IONIZATION AND ELECTRON TEMPERATURES IN CARBON

MONOXIDE AND HYDROGEN FLAMES WITH ADDED METHANE

DEREK BRADLEY AND K. J. MATTHEWS

Mechanical Engineering Department, University of Leeds, Leeds, England

Premixed gases were burned on an open tube burner at a pressure of 0.1 atmosphere. Gas temperatures were obtained from the recorded temperatures of silica coated P t - P t 10% Rh thermocouples. Double spherical electrostatic probes were employed to measure corresponding electron temperatures and positive ion concentrations. Gas velocities were estimated from previous theoretical calculations. The rate of chemical heat release and of the net rate of formation of positive ions per unit volume were derived from these values. I t was found that the zone of ion formation was narrower than that of chemical heat release and was associated with the later stages of reaction.

Electron temperatures in the hydrogen flame were close to the gas temperatures. In the carbon monoxide flame a narrow zone was observed in which the electron temperatures were very much larger than the gas temperatures. This is attributed to the acceleration of electrons to high energies by collision with electronically and vibrationaUy excited molecules of carbon dioxide.

I n t r o d u c t i o n

Much information on ionization in flame gases on flat flame burners, and often at low pressure, has been obtained using electrostatic probe techniques. Calcote and co-workers 1-a have measured ion concentrations and electron tem- peratures in hydrocarbon flames and found values for a positive ion recombination coef- ficient. Wortberg ~ has related the measured positive ion concentration in a methane-air flame at atmospheric pressure to the rate of chemical heat release, derived from gas temper- atures measured by an interferometer.

In the present work, the variation of the rate of formation of positive ions, the chemical heat- release rate, the gas temperature, and the electron temperature are presented throughout the reaction zone. Combustion of premixed gases was at a pressure of 7.6 cm Hg abs. on an open tube burner. All quantities, including gas velocity, showed both axial and radial variations. Gas temperatures were derived from the temper- atures of a silica-coated thermocouple by the application of a radiation correction. The low gas velocities at low densities could not be measured experimentally with sufficient accuracy and so theoretical values were computed from the metered flow rate of the unburned gas. Positive ion concentrations were derived from the current-voltage characteristics of a double Langmuir probe using the theory of Su and LAID_. ~

The two mixtures used, 2% CHd, 85.5% CO,

1% H~, 11.5% O~ and 2% CHd, 17.5% H2, 80.5% air each contained methane to ensure a satisfactory degree of ionization. I t was possible to relate significant differences in measured electron temperatures to the different reaction mechanisms of the two mixtures. Calcote 2 has reported electron temperatures of which some values are close to, and others far in excess of, the gas temperature. Some evidence, in conflict with that of Calcote, has been presented by Travers and Williams 6 in the form of a measured electron temperature close to the gas temperature. These latter workers pointed to possible errors in the single-probe technique and to some ad- vantages in the use of double probes. Von Engel and Cozens 7 have suggested that electrons collid- ing with the excessively excited molecules and radicals present in flame gases may be accelerated to electron temperatures in excess of the gas temperature. Evidence of such elevated electron temperatures in a coal gas-oxygen torch flame, but of electron temperatures close to gas tem- peratures in a bunsen flame, has been presented by Attard. s

One of the aims of the present work was to obtain reasonably reliable values of electron temperature and to relate these to unburned mixture composition, chemical heat-release rate, and ion formation rate.

Experimental Apparatus

Commercial grade gases were used. Oxygen and hydrogen, supplied by British Oxygen Co.

359

360 CttARGED SPECIES IN COMBUSTION PROCESSES

KHAUST GAS OUT

vIE TRAP

2/al. D. WATE R COOLED J FLAME TUBE

FOUR PROBE FLANGES /AT ONE FOOT PITCH

SPARK PLUG BOSS

WINDOW

SCREW JACK

] . I hD.WATER COOLED

BURNER TUBE

AXIAL POSITION I NDICATOR

RUBBER BELLOWS

FLAME TRAP

T GAS MIXTURE IN

Fro. 1. Diagrammatic representation of low-pressure burner.

