped functions with maxima at 1600-1800 K,but all our curves are shifted to lower temperatures by ~ 1 0 0 K. On both the lowtemperature side (the rising segment) and the high-temperature side (the falling segment), the initial saturated absorption levels
were clearly less than total. These observations can best be interpreted by assuming that the recorded light loss was due primarily to absorption by transient species, which are precursors to soot, and to a lesser extent by the soot generated during the later stage of the runs. Each absorbing species pesses through a bell-shaped concentration-time profile. The finite rate of initiation of two-, three-, and four-ring systems accounts for the initial lag, but in turn these species are removed by condensation to produce the higher homologues, so that the net absorption by the shock-heated sample shifts progressively to the red. This sequence is reproduced by the proposed mechanism shown in Table I and the computed mole fractions in Figure 1. It is reasonable to assume that the absorbing species at 632.8 nm are condensed ring entities such as the para sequence of the acene series16and/or radicals of similar structure (Figure 8), which would be present at considerably lower concentrations due to their higher heats of formation. For the 2% runs the calculated concentration-time profiles for C20Hll(designated as the representative absorber of were integrated, as per (1). These curves show all the salient features of the recorded (I/Io) traces, i.e. the sigmoid shape following a delay, a relatively sharp rise, and a slow approach to saturation (Figure 9). The indicated computed times selected for minimal detection of absorption check quite well with the measured values (at 3X noise level) (Table IV). 11. The @ group consists of a sequence of absorption curves over a range of wavelengths (400-800 nm), obtained with the linear (16)Clar, E. Polycyclic Hydrocarbons; Academic: London, 1964. Clar, E. The Aromatic Sextet; Wiley: London, 1972. Stern, E.S.;Timmona, C . J. Electronic Absorption Spectroscopy in Organic Chemistry;Edward Arnold London, 1970. Karcher, W . , et al., Ed. Spectral Atlas of Poly cyclic Aromatic Compounds: D.Reidel: Dorchester, England, 1983.
Energy & Fuels, Vol. 2, No. 4, 1988 453
Investigation of Soot Precursors
Table IV. Exoerimental Conditions for Reoresentative Shocks ~
~
~~
At," ps
composition 2.0% toluene
pl? Torr 110 95 85 65 85 95 85 65 65 90 130 130 115
2.0% benzene
1 % allylbenzene
Opl: total pressure of fuel plus
T.5,K
ul,"/rs 0.893 0.943 0.962 1.02 0.926 0.926 0.926 0.980 1-00 0.840 0.877 0.885 0.922
103P.5,g/cm9 1.93 1.64 1.82 1.37 1.41 1.55 1.41 1.51 1.56 1.75 2.32 2.28 2.12
1612 1700 1725 1812 1767 1767 1767 1910 1950 1591 1640 1652 1710
exptl 340 100 70 40 240 380 360 50 50 230 130
calcd 320 50 240 60 40
100 80
Ar. * A t : interval between the onset of the reflected shock and the toe of the absorption trace.
0 0
-
0 4 0 0 nm
0 0
0 0 0 rm
N
0 5 0 0 - 7 5 5 nm
0
0 0 0
m
2 0
Eg e 0 0
C
4
*
-1.7-
0 0
8
-1.9-
a
,
104/T
5.1
5.3
\
I
5.5
5.7
I
5.9
6.1
0 0
00
400
800
1200
1600
2000
2400
2800
0 4 0 0 nm
3200
0 500
Time (Microsec)
Figure 9. Calculated transmittances for 2% CBHBshock heated to various temperatures. The m o w s indicate the times selected as the start of observable absorption. array spectrometer" (Figure 5b). Typical timewavelength spectra at two extremes are shown in Figure 7b,c. While no significant differences appear over a spectral range of 20 nm, there are clear, significant differences between the spectral scans at 811 and 392 nm. First, the delay times are shorter and the times for the rise of absorption are faster for the shock speeds at the shorter wave lengths, indicating more rapid rates for generating the smaller species that absorb in the UV region, compared with the much longer delay times for the appearance of the larger species, which absorb in the near-infrared region. Second, a two-step process appears at 800 nm, where the initial relatively fast rise is followed by a slower continued increase in absorption, demonstrating subsequent growth, since the absorption edge continues to move toward the longer wavelength. Two types of information are presented by these plots (recorded at 100 nm intervals, from 400 to 800 nm): delay times (ti) that measure induction times for the development of absorbing species, and rise times [(tf- ti)/2], which are mean inverse rates of production of these species. For any specified reflected shock temperature (T&,a plot of ti vs mean X has a positive slope, as expected for a sequential growth of absorbers with leading edges progressing toward the red. For any specified A, tiis longer the lower the shock speed (T& Because of the restricted range of fmal densities covered in these experiments, the half-times for attaining the f i i t saturation level do not permit us to determine whether the global process is first or second order. However, graphs of In it, = 106(ln2)/tlI2 (ps) v9 1/T5 (Figure 10) clearly show nesting of points for the sequence of A's, as expected, assuming that the leading edges of the absorption curves measure the larger units at longer wavelengths.
nm
---
v600 nm0 7 0 0 nm
-
0800nm-----
3.6-
b-
' 5.2
10'IT
--+
I
I
5.4
5.6
I 5.8
I 6.0
I 6.2
Figure 10. Plots vs T' of the logarithms of (a) the inverse of delay times, and (b) pseudo-first-order rate constants for shock-heated toluene, measured over a range of wavelengths. At any T5,the ku's are consistently larger when derived from 400-nm traces compared with 800-nm traces. The mean half-times for generating the larger absorbers increase with increasing size. The 800-nm absorbers require a larger activation energy (=30 kcal/mol) compared to those that absorb at 400 nm (=24 kcal/mol).
Acknowledgment. We sincerely thank Dr. R. J. Kee for copies of the Sandia computer codes, Professor C. F. Wilcox, Jr., for extensive discussions of the possibilities and limitations of the Diels-Alder addition sequence, and Professor J. H. Kiefer for discussions of his shock tube pyrolysis experiments on toluene and ethylbenzene. Registry No. C&B, 71-43-2; C7H8,108-88-3.
