794
ROGERc. M I L L I K 4 N
present research. These data also make it possible to estimate similar functioiis for neighboring sulfides which have not been available previously. Some other values available in the literature also
Vol. 66
have been included for convenience and reference. Acknowledgment.-The authors are indebted to Professor Alan W. Searcy far his helpful comments on the manuscript.
NON-EQUILIBRIUM SOOT FORMATION I N PREbTIXED FLAMES BY ROGER C. MILLIKAN General Electric Research Laboratory, Schenectady, N . Y. Received August I?’, 1961
Soot deposition has been studied in a series of large, flat, premixed ethylene-air flames burning on porous metal burners. The mixture ratios bracketed those at which soot first appears. Using combined optical and probe sampling techniques, spatially resolved measurements were made of C2H2, CH,, OH, and soot concentrations as functions of burned gas temperature. It was found that acetylene and methane (in smaller amounts) are produced in the reaction zone and survive to form about 1% of the burned gas. The pyrolysis of these hydrocarbons competes with an oxidation reaction involving OH. The bdance between these reactions determines whether or not soot is set free. The dark space betrsTeen the reaction and soot zones is a region of high oxidizing power due to the OH concentration in excess of equilibrium. The critical mixture ratio a t which soot appears is temperature dependent. This temperature dependence leads to a value of -34 f 10 kcal./ mole for the difference in activation energy between the soot oxidation and formation reactions.
Introduction Soot formation occurs in many premixed flames when it should not-if chemical equilibrium were attained. The onset of soot deposition in certain ethylene-air flames can be observed for mixtures with a carbon to oxygen atom ratio as low as 0.58. If equilibrium prevailed, soot would not be set free until this ratio exceeded unity. That is, even though there is more than enough oxygen present to oxidize all the carbon to CO, this does not happen: Instead soot is liberated. Such non-equjlibrium behavior was established by Street and Thomas1 for premixed flames of a variety of fuels burning in air. This aspect of soot formation is but one of the puzzles2which adds scientific interest to a problem of obvious industrial importance. The purpose of this work was to find an explanation for the occurrence of non-equilibrium soot deposition, and to elucidate the mechanisms involved. Theories of soot formation are many. They have been well summarized,i.2 and will not be recounted here. The difficulty lies in the lack of sufficient experimental data on a well characterized system to limit the possibilities to one mechanism. We have chosen for study a set of large, flat ethylene-air flames vhose mixture ratios bracket those at which soot first appears. By using a combination of optical and probe sampling techniques, an experimental description of these flames has been obtained which is complete enough to severely limit speculation on the mechanisms of soot formation. A key point in the present work has been a determination of the temperature depeiidence of soot formation. This turned out to be an important factor which has not been recognized in earlier work. Experimental The flames studied were burned on porous metal burners of the type developed by Kaskan.8 Burners used ranged (1) J. C. Street and A. Thomas, Fuel, 81, 4 (1956). (2) A. G . Gaydon and H. G. Wolfhard, “Flames, Their Structure Radiation and Temperature,” Second ed., Chapman and Hall, Ltd., London. 1960, pp. 175-209.