Ltd., contained less than 0.5% impurities; carbon monoxide, supplied by I.C.][. Ltd., contained approximately 1% impurities; and methane, supplied by Air Products Ltd., con- tained less than 0.05% impurities. All gases were dried by passing through silica gel and the carbon monoxide was passed through activated charcoal to remove any iron carbonyl. The required mixture was made up in a storage tank, the gases being allowed to mix for 24 hours before being passed via a calibrated metering orifice and control valves to the vertical burner.

The mixture flowed up the 2 ft 6 in. long and 1 in. diameter water-cooled burner tube, which

was sufficiently long for the attainment of fully developed flow prior to combustion. The burner tube was a sliding fit inside the 5 ft high, 2 in. diameter, water-cooled flame tube, the joint being sealed by a rubber bellows unit as shown in Fig. 1. All tubes were of copper and the flame tube was fitted with four flanges at a pitching of 1 ft. Horizontally traversing probe units, for either the thermocouple or the electrostatic probes, could be fastened to any of these flanges. The burner tube could be moved vertically relative to the flame tube by means of a screw- jack which, in conjunction with the four flanges, enabled the flame gases to be probed at any accurately measured distance above the top of the burner tube. The combustion gases flowed out at the top of the flame tube and air was bled into the exhaust line through a needle control valve. This air served the dual purpose of cooling the exhaust gases and of providing some control over the combustion pressure. The gases flowed through a buffer tank before being finally ex- hausted to atmosphere by a 19 cu ft per min Pulsometer rotary pump. A retractable spark plug, connected to an induction coil, ignited the mixture at a pressure equal to about twice the working pressure which was 7.6 cm Hg abs.

Flame gas temperatures were measured using a Pt-Pt 10% Rh thermocouple with a hot junction formed by gas welding two wires of 0.0127 mm diameter. The fine thermocouple wire, 1 cm long, with the thermojunction central, was welded to two main support leads of 0.254 mm diameter, the junction position being ac- curately calibrated relative to the burner tube. To prevent catalytic reaction on the wire sur- face, the wire was coated with fused silica to an outer diameter of 0.071 mm by immersion in the flame of a mixture of 10% hexamethyldisiloxane and methanol on a wick burner.

The energy equation was applied to the fine wire in order to evaluate the magnitude of the error in temperature measurement arising from different causes. The larger diameter main leads were at a lower temperature than the fine wire and computations showed that the conductive cooling of the thermojunction by these leads was negligible for the length of wire used. I t was also calculated that in the central region the temper- ature of this length of wire followed that of the gas very closely. That is to say the application of a radiation correction, calculated in the usual way, to the thermojunction temperature gave the gas temperature, nothwithstanding the pressure of a temperature gradient in the gas.

The thermojunction was traversed across the flame tube with the fine wire horizontal. The emf was recorded on one axis of a Moseley Model

IONIZATION AND ELECTRON TEMPERATURES 361

7030 A Plotter and the radial position of the ]unction was accurately recorded, through the agency of a precision potentiometer linked to the traversing mechanism, on the other axis. The emf profiles across a diameter were obtained with the junction at different heights above the top of the burner tube.

The technique of the floating potential double probe was preferred to that of the single probe for the measurement of positive ion concentration and electron temperature, for the reasons given by Travers and Williams. 6 Consideration of cur- rent continuity shows that the current of electrons to one probe is limited by the value of the positive ion saturation current at the other. Because of this, any excessive drain of electrons from the plasma, which could result in a shift of plasma potential and consequent error in meas- ured electron temperature, is avoided. The double probe unit consisted of two spheres of 1.81 mm diameter formed on the ends of plati- num-10% rhodium wires of 0.178 mm diameter. Each of the two wires was electrically insulated from the surrounding gases by a fused silica tube within a 0.25 in. diameter boron nitride sheath 2.75 in. long. The end of the sheath was chamfered to an angle of 30 ~ and the end of the probe sphere was a distance of 6.3 mm beyond the apex of the sheathing. The insulated probes were mounted 0.5 in. apart on the traversing unit in the same horizontal plane and with the sphere centers at the same radial distance from the center of the flame tube. The position of the probes could be found from traversing unit cali- brations.

