1020
G. H. AIEGUERIBN
T-01. 67
THERMAL STUDIES OF COOL-FLAME OXIDATIOXS BY G. H. MEGUERIAN Research and Development, American Oil Co., Whiting, Indiana Received September g4, 1961
A thermal method for the study of cool-flame oxidations in laminar-flow systems has been developed. The empirical equation ATm = B [ F ]L ( [OZ] - [Onla} relates the maximum temperature-rise to oxygen and fuel concentrations in terms of two constants: the exothermicity constant, B, and the minimum oxygen requirement, The value of B varies directly with the residence time. The value of [ O ~ ]changes O slightly or remains [02]0. constant depending whether the residence time is varied by changing the gas floxr-rate or by moving the thermocouple along the axis of the reactor. The cool-flame oxidation of isooctane is enhanced by n-heptane and inhibited by propionaldehyde and aromatic amines. Initiation is suggested to result from the activation of oxygen molecules on the surface of the reactor through formation of *-complexes with the carbonaceous layer. The latter forms during aging of the reactor and is replenished during the oxidation.
Introduction Despite the large amount, of work done and several elementary reactions proven possible,l,Z the over-all mechanism of hydrocarbon oxidation in the cool-flame region, between 250 and 450') is still not clear. For this reason, interpretat,ion of results obtained by different methods under various conditions is very difficult. The task is even more difficult with comparative results of hydrocarbon reactivities, the order of which sometimes changes with experimental conditions. Further studies are needed to elucidate the effect,s of surfaces, temperatures, and other experimental variables upon the various steps of oxidation. The gas-phase oxidation of hydrocarbons proceeds by a series of chain reactions of the degenerate-branching type.3 The over-all reaction js highly exothermic, and may produce cool flames and explosions. This exothermicity puts restrictions on the kinetic methods us.ed. In the static method, to assure isothermal conditions or avoid explosions, oxidations must be carried out a t sub-atmospheric pressures, low concentrations, or low t,emperatures. Under these coiiditioiis the kinetics of the reaction can be conveniently determined from pressure rise, but the analysis of the reaction mixture is very difficult'. Also, hydrocarbons of greatly different reactivities often must, be studied a t different temperatures4 and pressures. Because the relative iniportance of the various steps of oxidation Do the overall reaction may change with pressure and especially with temperature, this procedure may lead to inaccurate comparisons. These difficulties can be alleviated by use of the flow method. By varying the residence time, the extent of oxidation can be kept low and, therefore, wider ranges of pressures, concentrations, and tempera,tures can be used. This flexibility may also make it possible to exaggerate and study the various steps of oxidation. Cool-flame oxidatioiis in flow systems have been followed by measuring consumption of reactaiit's or formation of products, such as C0.5,6 Because of in(1) A. S. Sokolik. "Autoignition, Flames. and Detonation in Gases." Academy of ScienceE, U.S.B.R., Moscoa, 1960; C. F. 13. Tipper, Quart. Rev., 11, 313 (1957). (2) N . N. Semenov, "Some Prohlenis in Cheriiical Kinetics and Reactivity," Academy of Sciences, U.S.S.R.. Moscow, 1954. (3) N. N. Semenov, "Chemical Kinetics and Chain Reactions," Oxford Univ. Press, 1935. (4) C. F. Cullis and C. N. Hinshelwood, Discussions Faraday Soc., Z, 117 (1947). ( 5 ) M. R. Nelman and A. F. Lukovnikov. "Chain Reactions of Iiydrocarboil Oxidation in the Gas Phase," Academy of Sciences, U.S.S.R., l\Ioscowr, 195:, p. 140. (6) K. C . Salooja, Combustion and F l a m e , 4 , 193 (1960).
