593
OXIDATION OF SOOTBY OH RADICALS
Oxidation of Soot by Hydroxyl Radicals
by C. P. Fenimore and G. W. Jones General Electric Research and Deaelopment Center, Schenectady, New York
(Received August 12, 1966)
-
I n flame gases of temperatures from 1530 to 1890°K and PO, to 0.3 atm, the rate of oxidation of soot does not depend very strongly on PO,. The observations are consistent with the assumption that about 0.1 of the collisions of OH with the soot remove a carbon atom, the number of collisions being calculated from POHby kinetic theory.
Millikan' inferred that tiny soot particles in fuelrich hydrocarbon flames a t 1700 to 1800°K were oxidized by OH radicals rather than by the much larger concentrations of COz or HzO. Graham and coworkers2 also considered OH to be a very reactive oxidant. They proposed that the gasification of massive pieces of carbon in a very hot stream of fuel-rich combustion products was due to the removal of a carbon atom in about 0.03 of the collisions of OH with pyrolytic carbon, or in about 0.1 of those with commercial graphites. When O2 is present, OH might be less important. Lee and co-svorkers3 reported a rate of soot oxidation of
1.085 X 104Po,T-'/2e-39.3 koal/RT g ern+ sec-'
(1)
in flame gases, with PO, = 0.05-0.10 atm and T = 1350-1650"Kl as if O2 were the only important reactant when PO, was considerable. Neither Pon nor T was constant in any experiment, however, and it is uncertain how well the separate effect of each variable was untangled from the observations. We have measured soot oxidation in flame gases of constant PO,and T during each run, so that the variables were separated experimentally. I n the ranges PO,= 0.04-0.30 atm and T = 1530-189OoK, we find little dependence on PO,. Even in slightly fuel-rich gas where PO,is very small, soot oxidation remains considerable. Our rates are about five times faster than those reported by Lee and co-workers a t the same temperature and PO,. One way of explaining our weak dependence on Po, is to suppose that soot is mainly oxidized by OH in our flames. We measured the partial pressure of this radical in some of our runs and found that the observed oxidation rates could be accounted for by a collision
efficiency of OH on soot of about 0.1. If OH was important, the fact should be revealed by comparing the rate of oxidation in flames with the rate of a suspension of soot in dry gas where OH was absent. We could not make this comparison at the same temperature as we would have liked to. However, soot suspended in dry gas at 1200°K oxidized more slowly than would be expected from an extrapolation of our flame results. Furthermore, the rate of oxidation in dry gas depended on PO,. We think an additional mechanism besides oxidation by O2 is important in our flames, and we favor an oxidation by OH. The slower rates observed by Lee and eo-workers may reflect differences in the flames used; POHwas doubtlessly smaller in their gases than in ours a t the same PO,and T . It is possible that when PORis smaller, a reaction with O2 became more important] as is implied by their reported dependence on Pol.
Experimental Section Experimental Flames. Figure 1 is a sketch of the apparatus used to obtain fuel-lean sooty flames. A rich mixture of molecular proportions of C2H4 1.202 2.OAr was burnt on a porous burner4 to give a gas containing soot equal to about 1% of the carbon fed as ethylene. Sixteen per cent of the fuel remained in the products as hydrocarbons, mostly as acetylene and the rest as carbon oxides, steam, and hydrogen. The products were cooled, mixed with additional constituents, and finally burnt on a second burner. The
+
+
(1) R. C. Millikan,
J. Phys. Chem., 6 6 , 794 (1962). (2) J. A. Graham, A. H. G. Brown, A. R. Hall, and W. Watt, Sac. Chem. Ind. (London), 1958. (3) K. €3. Lee, M. W. Thring, and J. M. Beer, Combust. Flame, 6 , 137 (1962). (4) W. E. Kaskan, Symp. Combustion, Gth, 134 (1957).