At any required position a range of potential differences was applied to the probes by a dry cell. The voltage and corresponding current in the floating double-probe circuit were recorded on the separate axes of the Moseley Plotter. The current measuring input to the Plotter was the output from a Hewlett-Packard Model 425A microvolt-ammeter. Great care was taken to avoid errors arising from leakage currents in the circuit. Double-probe theory shows that with positive ion saturation at one probe the second probe is closer to plasma potential. Changes in potential difference are accompanied by a change in potential of the first probe relative to plasma potential, with very little change in potential of the second probe. I t was found that insig- nificant leakage occurred provided the instru- ment was connected in the probe lead closer to plasma potential. Having taken such precau- tions, probe current-voltage characteristics were obtained which were symmetrical about the origin when conditions at the two probes were the same.

Experimental Results

Bradley and Entwistle 9 have presented theo- retically computed data on the developed axial velocities and gas temperatures for the upward flow of high-temperature air in a cooled circular tube, making allowance for the property vari- ations of air with temperature, natural convec- tion, axial momentum change, and axial con- duction. Dimensionless velocity and temperature profiles across the diameter were presented as a function of Reynolds number (Re), Grashof number (Gr), gas mixed mean temperature, and wall temperature. These data were used to derive the gas velocities of the burned gas from the metered flow rate and measured gas tempera- tures. Typical values of Re and Gr, referred to the flame tube were 30 and 380, respectively. An axial velocity profile was also calculated for the developed laminar flow in the burner tube. Flow development in the reaction zone was com- plex, but axial velocities in this region were obtained by an interpolation which blended exit- gas velocities from the burner to burned gas velocities at the same radius. This approach was most accurate close to the center line of the tubes, and the velocity u at a radial distance of 0.25 in. obtained in this way for the CO mixture is shown in Fig. 2. Throughout the present work values of transport properties of the constituent gases were taken from Svehla. 1~ Mixture vis- cosities and conductivities were found using the expressions of Wilke n and Mason and Saxena, TM

respectively. Gas temperatures (T, ~ were derived from

thermocouple temperatures by equating con- vective heat transferance to the wire with net radiative emission from the wire. The convective heat-transfer coefficient was obtained from Kramer's formula, quoted by Hinze. ~3 Reference to Spangenberg 14 showed the Knudsen number effect to be small. The total emittance of the wire at different temperatures was found theoretically from the values of monochromatic radiation absorption coefficient of silica and metal wire resistivity) 5 A typical value of emittance for the coated platinum-10% rhodium wire was 0.176 at 1550~ Some typical gas temperatures, derived in this way for the CO mixture, are shown for different heights above the top of the burner tube in Fig. 3. The gas temperatures at a radial distance of 0.25 in. are shown in Fig. 2, and, for the H2 mixture, in Fig. 4.

Positive ion number densities n+ were obtained from the positive ion saturation region of the double-probe current-voltage characteristic using the theory of Su and Lam3 This theory makes no a priori assumptions concerning sheaths around

362 CHARGED SPECIES IN COMBUSTION PROCESSES

n,

O o x T i IO

8

4

O

U

II 0.2

w

O.41

0 , 6

0 . 8

I.O

�9 K

i60

140

2Ot

OO

O

T

II /11 I ores co

~ 1 7 6 1 7 6

,i g i Jb t2 ti i~s ~'8 2 b 2 ~ O~STANCE ASOVE ~UR,E~-- M~

FIG. 2. Temperature and positive ion axial profiles at a radial distance of 0.25 in. for CO flame.

the probes, but is based on Poisson's equation and continuity equations for positive ions and electrons which include ordinary diffusion and electrostatic field mobility terms. Theo- retical double-probe characteristics were con- structed from the single probe data of Su and Lam. By matching experimental and theoretical curves in the region of positive ion saturation, a value of D+n+ was found, where D+ is the positive ion diffusion coefficient. D+ was calculated from the expression of Chapman and Cowling 16

