The role of iron additives in sooting premixed flames

Twenty-Fourth Symposium (International) on Combustion/The Combustion Institute, 1992/pp. 1007-1014

THE

ROLE

OF IRON

ADDITIVES

IN SOOTING

PREMIXED

FLAMES

DAVID W. HAHN nNo TRYFON T. CHARALAMPOPOULOS Mechanical Engineering Department Louisiana State University Baton Rouge, Louisiana 70803 USA

In-situ light scattering measurements are combined with chemical analysis of sampled soot particles to elucidate the mechanisms through which iron addition affects soot formation and growth processes in fuel-rich premixed propane/oxygen flames. Soot particle optical inhomogeneity resulting from the addition of iron is accounted for in the light scattering analysis using an effective refractive index model. The influence of iron addition on soot particle diameters, number densities, volume fractions, surface area, and specific growth rates is investigated. In addition, X-ray photoelectron spectroscopy was used to determine the chemical state of iron species throughout the flame. The analysis revealed that the iron oxide Fe203 is the dominant species within the soot particles, corresponding to residence times from 10 to 32 milliseconds. The role of iron additives as soot suppressants is discussed, and mechanisms are proposed to explain the soot suppressing behavior associated with iron in practical combustors.

Introduction The formation of particulate soot in flames can be both beneficial and detrimental, depending on the nature of the combustor. For example, in boilers and furnaces the presence of soot is favored because it promotes high heat transfer rates. However, the presence of soot in systems such as diesel engines is highly undesirable because it constitutes a considerable source of particulate pollution. The future use of heavily sooting alternative fuels can only compound the problems associated with unwanted soot emissions. Thus, better understanding of the soot formation and control processes is of primary importance. One option available for the control of flame generated soot is the use of metal based fuel additives, which provide an economically sound and relatively simple alternative for emission control. Numerous 15 68 studiesand several reviewshave been prepared concerning the effects of metal additives on soot suppression and control. Finfer6 conducted a review of fuel oil additives for controlling emissions, concentrating on the few metallic additives that were commercially available. Finfer concluded that "improved additives are required," and appears to be the first investigator to state that the toxicity of additives should be addressed. Of the metals reported in the literature, iron, manganese and barium form the basis of metal compounds which are often cited to be highly effective in the reduction of soot emissions. While the effects of metal containing fuel addi-

tives on soot emissions have been well documented, 1-8 study of the mechanisms through which they function continues to be an area of active research. 9 The formation of metal oxides, particularly with iron addition, and their subsequent oxidation of solid carbon to CO and CO2 has been proposed as a soot reduction mechanism. 1A~ Alternatively, metal oxides of the main transition metals iron and manganese can catalyze the oxidation of soot by oxygen. 11A2 In contrast, catalysis of carbon deposition is associated with metallic iron. 13 Ionic activity with the addition of iron to flames has been reported to be insignificant,2 thereby eliminating a possible ionic mechanism of soot reduction. In summary, iron is one of the metal additives most considered for emission control and is among the metals which pose the least threat to the environment. Nonetheless, the function of iron as a soot suppressant has yet to be fully understood. The objective of the present study is to combine in-situ light scattering measurements with chemical analysis of sampled soot particles to elucidate the role of iron additives in sooting flames.

Experimental Facilities and Measurements Premixed propane/oxygen flames with fuel equivalence ratios 2.4 and 2.5, diluted 48 percent by volume with nitrogen, were used in the present study. The laminar flames were supported on a 57.2 mm diameter ceramic honeycomb burner (400 cells per square inch). The source of iron for the present