herent difficulties, the most direct method, namely the measurement of temperature rise, has attracted little attention. However, it has been applied with some success to kinetic studies in static systems78and recently, in this Laboratory, to the comparative study of hydrocarbon reactivitiesa9 This method has now been refined and adapted to more precise studies of cool-flame oxidations. A cylindrical reactor was modified to give laniiiiar flow and to minimize heat losses. An empirical expression for the maximum temperature rise was derived in terms of initial oxygen and fuel concentrations. Although this expression can be used to compare reactivities of various hydrocarbons, in this study only results with isooctane were used to demonstrate the method and to suggest a mechanism for the thermal initiation of cool-flame oxidations on the reactor surface. Experimental Isooctane and n-heptane, both ASTRl grade (Phillips) were further purified by percolation through silica gel and A1203 before use. Propionaldehyde (Eastman) was similarly purified. Aniline, N-methylaniline, and N,N-dimethylaniline (all Eastman) were distilled under reduced pressure, and a middle fraction of each amine boiling a t constant temperature was collected and stored under nitrogen. Fresh mixtures were made as they were needed. Cylinder oxygen and nitrogen of high purity were used without further purification. The apparatus consists of fuel measuring and delivery systems, a preheater-reactor assembly, a furnace, instrumentation for controlling and measuring temperatures, and the conventional system for handling and metering gases. The final choice of design was based upon trial and error to achieve maximum reproducibility and the greatest ease of operation. By means of a Rideal Microdoser'o submerged in a constanttemperature bath the fuel is metered and, through a capillary tubing, delivered t o the upper bulb of a vertical carburetor. The two bulbs of the carburetor, which can be heated separately, assure the complete evaporation of the fuel and its better mixing with the stream of nitrogen. To obtain a constant and regular flow of fuel, the bath temperature is kept constant within 0.005" a t 25', and the carburetor temperatures are adjusted so that the fuel evaporates completely without causing bubbles to form in the capillary side arm of the carburetor. From the carburetor, the fuel-nitrogen mixture enters the outer chamber of the 70 cm.-long preheater and flows to the mixing tip, as shown in Fig. 1. Metered oxygen flows through the inner chamber. The gases, mixed by the reversal of the oxygen flow, enter the reactor through three slots, which are cut into the tip and are 1 mm. wide and 3 mm. long. The gases are released into the air through an exhaust tube. The preheater and the reactor are connected by means of tightly fitting standard tapered (14/35) ground joints. To minimize heat losses through the walls, the reaction chamber is enclosed in a jacltet which is (7) M. Vanpek, Bull. soc. ctmm. Belg., 64, 235 (1956). (8) R. H. Burgess and J. C. Robb, T r a n s . Faraday S o i . , 64, 1015 ( l Y 5 8 ) (9) J. F. Bussert, Combust. Flame, 6, 245 (1962). (10) R. 0. King and R. R . Dai-idson, C a n . J . Res., ZlA, 65 (1943).
May, 1963
THERMAL STUDIES OF COOL-FLAME OXIDATIONS
silvered from the inside, evacuated, and sealed. To minimize heat losses through the mixing tip, it is placed just outside the chamber. To prevent reactions from occurring before gases enter the chamber, the clearance between the tip and the reactor wall is made less than 1 mm. Temperatures of reactant mixture and exhaust gases are sensed by thermocouples TMand TE and recorded continuously by a duplex recorder. Chromel-alumel thermocouples with bare junctions, about 1 mm. in diameter, are used; sheating of the junctions with Pyrex glass gave identical results. The preheater-reactor assembly is placed in an electrically heated cylindrical furnace. A program controller provides a constant rise of 5’ per minute over the entire temperature range studied. The operating procedure was carefully standardized for maximum reproducibility. With the rates of nitrogen and oxygen flows adjusted, the furnace is heated to about 150” and kept there until equilibrium between mixture and exhaust temperatures is established. The program controller is then activated, the fuel is turned on, and the run is continued until the mixture temperature reaches 450’. The reactor is cooled by blowing air from the outside while a slow flow of nitrogen is maintained through it, and the procedure is repeated for the next run. From the tracings of mixture and exhaust temperatures, the temperature rise, A T , due to the oxidation of the fuel, is easily determined a t any mixture temperature. Irregular fuel-flows due either t o wide fluctuations of the bath temperature or to badly adjusted carburetor-temperatures give wavy traces of exhaust temperatures; they are poorly reproducible and may lead to erroneous conclusions. With good operative conditions smooth traces are obtained which give reaction temperatures reproducible within 3 O and maximum temperature rises, within 1O . Because of breakage, three reactors with volumes of 11, 13, and 14 ml. were used. All three had about the same inside diameter of 22 mm. but, because of the tapered construction on both ends of the reactor, the surface to volume ratios were not the same. S o attempt was made to determine these ratios precisely. All new reactcrs were washed with nitric acid and thoroughly rinsed with distilled water before use. Freshly cleansed reactors gave, a t first, weak exothermic reactions. The intensities increased with subsequent runs, but reproducible results were obtained only after 4 hours of continuous oxidation of isooctane a t 350”. During “aging,” the reactor surface was coated with an invisible carbonaceous material that could be removed by oxidation with oxygen rapidly above 350°, but only slowly a t lower temperatures. Washing with nitric acid or acetone also destroyed the coating.