D+ = ( 3~r/8NQ) [ (kT/27r) (m+ -~ + m-l)] 1/2,

where N is the total number density, Q the ion- neutral collision cross section, k is Boltzmann's constant, and m+ and m the molecular mass of the ion and neutral molecules, respectively. The ion was assumed to be H~O +, usually the dominant positive ion in flame gases. 2,3 Using the data of Calcote, 2 the computed value of Q for this ion was 4.2X10 --15 cm 2 for the CO mixture and 4.1 X 10 -15 cm 2 for the hydrogen mixture. Knowing the value of D+ the value of n+ was found. Some typical values of n+ at different heights above the burner for the CO flame are given in Fig. 3. Electron temperatures T_ were

IONIZATION AND ELECTRON TEMPERATURES 363

T o K

'; I o.3e"

0.38" MIXTURE

0 -855 CO o.I 15 O2 0.020 CH 4 o.o io H 2

120(

100(

80(

60(

IS

]t 4 _z

! a .

2 .

I .

0

o

o.2"/ / j,"

. l~§ ~,~ DZSTANC E ABOVE

~ BURNER GIVEN IN INCHES

"--.. \

// /

0

RADIAl PQSrflON -- MM

Fro. 3. Temperature and positive ion radial profiles for CO flame.

derived from the current-voltage characteristics by the method developed by Johnson and Malter 1~ for low pressure gases. As a check on this approach T_ was derived from the theory of Su and Lam 5 and also from the expression of At- tard, s applicable to higher pressures. Both these approaches gave values of T_ in ~ubstantial agreement with those of the Johnson and Malter theory. Ion number densities and the ratio of gas temperature to electron temperature T / T _ at a radial distance of 0.25 in. are shown for both the CO and H~ mixtures in Figs. 2 and 4.

Discuss ion

Assuming that radial flow velocities, radiation from the gases and net enthalpy fluxes arising from diffusion velocities may be neglected, the

energy equation is:

Ou E ~ih, Ok(OT/Ox) r- ~ Ohr(OT/Or) q -- Ox Ox Or

where q is the rate of chemical heat release per unit volume and ~ indicates a summation over all chemical species present. Values of q at dif- ferent positions were found for the CO flame by graphical differentiation based on gas temper- ature and velocities fields. These values are shown in Fig. 5 for different heights above the top of the burner tube at a radial distance of 0.25 in. Chemical reaction extended, with diminishing intensity, to a distance of 18 ram. Previous work 18 has shown that atomic oxygen and excited molecules of carbon dioxide, with accompanying visible emission, are present in the afterburning region.

364 CHARGED SPECIES IN COMBUSTION PROCESSES

n +

O

% X

8

6

4

2

0

FIG. 4. Temperature and positive ion

I m I"

o,iO1+oi++i+ .+...

"~ I " ' . " : , " o ~ , i g ~ ~b ~ ~4 I~ li~ 2o 2a 24 2i~

DISTANCE ABOVE. BURNER--MM

axial profiles at a radial distance of 0.25 in. for H~ flame.

From a knowledge of the distribution of positive ion number density n+, gas temperature, velocity, and electron temperature, it was possible to derive the net rate of production, or of decay, of positive ions per unit volume ~/+, at different positions. The continuity equation for positive ions, neglecting any small radial flow velocity but allowing for axial and radial ambipolar diffusion, is

On+u 0 (Do~T) (On.+T/Ox) n+ = Ox Ox

_ r- 10 ( rDJT) (On+T/Or) Or

Do, the positive ion ambipolar diffusion coef-

ficient, is related to the ordinary diffusion coef- ficient D+, by 1~

D_/D+ + T _ / T = (1 + T_/T)D_/Do.

As

D_/D+ >> T_/T,

with sufficient accuracy, we may write that

D, = D+(1 + T_/T)

Values of 4+ were found for the CO flame by graphical differentiation, and these results are shown in Fig. 5 for different heights above the

IONIZATION AND ELECTRON TEMPERATURES 365

IB .