from a 2.5 X 5 em. rectangular one to circular ones 7 cm. in diameter. All were of the compound design so that the flame could be shielded from its surroundings by a sheath of Nz ga;. The use of such porous metal burners aided this work in two ways They gave large flat flames which could be probed optically with good spatial resolution. More important is that on such burners the burned gas temperature becomes a variable independent of mixture ratio. Consequently we have been able to observe the temperature dependence of soot formation at constant over-all composition. Since the gas flow is normal to the burner a t a calculable velocity, the scale of distance from the burner can be converted to a time scale. For all the flames considered here this conversion is of the order of 1 msec. per mm. Measurements have been made of burned gas temperature, GH2, CH,, OH, and soot concentrations. In addition, the critical mixture ratio a t which soot luminosity appears has been determined. Most of the techniques of measurement have been described previously. The following is only a brief account of the procedures used. A. Apparatus.-The ethylene was of C.P. grade. Air was supplied from a 2000 p.s.i. reservoir and compressor. Both were metered with calibrated critical flow orifice meters to within 0.5%. Details of the narrow beam optical setup have been given.*r6 B. Flames.-The ethylene-air flames used all fell within the region defined by the following ranges of these variables: mixtureratio 0.5 < atomic C/O < 0.7, linear gas velocity 7 < uZ5 < 20 cm./sec. NTP, temperature 1600 < T < 2000 K. The characteristics of some of the indivldual flames used are given in Table I. Noted there for each flame is the atomic C/O ratio +, the linear gas velocity entering the flame, vTg, and the maximum burned gas temperature. C. Temperature Measurements.-Gas temperatures were measured in one case by a fine wire thermocouple. Such couples soon became brittle and broke when exposed to these reducing flames. Subsequent measurements were made by the sodium &line reversal m e t h ~ d . They ~ are believed accurate to f 5 0 ” K . Soot luminosity and scattering from even the richest of our flames were so feeble that no errors in the temperature measurements arose from that source. D. Concentration Measurements.-The OH radical concentration was determined by the ultraviolet line absorption method of Kaskan.6 Since the level of OH was low, it was necessary to pass the light beam twice through a 7 cm. (3) W. E. Kaskan, “Sixth Symposium (International) on Combustion,” Reinhold Publ. Corp., New York, N. Y., 1957, p. 134. (4) R. C . Millikan, J. Opt. Soc. Am., 61, 535 (1961). (6) R . C. Millikan, Combustion and Flame, 6, 349 (1961). (6) W. E. Kaskan, ibid., 2, 229 (1958); J . Chem. P h w , 29, 1420 (1968).
May, 1962
795
KON-EQUILIBRIUM SOOTFORMATION IN PREMIXED FLAMES TABLE I FLAME CONDITIONS
Atomic C/O
vzs, cm./seo.
0.583 .53 ,663 .577 .607 .640 ,673 ,710
7.0 19.5 14.0 11.5 11.6 11.7 11.7 11.8
I800 r Tmax,
OK.
1720 1950 1900 1820 1820 1820 1815 1810
diam. flame to secure enough absorption for adequate measurement. Acetylene and methane were measured at high levels in the flames by extracting samples with fine quartz probes and analyzing them with a mass spectrometer. Similar analyses for COz were shown to agree with values derived from equilibrium calculations,6 so that the probe analyses are thought t o be reliable. Concurrently, measurements were made of the emission and absorption of infrared light by acetylene at 13.7 p . The results of the probe analyses were used to calibrate the infrared method of measuring CzHz. The infrared method then was used to determine the spatial distribution of CzHz in different flamesS5 The soot produced has been characterized by measurements of light absorption and emission by the particles in the flames over the 4000-10,000 A. wave length range.4 Soot deposits from the flames were studied similarly by light absorption. Particle sizes were determined by electron microscope examination of those caught on a cool target in vucuo after being extracted from the flame by a quartz critical flow orifice.7-8 The equilibrium composition for the burned gases of all our flames have bemen obtained from the published tables of Fremont, et ul.,g u.sing the measured values of flame temperatures.
Results The temperature and emission profiles for flame 1 are given in .Fig. 1. These are for a highly quenched flame that is just rich enough to show soot luminosity. There is a distinct dark space at 4 mm. from the burner which is bounded on one side by the reaction zone emission, and on the other side by the increasing emission of soot. Such a dark zone has been olbserved previously by others using flat flames.lOJ1 Since it represents a delay in the deposition of soot, the dark space is an important feature of the flame which must be accounted for by any proposed theory of soot formation. It should be noted t8hatthe ordinates in Fig. 1 are different for the d.ifferent species. The sensitivities for CH and aioot are very great, while that for CzHz is low. The behavior of the OH radical concentration is shown in Fig. 2 for a non-luminous flame above and for a sooty flame below. The dashed lines give the calculated equilibrium OH concentrations. The equilibrium values fall off slowly due to the gradual decrease in burned gas temperature (cf. Fig. 1). (7) R. C. Millikan, J . Opt. Soc. A m . , 51, 698 (1901). ( 8 ) R. C. Millikan, “Fourth Symposium on Temperature, Its Measurement and Control in Science and Industry,” Vol. 3, in press, 1901. (9) H. A. Fremont, H. N. Powell, A. Shaffer, and S. N. Suoiu, “Properties of C o m h s t i o n Gases, System CnHzn-Air,” General Eleatric Company, 19!%. (10) C. P. Fenimore, G. W. Jones, and (2. E. Moore, “Sixth Symposium (International) on Combustion,” Reinhold Publ. Corp., New York, N. Y., 1957, p. 242. (11) J. M. Singer and J. Grumer, “Seventh Symposium (International) on Combustion.,” Butterworths Scientific Publ., London, 1959, p. 559.
rnrn FROM BURNER
80 . [-I t
t
Y)
CH 4312
SOOT 57008 0 v)
=B
20
2
0
4
6
8
1
0
rnrn FROM BURNER.