Results A typical plot of temperature rise, AT, a t various mixture temperatures is shown in Fig. 2 for isooctane. This plot, the thermal profilell of the cool-flame region, shows that the highly exothermic cool-flame reactions begin a t temperature Ti and reach maximum intensity Beyond this temperature their intensity a t T,. gradually decreases to zero. The steep S-shape of the low-temperature half of the thermal profile also shows that the rate of temperature change increases with temperature and passes through a maximum; the maximum temperature rise, AT,, is attained a t d(AT)/dT = 0. Although any AT or d(AT)/dT on the thermal profile can be used to study cool-flame reactions, AT, has been preferred because of the greater reproducibility of both T , and AT,. Effect of Concentrations of Reactants.-The thermal properties of cool-flame reactions for several series of runs a t various oxygen and isooctane concentrations are given in Table I. Nitrogen was used as diluent to bring the total gas flow to 512 cc. per minute a t 25”. Although increasing oxygen concentrations reduces the temperature, T,, a t which exothermic reactions begin, the effect is small; the difference between the (11) G. H. Meguerian and J. F. Bussert, J. Chem. Eno. Data, 7. 127 1962)
1021
PREHEATER
Fig. 1.-The 40
REACTOR
thermal apparatus.
I
I
I
I
375
400
425
30
6-1 20
IO
0
300
325
350
MIXTURE TEMP., “c.
Fig. 2.-Thermal
IO0
properties of cool-flame reactions.
125
I50
LO;], x
Fig. 3.-Effect
io5,
2 00
I75
2 25
MOLESKCA T 2 5 ~ .
of oxygen concentration.
highest and the lowest temperatures being only 15’. The effect of isooctane concentration on Ti seems to be within the experimental error. The maximum temperature rise, AT,, occurs over a temperature range of 5 to 6’. Values of Tm vary in general with the size of AT,, the highest AT, occurring a t the lowest mixture temperatures. The maximum temperature rise, AT,, depends on the first power of either reactant concentration when the other is kept constant. Figure 3 shows the existence of a limiting value for the oxygen concentration, [ 0 2 ] 0 , given by the intercept, below which no exothermic reactions occur regardless of the isooctane concentration in the mixture. However, no such limit exists for the isooctane concentration, since all the lines pass through the origin. These relations can be expressed mathematically
AT,
=
B [F]i
lo~li-
[02]0
1
(1)
where [F]i and [ 0 2 ] i are the initial concentrations of isooctane and oxygen, and B is a proportionality coiistant, called the “exothermicity” constant. An average
G. H. NEGUERIAN
1022
value of 3.15 f 0.08 X 10I2 deg. cc.2/moles2was obtained for B a t 4 isooctane and 5 oxygen concentrations. TABLE I EFFECT OF OXYGEXA N D ISOOCTAXE CONCENTRATIONS O N COOLFLAME REACTIONS Isooctane, moles/cc.
x
106
0.57 .57 .57 I57 1.18 1.18 1.18 1.18 1.18 1.70 1.70 1.70 1.70 2.22 2.22 2.22 2.22
Oxygen, moles/cc.
x
106
1.60 1.80 2.20 2.40 1.40 1.60 1.80 2.00 2.20 1.40 1.60 1.80 2.00 1.40 1.BO 1.80 2.00
T,,“C.
330 327 324 324 335 328 324 323 322 330 327 326 326 330 330 324 320
T,, O C .
ATm, ‘ C .