7 oK |

~ U S 16(30 I 0

4 114001 0 . 2

~!j lmo.j 0 . 4

2 1 1 0 0 0 1 0 . 6

I I 1 0 - 8

LUMINOUS ZONE

M IXTURE

0 - 8 5 5 C O 0415 0 2 o~2o c ~ o . 0 1 0 H 2

I

\ n § \

OI I 1 . 0

t ,o 2 ~ / 6 /a 11o~2//r4 i m 22

- 4 ,, o ;

DISTANCE ABOVE BURNER- MM

Fro. 5. Derived values of chemical heat release and net rate of positive i on

production for CO flame.

top of the burner table at a radial distance of 0.25 in.

Errors in these results may arise from a number of causes. The electrostatic probe theory made no allowance for a temperature gradient around the probes or for a finite gas velocity. I t is un- likely that excessive error arises on these counts but greater errors occur due to the highly non- uniform condition of the gas in the reaction zone. The electrostatic probe theory assumed the probes to be surrounded by a uniform plasma, whereas the gradients of n+ and T_ were some- times very large relative to the probe size. The computational process of double graphical dif-

ferentiation further adds to the error in the derivation of q and ~/+. Nothwithstanding these limitations certain points of interest emerge. Figure 5 shows that the zone of net production of positive ions is narrower than is suggested by the values of n+ in Fig. 2. Ions may be present at a given position because they diffuse there, not because they are produced there. Figure 5 also shows that ion production is associated with the later stages of chemical reaction.

In the absence of any production of positive ions, the rate of ion decay is often expressed by

_ ~ + = ~ _ 2

366 CHARGED SPECIES IN COMBUSTION PROCESSES

where a is an ion recombination coefficient. Values of a found from this equation for dif- ferent positions are shown at appropriate points on the curve of ~+ in Fig. 5. The equation is not applicable in the region where ions are being produced. If n+v is the rate of positive ion pro- duction per unit volume then in this region

n+p = r + an+ ~

In regions where the assumption, n+v = 0, was justifiable a was found to be 1.3 X 10 -~ cm 3 sec - i . If ~ is assumed constant for all regions then n+v may be obtained from ~i+ and n+. Values of a, in cm a sec -1, which have been obtained by other workers are 2.4 :t: 0.4 X 10 -~ by Calcote, Kurzius and Miller, 3 2.2 :t: 1.0 X 10 -~ by Green and Sugden 2~ and Semenov and Sokolik, ~ and 1.1 -4- 0.1 X 10 -~ by Wortberg. 4 Values were almost constant throughout a particular flame and showed no discernable pattern of change with temperature and pressure.

Ionization in the carbon monoxide and hydro- gen mixtures may be compared by reference to Figs. 2 and 4. The velocity of the unburned gas was the same for both mixtures, but the carbon monoxide flame attained a higher temperature. The maximum value of n+ was higher for the CO flame. For both mixtures, the ionization origi- nates from the presence of 2% methane and is based largely on the generally accep~d mechan- ism 2~

CH + O --~ CHO + -4- e- (1)

CHO + + H~O -* CO + H.~O + (2)

e- -{- HaO + --* HeO ~- H, (3)

the last reaction being one of positive ion recom- bination. A noticeable difference between the two mixtures lies in the existence of a narrow zone of very high electron temperatures, with a peak value of 30,000~ in the CO flame. In the H2 flame, electron temperatures were close to gas temperatures.

An explanation of this difference between the two flames can be advanced, in terms of dif- ferences in the chemical reactions. Suggested reaction schemes for the oxidation of carbon monoxide have involved electronically and vibrationally excited molecules of carbon dioxide, with associated chemiluminescent radiationY ,24 Excited molecules have not been involved in any comparable way in reaction schemes for the oxidation of hydrogen? 3 Clyne and Thrush 24 have proposed a scheme for carbon monoxide oxidation in which carbon monoxide and atomic

oxygen react in a termolecular reaction to give carbon dioxide in an electronically excited triplet state. This is followed by a radiationless transi- tion to an excited singlet state, which decays by emission to a vibrationally excited linear ground electronic state. I t was noticed in the present work that ionization increased with the propor- tion of methane in the CO mixture. However, both the visible radiation and atomic oxygen concentration, indicated by catalytic heating of uncoated thermocouple wires, TM decreased show- ing an increased reaction of 0 atoms by Re- action (1).