Volume 71, Number 3 February 1967
C. P. FENIMORE AND G. W. JONES
594
I/
FUEL-LEAN,SOOTY FLAME
e -FUEL-LEAN
MIXTURE OF H2 02, CO2
COOLED COPPER TUBE, 8 5 c m LONG
+
-
WATER COOLED POROUS BURNER
t
FUEL RICH MIXTURE OF C2H,,Oz,Ar
Figure 1. Arrangement of apparatus to get a fuel-lean sooty flame.
soot gasified in an environment of constant PO,and T above the second burner. The second burner was a copper disk 2 cm in diameter by 0.6 cm thick with 220 0.1-cm holes running through it and a copper cooling coil soldered around it. Its flame consisted of a series of small cones, one per hole, whose flows merged into an approximately onedimensional stream of burnt gas 0.1 cm above the surface. The disks were made by W. E. Kaskan when he worked in our laboratory several years ago. The second flame usually became unstable if its temperature fell below 1600"K, although one acceptable run was obtained a t 1530". Temperature was measured by a fine quartz-coated thermocouple. The extent of reaction was determined by sucking a measured volume of gas through a quartz probe held a t various distances above the disk and catching the soot on a plug of Vycor wool in the cool leg of the probe. The probe was evacuat4ed a t "room temperature and then heated with a flame while oxygen was passed through it. The stream of oxygen and carbon oxides was run through hot copper oxide and the carbon dioxide trapped a t 113°K (isopentane slush). The trap was cooled further in liquid nitrogen and evacuated, then warmed, and its contents expanded into a mass spectrometer where the carbon dioxide was measured. The Jozirnal of Phuaical Chemistry
A correction was established by analyzing over-all fuel-lean runs in which the first burner was not ignited, so that all the reactants burnt a t once on the disk in a soot-free flame. The correction was 8% of the soot measured close to the disk, but a larger percentage further downstream. Measurements were not attempted after the correction had grown to 30%, which means that the region of study rarely extended farther than 1 cm downstream of the disk. Soot samples were also caught on carbon substrates in a similar probe and submitted to E. F. Koch of our laboratory for electron micrographs. These samples were strings of uniform spheres about 200 A in diameter before burning in the second flame and about 100 A in diameter after the soot content had fallen by oxidation to 0.22% of the carbon in the ethylene. Arrangement for Oxidation in Dry Gas. Soot, collected from the same flame used to generate it for the flame oxidations, was evacuated overnight at 5OO0K, then cooled and loaded into a closed vessel equipped with a sealed stirrer to disperse it. A small flow of NZ through the stirred vessel carried entrained soot to mix with a large volume of dry preheated gas. The large volume of preheated gas was NZ or Nz-OZ mixtures piped through a quartz tube of 1.3 em inside diameter. The tube was heated by two electric furnaces each 30 cm long. The gas was preheated in the first furnace, then mixed without much drop in temperature with about 0.1% of its volume of the soot suspension from the stirred vessel. The hot dilute suspension passed through the second furnace, and the surviving soot was caught on a plug of Vycor wool and measured as before. The soot added to the preheated gas was taken to be that which was measured when Nz containing less than 0.2% O2 was used in the quartz tube. This amount was a constant = t l O % , independent of the gas temperature in the second furnace, 1000 to 1200°K when a constant flow was passed through the dispersing vessel. The extent of reaction was measured by the decrease in soot recovered when the hot dry gas contained oxygen. However, the uncertainty of *lOojO in the soot added meant that the minimum reaction g of carbon rate observable was about 0.7 X cm+ sec-'. At 1140°K or less, the rate was certainly not larger than this when PO, = 0.21. At 1200"K, measurable rates were obtained for PO,= 0.12 and 0.21 but not for PO,= 0.05. Other Experiments. A few measurements were made of the rate of oxidation of soot containing 1% manganese by weight. This was made by saturating the argon flow to the soot-forming burner with volatile manganese-methylcyclopentadienyl-tricarbonyl. The
OXIDATIONOF SOOTBY OH RADICALS
595
y;,
\ SOOT IN FLAMES
Po2'. 29 5
Po, (AT EQUILIBRIUM) 0.30
0.3
1690 O K
=(0.4TO 81x
LEE i t . a]. \ \
Pa, = 0 . k TO 0 IO
'4
SOOT SUSPENDED ' o'2' IN DRY GAS OP0, z 0 . 1 2
5
Figure 3. Reaction rate us. 1/T. Flame data from Table I. The range of observations by Lee and co-workers in the soot cloud above a hydrocarbon diffusion flame is indicated. The lower two points give measurements of oxidation rate in dry N r O t mixtures.