360-365 360-365 360-365 355-360 355-360 355-360 352-357 355-360 355-360 353-358 350-360 348-354 348-352 353-357 353-357 318-3 52 345-350
6.5 10 18 21 6 13 20.5 28 35 9.5 20 31.5 42.5 12 25.5 39.5 53
Effect of Residence Time.-The effect of residence time was studied by varying (a) the total flow rate of the gases through the reactor, and (b) the position of the thermocouple along the axis of the reactor. The last method was based on the assumption that under the laminar-flow conditions used there was little or no longitudinal mixing within the reactor. When the flow rate was varied, the thermocouple position was held a t its usual position-1 cm. from end of reactor-and when the thermocouple position was varied, the flow rate was held constant a t 512 cc. per minute. The residence time, At, was defined as the ratio of reactor volume to total flow rate per second a t 25’. The reactor volume was estimated from the position of the thermocouple: a position of 4 cm. from the tip of the preheater being equivalent to 12.6 ml. Results a t six different flow rates giving residence times from 0.74 to 2.52 seconds show that the oxygen requirement varies inversely with the residence time, but the effect is not pronounced: about 3-fold increase in the flow rate increases the value of [Ozlo by only 20%. The exothermicity constant, on the other hand, increases linearly with the residence time according to the relation
B = k,(At) (2) Substitution of this expression for B in equation 1gives, after rearrangement
where IC, is an empirical constant related to the over-all rate constant for exothermic reactions a t the peak temperature. A value of 2.04 X 10I2deg. cc.2/moles2 sec. for k , has been determined from equation 2. Results for four different positions of thermocouple show that the oxygen requirement appears to be independent of the therinocouple position, the variations being within the experimental error. The exothermicity constant, on the other hand, increases as the thermo-
T’ol. 67
couple is moved farther away from the preheater tip, or as the effective length, L, of the reactor is increased according to the expression
B
=
CL or B
=
k,(At)
since L determines the effective volume of the reactor and hence, the residence time. From results a t four different locations, Ice is calculated to be 2.22 X 1OI2 deg. cc.2/moles2 see., which compares very well with that obtained by varying the flow rate. Mixtures of Isooctane with Various Compounds.Compounds, such as n-heptane and propionaldehyde, that undergo intense cool-flame reactions should enhance the reactivity of isooctane. The minimum oxygen requirement and the exothermicity constant for isooctane are compared with those of isooctane containing 2.5, 5.0, and 10 mole % ’ n-heptane in Table 11. Two different fuel concentrations of the 5 mole % nheptane mixture, like pure isooctane, give the same value for [OZ]O. Although the value of B increases with the amount of n-heptane in the mixture, the effect is much larger on [ 0 2 ] 0 : with only 10 mole % n-heptane, the value of [OZ]O was reduced by about 50%. TABLE I1 EFFECTOF IIL-HEPTANE O N COOL-FLANE REACTIOXS OF IsoOCTANE
n-Heptane In mixture, mole %
Total fuel concn., moles/cc. X lo6
moles/cc. X IO5
Sone 2 5 5 0 5 0 10 0
1 182 0 654 1 182 0 654 0.654
1 12 0 93 .70 .71 .48
~0210,
B X 10-12, deg. cc.Z/moles2
3 3 3 3
02 34 94 92 4 63
Propionaldehyde, on the other hand, inhibits the initiation of the exothermic reactions and enhances their intensity only slightly. Five and 20 mole % propionaldehyde in isooctane increase the minimum 0 2 requirement from 1.23 X mole/cc. to 1.43 X and 1.59 X and the constant B, from 3.17 X 10I2 deg. cc.Z/moles2 to 3.24 X 1OI2 and 3.60 X Furthermore, a run with isooctane alone immediately following this series gave a low AT,. The correct value for AT, was obtained only after re-aging the reactor with isooctane for about one hour. Oxidation inhibitors, such as aromatic amines, raise the niininium oxygen requirement and lower the value of B. Results of experiments with isooctane containing various amounts of aniline, N-methylaniline (XXA), and S,X-dimethylaniline (DMA) are given in Table 111. Both aniline and NMA reduce the exothermicity constant, by 20% at 1 mole yo concentration. But D3IA is relatively ineffective, reducing the value of B by only 7yoa t 4 mole yoconcentration. Beyond 1 mole yo.adding more amines to isooctane appears to affect B only slightly. The minimum oxygen requirement increases linearly with the concentration of the amines. Again the inhibitory effect of DMA is very weak; but KMA seems to be slightly more effective than aniline. Discussion The reactor assembly was designed and flow rates were chosen to assure essentially a laminar flow within the entire length of the reactor. That this was achieved is shown by the same values of the constant IC, deter-