Fig. 1.-Above, temperature profile for soot forming flame 1: 0, determined with a thermocouple. A determined by Na D-line reversal. Below, emission profiles for the same flame. The location of the lower abscissa scale is uncertain with respect to the upper one by 0.5 mm. Ordinate scale is different for each species.
-k
C/o ~0.53 T = 1950’K
----EQUILIBRIUM^
-2--2-
kl REACTION ZONE
2
4
6
8
IO
m m FROM BURNER, Fig. 2.-Concentration profile for OH: o,non-luminous flame 2; 0, soot forming flame 3. Location of reaction zone indicated by CH emission.
The agreement between the calculated equilibrium values and the measured ones at large distances from the burner is within the experimental accuracy. The onset of soot luminosity occurs abruptly at a reproducible mixture ratio when one begins with a non-luminous flame and gradually makes it richer in fuel. We found that this critical mixture ratio,
ROOERC. MILLIKAN $CR, at which
soot deposition begins varies with the The important variablc which is being changed when the flow velocity is varied is thc burned gas tcmperaturc. This temperature was detcrmined for five of the critical ratio flamcs, and the results are shown in Fig. 3. Thc mixture ratio at which soot appcars is sccn to be tempcraturc dcpendcnt. This observation should he quantitativcly cxplaincd by an adcquate theory of soot deposition. The amount of acetylene and methane produced in the reaction zone and surviving into the burned gas next was measured for flames of different mixturc ratios, but all with temperatures between 1800 and 1820'K. The data obtained by probe analyses 12 mm. above the burner are shown in Fig. 4. It is clear that as the mixture ratio is increascd through that a t which soot appears, the acetylene concentration continues to rise smoothly. The same is true of the methane, but the rise is more gradual. Other hydrocarbons were not detcctcd by thc mass spectrometer. It is possible on the porous burner to burn a flame which shows considerable soot luminosity at low flow veloeitics, but which hecomcs nou-luminous at high flow velocitics, the mixture ratio being kcpt constant. The visible emission of soot and thc infrared emission of C2H2were measured 8 mm. above the burner for a set of such flames of constant composition. The data are givcn in Fig. 5 plotted against the measured temperature of the flames. The low tcmperature portions of the curves are dotted since those flamcs wcre wcak and unsteady. Edge effectswere more important for thcm, and no doubt account for the fall in soot emission for the 1625O point. The data are plotted as obtained, and have not been adjusted to account for the tempera, ture dependence of thc black body function. Were this done, the C2H1 curve in terms of concentration would he little affected because of the long wave length, but the soot curve would become much steeper. Evidently the acetylene concentration is little affected by the flame temperature, but the soot concentration is very sensitive to it. The data of Fig. 2 showed that the concentration of OH was considerably above equilibrium for a short distance downstream of the reaction zone. This suggestcd that the dark zone might be a region in which oxidation processes still are dominant. To see if this were so, the following experiment was performed. A sooty flame similar t o flame 1 , which showed a distinct dark space, was lit. A one mil platinum wire was slowly lowered into the soot zone while the tip of the wire was observed through the telescope of a micro optical pyrometer. During the coursc of a few minutes the tip became covered with a deposit of soot. This could he distinguished because of the different emissivities of platinum and soot. The wire next was lowered until its tip extended into the dark space of the flame, but not into the reaction zone. The soot deposit on the end of the wire was seen to slowly disappear until the wire was hare. This deposition of soot in the luminous zone and its removal in the dark space could he repeated over and over. It is no wonder that soot cannot be deposited in the dark space, gas flow rate through the porous burner.
Z
SWT LUMINOSITY OBSERVED
1750
ATOMIC