TIME, MILLISECONDS
Figure 2. Cube root of the soot content us. time. Soot k expressed as a percentage of the carbon fed in the ethylene; time is inferred from the known gas velocity.
yield of soot was approximately the same whether it contained manganese or not.
Results We assume that the soot consists of uniform spheres which react, at a constant rate per unit of area, with a surrounding gas of fixed temperature and composition. From geometrical considerations, the gasification can be expressed cent soot ) (-per per cent sooto
= 1 -
2R(t
- to)
9
I O4IT
6'
(dldo) =
7
(2)
PdO
where d is the diameter of a particle, per cent soot equals the carbon present as soot, expressed as a percentage of the carbon fed in the ethylene, R the reaction rate in g of carbon ern+ see-l, t the time, and p the density of a particle. The zero subscript refers to some arbitrary point at which d = do. Figure 2 gives the flame data in plots of (per cent soot)*" us. time. From each curve, a value of (2R/ p d 0 ) 0 . 6 appropriate to eq 2 was computed and listed in
Table I. I n these values do is the d when (per cent soot)'/* = 0.6; it is about 1 X cm and unchanged from run to run; therefore, the values of (2R,/pdo)o.6 are about lo6times R itself. Values of 10-6(2R/pdo)o.6= R are plotted against the reciprocal of temperature in Figure 3. The six runs with appreciable PO,establish a line with activation energy of about 22 kea1 mole.-' The three slightly fuel-rich runs lie below this line, but surprisingly close if the oxidation is truly by 02. The slower, more oxygen-dependent rates reported by Lee and co-workers are indicated. Figure 3 also contains two points for the oxidation of soot in dry gas at lower temperatures. These rates are less than the values expected from a long extrapolation of the flame data. Furthermore, the oxidation in dry gas depends more strongly on PO,than does our flame oxidation. It is probable that our flames possess an additional mechanism besides oxidation by 02. Figure 4 shows a plot of (2R/pdo)o.sagainst the calculated equilibrium POH.Since POH was doubtlessly larger than the equilibrium pressure,b we determined ( 5 ) W.
E.Kaskan, Combust. FZame, 2 , 229, 286 (1958).
Volume '71, Number 9 February 1967
C. P. FENIMORE AND G. W. JONES
596
Table I: Rate of Gasification of Soot in Flame Gases Gas composition, mole fractionCOZ HsO
T, 02
OK
0.437 0.608 0.561 0.510 0.676 0.658 0.574 0.586 0.591
0.295 0.149 0.143 0.145 0.039 0.038 0.04 x 10-4 0.12 x 10-4 0.83 x 10-4
1690 1530 1745 1887 1670 1710 1726" 1800" 1820"
0.166 0.132 0.161 0.187 0.158 0.168 0.213 0.232 0.210
----. Ar
GO
aec -1
0.102 0.149 0.135 0.158 0.127 0.136 0.155 0.145 0.182
...
620 310 700 1090 460 520 270 450 510
... ...
... ... ... 0.026 0.034 0.016
PO, is the calculated equilibrium pressure for the three slightly rich runs.
/
/
/
corrected curve would be consistent with what is known about excess OHP i e . , the ratio of actual POH to the equilibrium pressure would be larger, the lower the temperature or the smaller PO,.
/
y 7 4 5 P I690
4 - 1
Table 11: Ratio of Actual POH to the Calculated
1690
Equilibrium Pressure in Three Runs
OK
Po2
POH, equil
POH,actual POH,equil
1690 1530 1710
0.295 0.149 0.038
5.3 x 10-4 1.2 3.8
4 zk 0.8 8zk2 5fl
T,
I
I I
0
2x10-3
POH, ATMOS.