May , 1963
THERMAL STUDIES
OF
COOL-FLAME OXIDATIOXS
TABLE I11 EFFECT OF AROMATIC AivfIms o s COOL-FLAME REACTIOXS Amine in isooctane, [0210, B x 10-12, 70 0.00
moles
0.94 1.94 3.02
moles/oc. X IO5
deg. cc.2/molesl
1.24
3.08
A. Aniline 1.55 1.80 2.12
2.47 2.50 2.49
40
1023
I
I
I
I
I
2
3
4
30
P +-
20
a
B. N-methylaniline 0.51 1.01 1.49 1.95 2.46
1.51 1,68 1.92 2.16 2.52
C. 2.03 5.0
IO
2.72 2.48 2.65 2.24 2.57
N,S-Dimethylaniline 1.34 1.40
0
Fig. 4.-Reaction
2.88 2.86
mined by the two methods used to vary the residence time. Under good laminar conditions, the hydrocarbon-oxygen mixture within a small volume of gas flowing through the reactor probably reacts independently of the rest, so that the heat generated increases the temperature of the gas only within this same volume. With the temperature of the mixture entering the reactor held conststnt, a steady-state temperature gradient is soon established along the axis of the reactor. Figure 4 shows several evamples of this gradient in terms of temperature rises a t various positions along the axis of the i'eactor at four different inlet temperatures. Plotting temperature rise at any position as a function of the inlet temperature should give the thermal profile for that position. However, the profiles used in this study were obtained by continuously increasing the inlet mixture temperature. Severtheless, because this rate was slow compared with the halflife for the estabiishment of thermal equilibrium' l 2 the resulting profiles are also considered to be for stehdy-state conditions. Figure 4 also shows $hat exothermic reactions begin a t the reactor entrance. Thus, temperature rises making up the thermal profile results from the exothermic reactions occurring throughout the entire residence time. If during this time An moles of oxygen, or hydrocarbon, are consumed causing a temperature rise of AT, the t8woquantities are related by
An.AH
=
C,AT
+ KAT'.At
(4)
where AH is the heat of the reaction per mole of reactant consumed; C,, the hleat capacity of the system; K , the heat transfer coefficient; A T ' , the difference between the temperatures of the gas and the surrounding, and A t , the residence time. In setting up equation 4, AH and C, are assunied to be independent of temperature and concentration changes and At is assumed constant. These assumptions are justified because A T , is kept below 40°, which corresponds to about 15% conversion a t an isooctane concentration of 1.2 X 10-6 mole/cc. Heat loss by radiation through the silvered m-alls is negligible and has been omitted. II AT
Because at constant residence time AT/At is proportional to A T , the thermal profile is similar in shape with curves obtained by plotting average rates of consumption against temperatures. l 3 An empirical expression for the maximum average rate is obtained from equations 3 and 5
Thus, ke, determined by the thermal method, depends not only on the reaction rate but also on the thermal properties of the system. When they are held constant, ke can be used to measure the relative reactivities of hydrocarbons with respect to their exothermic reactions. Calculatioiis show that the thermal properties change little with the kind and concentration of fuels if nitrogen is used as djluent gas and the reactions are carried out in the same reactor. A 4-fold change in isooctane to 2.215 X concentration from 0.567 X mole/cc. results in about 20% change in the thermal capacity of the gaseous mixture. The change should K ' ) since the reactor be smaller for the value of (C, also contributes to it, so that the effect on the value of ke, or of B a t constant At, could be within the experimental error. Actually, the values of B determined a t the two extreme concentrations differ by only 7%. The Minimum Oxygen Requirement.-Although the lower limit for the oxygen concentration may arise from the effect of oxygen either upon the reactions or the formation of the degenerate-branching agent, the experimental results argue in favor of the latter. These
+
(13) B. V. Aivasov and M. B. Neiman, Zhur. Fiz. Khzm., 8, 88 (1938).