I t is unlikely that the ejection of an electron with excess kinetic energy in Reaction (1) can be the explanation of the high values of T_ in the CO flame, when the same reaction does not result in an increase of T_ above the gas temperature in the H2 flame. A better explanation of elevated electron temperatures is along the lines proposed by yon Engel and Cozens. ~ In the CO flame, collisions of chemically formed electrons with electronically and vibrationally excited molecules of carbon dioxide could accelerate the electrons to the velocities of the elevated electron temper- atures. There would be no counterpart of this in the H~ flame, with its lower concentration of carbon dioxide.

Von Engel and Cozens 25 have further sug- gested that small concentrations of electrons, arising from thermal ionization, could gain energy by collision with excited molecules to such an extent that electron concentrations could be further raised as a result of ionizing electron collisions. The view that such collisions are a predominant cause of the high ionization in flames is not supported by the conditions in- vestigated in the present work. At the peak ob- served electron temperature of 30,000~ the average kinetic energy of an electron corresponds to 3.9 eV. Using the data presented by Clyne and Thrush, 2a an electronically excited molecule of carbon dioxide may possess excitation energy equal to approximately 5.6 eV. The energy available for ionization in a collision between a high-temperature electron and an excited mole- cule of carbon dioxide would be approximately 9.5 eV. According to Cobine, 2s the ionization potential of a normal molecule of carbon dioxide is 14 eV. I t would therefore seem, even when allowance is made for the presence of excited species, that electron ionizing collisions can only make a small contribution to the observed ionization. There is even less probabili ty of collisional ionization in the hydrogen combustion. Figure 5 shows that the region of maximum positive ion production precedes the region of maximum electron temperature. This gives some support to the view that h igh electron temper-

IONIZATION AND ELECTRON TEMPERATURES 367

atures are a result of chemi-ionization rather than a cause of collisional ionization.

Nomenclature

D diffusion coefficient Gr Grashof number h specific enthalpy per mole k Boltzmann's constant m molecular mass

mole number density n+ positive ion number density fi+ net rate of production of positive ions per

unit volume n+p rate of production of positive ions per unit

volume N total number density q rate of chemical heat release per unit

volume Q ion-neutral collision cross section r radial distance Re Reynolds' number T gas temperature (~ Tc coated-wire temperature (~ T_ electron temperature (~ u gas velocity x axial distance a ion recombination coefficient k thermal conductivity

Subscripts

- - electron -[- positive ion a ambipolar i chemical species "i".

ACKNOWLEDGMENT

The authors thank the Science Research Council for a Research Studentship Grant to K.J.M.

REFERENCES

1. CALCOTE, H. F.: Eighth Symposium (Inter- national) on Combustion, p. 184, Williams and Wilkins, 1962.

2. CALCOTE, H. F.: Ninth Symposium (Interna- tional) on Combustion, p. 622, Academic Press, 1963.

3. CALCOTE, H. F., KURZIUS, S. C., AND MILLER, W. J.: Tenth Symposium (International) on Combustion, p. 605, The Combustion Institute, 1965.

4. WORTBERG, G.: Tenth Symposium (Interna- tional) on Combustion, p. 651, The Combustion Institute, 1965.

5. Su, C. H. AND LAM, S. H. : Phys. Fluids 6, 1479 (1963).

6. TRAV~RS, B. E. L. AND WILLIAMS, H.: Tenth Symposium (International) on Combustion, p. 657, The Combustion Institute, 1965.