The number of collisions of OH radicals with the soot is calculable from POH by kinetic theory. If a fraction, CY, of these abstracts a carbon atom, R = 3 X 1 0 - a ~g~of carbon cm-2 sec-' when PoH = 10-3. Inserting this R into
Figure 4. Upper curve is a quantity proportional to R plotted against the calculated equilibrium POH. Lower curve is the same quantity us. the measured POH for three of the runs.
the actual POHin three lean runs by substituting NzO for part of the COz added in the mixing chamber and observing the rate of reaction 3 in the region of interest
0
+ NzO
2N0
(3)
The [O] was determined from the measured [NzO], d[NO]/dt, and T , the rate constant being known,6 and POHwas then inferred from equilibrium (reaction 4)which was certainly balanced.6
0
+ HzO = 2 0 H
(4)
The results of the measurements are given in Table
I1 and a curve for these data is also drawn in Figure 4
(2R/pdo)o.6 C% 300 and putting p = 2, do = 1 X we get a collision efficiency of CY cv 0.1. Comparison with Lee and Co-workers. PoH was doubtlessly smaller in Lee's gases than in ours even for the same final PO,and T. We added Hz to help stabilize our flames on the second burner; its combustion generated many radicals, but Lee, et al., did not have a similar source of radicals in their experiment. Furthermore, their observation times were 10 to 25 times longer than ours, so any excess radicals should have decayed farther toward equilibrium. Their linear dependence of the rate on Po,, expression 1 in
passing through the origin with a slope of (2R/pdo)o.e = 300 f 100 per atm of POH. A displace(6) C. P. Fenimore and G. W. Jones, Eighth Symposium on ment of the other points so that they too fall on the bustion, Williams and Wilkins, Baltimore, Md., 1962, p 127.
Com-
OXIDATION OF SOOT BY
'i 0.3
OH RADICALS
597
l/402
*O A0
-
1820 O K
+ '/2H20= OH
AH
=
37.8 kcal
(5)
oxidized their soot with a collision efficiency, a. The expected rate under these assumptions is
Po2 N ZERO
O A
1-63 )( 104aPo,'/4PH,o'/2T'/*e-3'1*8/RT
A*A
(6)
which has values of the same order as expression 1. When a = 0.1,Poz= 0.075,and PH,O= 0.2,expression 6 predicts rates 60% as large as expression 1 in the range
1710'
1350-1650°K.
TIME, MILLISECONDS
Figure 5. Circles represent data already shown in Figure 2; triangles r epresent similar measurements in the same flames for soot containing 1% Mn.
the introduction, suggests that when POB. is only the 0.025, equilibrium pressure and when PO,= 0.075 the oxidation by 0 2 is important. Only a small range of PO,was examined by them, however, and perhaps the dependence on this variable was not established beyond question. We want to add that if their rates did not really depend much on PO,,it is possible that an oxidation by OH was important for them too. To show this, we write out the expected rate of reaction if OH, equilibrated according to reaction 5
*
We have no reason to doubt that an oxidation by O2 was important in Lee's work, but an oxidation by OH was probably important too. Catalysis by Manganese. Soot containing 1% manganese by weight oxidized at a lower temperature than soot free of manganese. When heated in a stream of oxygen, initially at room temperature, and the temperature raised 6" min-I, the maximum rate occurred a t 680°K for the catalyzed and at 800°K for the uncatalyzed soot. Figure 5 shows that manganese also increased the rate of reaction in flame gases having appreciable PO,, but by only about 20%. Although the catalysis may not look very clear-cut in Figure 5, it was easy to see in the flames themselves that the addition of manganese in the soot contracted the yellow carbon zone. I n the slightly rich gases, the appearance of the carbon zone was unaffected by manganese in the soot, and no catalysis is seen in Figure 5 for the rich gas. Since the catalysis only occurred in lean gas, the catalytic reaction was probably an oxidation by Or. Perhaps it is reasonable to add that the oxidation in slightly fuel-rich gas was probably not a reaction with 0 2 because it was not catalyzed. Since the rate in slightly fuel-rich flame gases was of the same order as the rate when Pozwas appreciable, this may be an additional reason for believing that most of the reaction of soot occurred with some other oxidant than O2 even when Pozwas appreciable. Achowledgment. We are grateful to G. E. Moore for helpful discussions.
Volume 71, Number 3 February 1067