G. H. NEGUERIAN
1024
results are: the value of [O2lO(a) is independent of isooctane concentration (b) is greatly affected by surface conditions; (c) is independent of residence time a t constant surface to volume ratios but varies slightly with the linear velocity of the gas; and (d) increases linearly with the concentration of inhibitors. Furthermore, exothermic reactions appear to start always at the entrance to the reactor. A plausible mechanism for the oxygen effect may result from the special role played by the reactor surface in cool-flame reactions. This surface can both initiate and break chains.14 If the onset of the branching reactions depends upon the formation of a critical concentration of the degenerate-branching agent, and if the rate of formation of this agent is determined by the balance between initiation and termination reactions, then the balance must be in favor of chain initiation from the very beginning of the laminar-flow reactor for branching reactions to occur. Otherwise no branching agent will be formed, and exothermic reactions will not occur a t any other position within the reactor. The carbonaceous material produced during aging and adsorbed on the surface may inhibit termination reactions on the surface and/or, most importantly, may enhance the initiation of chains. The former effect may result from covering the Pyrex surface, which is a relatively good chain breakerlE; the latter effect may arise from the interaction of oxygen with the carbonaceous material. Because of its mode of formation, this material should contain unsaturated bonds that can activate oxygen niolecules through the formation of complexes
s + 0 2 J_ 8-02” ICkl
i
+ S-02” * R.+ S-02H
(6)
The rate of initiation is given by
E . - - -d [ R . ] - k&i[S][021[RHl 1 -
k-i
dt
+
16,
[RH]
or, with k s > > k-i Ri
= k i [SJ [Of]
I n static systems degenerate branching is assumed to begin after a critical concentration of the branching agent is formed during the iiiductioii period. In the laminar-flow reactor, this critical concentration must be attained as soon as the reactants enter the reactor. This requirement sets a lower limit t o the rate of formation of the branching agent. If, for simplicity, this agent is an aldehyde formed by the sequence
+
R. 0 2 --+ ROz. R02. -i- R’CHO R’O. R”0. R H --+ R. R”OH RO2. +products RO. --+ products
+
then with k9
or, a t the limit
which is in agreement with the experimental results. Effect of Various Compounds.-The effects of nheptane and aromatic amines 011 the constants B and [02], are as expected. n-Heptane, being very active in the cool-flame region,12 should enhance the reactivity of isooctane. Aromatic amines, on the other hand are well known to slow down the vapor-phase oxidation of hydrocarbons. Aniline and niethylaniline, because of their reactive amino-hydrogens, can react more effectively than dimethylaniline with radicals that are essential for the initiation and acceleration of cool-flame reactions. Dimethylaniline may deactivate these radicals either by reacting through its methyl-hydrogens or by forming n-complexe~.~~ ,4mines may also increase [O2l0by favorably competing for the complex in reaction 6
S-02*
If
k13
+ ArKHR +ArKR + S-02H.
>> k6, the rate of
+ +
>> k l l
(7) (8) (9) (10) (11)
(13)
initiation becomes
Substitution of this expression in equation 12 leads, a t constant hydrocarbon concentration, to the general relation [0210
[ArXHR]
fast
RH
Vol. 67
=
const’ant
vhich confirms the linear dependence of [02], upon amine concentration. Although propionaldehyde greatly enhances the initiation of cool-flame reactions in static systems,l8 the opposite effect is obtained in the flow system. This inhibitory effect appears to result from the deactivation of the reactor surface by the highly polar propjonaldehyde. Details of this deactivation are being further studied. Conclusion The thermal method gives simultaneous information on both the activity and iiiitiation of exothermic oxidation of hydrocarbons in ternis of two empirical constants : the exothermicity constant and the minimum oxygen requirement. The exothermicity constant affords a simple and convenient means of determining the relative reactivities of hydrocarbons and the effects of inhibitors and accelerators. The existence of an oxygen limit and its independence of fuel concentration strongly suggest initiation of chains on the surface. The flow system appears to accentuate surface effects upon the initiation step and could be a suitable tool for their study. Acknowledgment.-TEe valuable help of Mr. George Hajduk in setting up the apparatus and carrying out the experiments is greatly appreciated. (16) G. H. N Chamberlain and .4. D. Walsh, tbzd.. 45, 1032 (1949),
I