7. YON ENGEL, A. AND COZENS, J. R.: Nature 202, 480 (1964).

8. ATTARD, M. C. : International Symposium on Magnetohydrodynamic Electrical Power Gene- ration 1964, Vol. 1, p. 21, European Nuclear Energy Agency and Organization for Economic Co-operation and Development.

9. BRADLEY, D. AND ENTWISTLE, A. G.: Intern. J. Heat Mass Transfer 8, 621 (1965).

10. SVEHLA, R. A.: Estimated Viscosities and Thermal Conductivities of Gases at High Tem- peratures, NASA TR R-132, 1962.

11. WILKE, C. R.: J. Chem. Phys. 18, 517 (1950). 12. MASON, E. A. AND SAXENA, S. C.: Phys. Fluids

1,361 (1958). 13. HINZE, J. O.: Turbulence, p. 76, McGraw-Hill,

1959. 14. SPANGENBERG, W. G.: Heat Loss Characteristics

of Hot Wire Anemometers at Various Densities in Transonic and Supersonic Flow, NACA TN 3381, 1955.

15. BRADLEY, D. AND ENTWISTLE, A. C.: Brit. J. Appl. Phys. 17, 1155 (1966).

16. CHAPMAN, S. AND COWLING, T. G.: The Mathe- matical Theory of Non-Uniform Gases, p. 245, Cambridge University Press, 1960.

17. JOHNSON, E. 0. AND MALTER, L.: Phys. Rev. 80, 58 (1950).

18. LEAH, A. S. ROUNTHWAITE, C., AND BRADLEY, D. : Phil. Mag. xli, 478 (1950).

19. SAMARAS, D. G. : Theory of Ion Flow Dynamics, p. 332, Prentice-Hall, 1962.

20. GREEN, J. A. AND SUGDEN, T. M.: Ninth Sym- posium (International) on Combustion, p. 607, Academic Press, 1963.

21. SEMENOV, E. S. AND SOKOLIK, A. S. : Zh. Tektm. Fiz. 32, 1074 (1962).

22. SUGDEN, W. M.: Tenth Symposium (Interna- tional) on Combustion, p. 539, The Combustion Institute, 1965.

23. GAYDON, A. G.: The Spectroscopy of Flames, Chapman and Hall, 1957.

24. CLYNE, M. A. A. ANn THRUSH, B. A. : Ninth Symposium (International) on Combustion, p. 177, Academic Press, 1963.

25. VON ENGEL, A. AND COZENS, J. R.: Proc. Phys. Soc. (London) 82, 85 (1963).

26. COBINE, J. D.: Gaseous Conductors, p. 83, Dover, 1958.

368 CHARGED SPECIES IN COMBUSTION PROCESSES

COMMENTS

Prof. S. H. Bauer (Cornell University): The very high electron temperatures which the probe meas- urements indicate for the CO/H2/CH, flames are surprising indeed. The proposal that the electrons derive their excessive kinetic energy by energy transfer from vibrationally excited CO~ requires substantiation. Several tests can be made. There are curreut measurements of collision cross reac- tions for energy losses suffered by slow electrons (10 eV range) with small molecules. Hence, esti- mates of efficiencies for the reverse process can be made. Also, since the flame gases contain unburned CO, the reaction

CO + e- (fast) ~ CO(') + e- (slow)

provides coupling between probe measurements of the electron temperature and the vibrational excitation of the COCV). Relative intensities of the various CO bands may serve as the desired check.

May I suggest that the difference observed be- tween flames with and without CO is due to their different luminosities? There may be considerable numbers of photoelectrons ejected from the probe, particularly when the luminosity is high, and these distort their distribution in the vicinity of the probe.

Dr. D. Bradley and Dr. K. J. Matthews: As yet, we have been unable to put the proposed energy transfer from excited carbon dioxide molecules to electrons on a reliable quanLitative basis, due to incomplete data. In addition to this effect, we agree that there may be some energy transfer between electrons and vibrationally excited mole- cules of carbon monoxide.

The suggestion of photoelectron emission from probes due to radiation, visible or otherwise, is a novel one, and we believe it warrants further investigation.