1007

1008

SOOT

study was iron4Pentacarbonyl, Fe(CO)~, which has been reported to have a half-life for stripping the carbonyl groups of less than 1 ns at flame temperatures. The additive was introduced to the burner by diverting a percentage of the diluent nitrogen flow through a constant temperature cylinder which contained the liquid iron pentacarbonyl, and then remixing the nitrogen and iron pentacarbonyl vapor with the bulk nitrogen supply. The system was calibrated and allows the iron mass flow rates to be varied from 0.06 to 0.37 percent by weight of fuel, reproducible to within 3 percent, while maintaining an overall constant nitrogen flow rate. The axial temperature profiles of the flames were recorded using platinum vs. platinum-10% rhodium thermocouples in both unseeded and iron seeded flames. The maximum temperature difference between the flames was less than 2.5 percent, which is within the limits of experimental error, and ensures that no thermal effects are introduced by the addition of iron. In order to fully assess the effects of iron addition to the flames, both soot particle sampling and in-situ light scattering measurements were utilized. A vertically polarized, 3 watt argon-ion laser was used for light scattering and extinction measurements at the wavelength 488 nm. More details about the experimental facilities and measurement techniques have been presented in previous work. 1~ A water cooled, stainless steel probe was used to collect soot from various axial locations in the flame. The soot samples were collected on 10 mm diameter glass filters for 5 minutes, and stored under argon prior to chemical analysis.

Experimental Analysis and Results Light scattering measurements were performed in premixed flames of fuel equivalence ratios equal to 2.4 and 2.5, designated as Flames 1 and 2, respectively. Measurements were also carried out in identical flames seeded with 0.16 and 0.32 percent by weight iron to fuel, designated as Rates i and 2. Soot samples were collected from Flame 1 seeded at Rate 2 at axial locations 4, 8, 12 and 16 mm above the burner surface, which correspond to residence times of 10.7, 17.4, 24.5 and 32.0 msec, respectively.

Chemical Characterization of Particles: The collected iron seeded soot samples were analyzed using X-ray photoelectron spectroscopy (XPS). The XPS spectra of the samples all exhibited nearly identical peaks, which are summarized in Table I. The absolute values and spacing of the XPS peaks are in agreement with the values reported 16 for the

TABLE ! XPS results for Flame 1, seeded with iron pentacarbonyl 0.32 percent by weight iron to fuel (Rate 2)

Height (mm)

Peak 1 (eV)

Peak 2 (eV)

Dominant iron species

4 8 12 16

712.1 712.2 712.0 712.2

725.5 725.4 725.1 725.6

Fe2Oa Fe2Oa Fe2Oa Fe2Oa

iron oxide Fe2Oa. No significant peaks were identified corresponding to metallic iron or any other iron species. The XPS analysis indicates the existence of iron as Fe2Oa at all four locations within the flame, corresponding to flame residence times from 10.7 to 32.0 msec.

Light Scattering and Extinction: The differential scattering and extinction coefficients were measured as a function of height above the burner surface for Flame 1 both unseeded and seeded at Rates 1 and 2, and for Flame 2 unseeded and seeded at Rate 2. In the present study, the cross sections are calculated using the full Mie solution for spherical particles, and the size distribution function utilized is the zeroth-order logarithmic distribution (ZOLD). 17 The values rg = 1.5 - 0.4i for the refractive index and or0 = 0.18, the width and skewness of the ZOLD, are utilized in the present analysis, and were determined specifically for the flames in the present study using a technique developed recently, is Data analysis yields the soot particle modal radius rm and the particle number density N. Knowledge of the soot size and number density enables calculation of the soot volume fraction and total surface area by integrating over all particle radii using the ZOLD function.

Effective Refractive Indices: Before analyzing the data obtained for iron seeded flames, the effects of particle optical inhomogeneity resulting from the inclusion of iron species within the soot particles should be addressed. Bitrievi1~ hypothesized that iron added to premixed flames will nucleate in advance of soot inception, and subsequently provide sites on which soot inception may occur. Depth profiling of collected soot particles verified the inclusion of iron within the soot partides. Further evidence in support of iron inclusion has been reported recently, 9 in a study of soot collected from burning ferrocene seeded crude oil.

ROLE OF IRON ADDITIVES IN FLAMES Since iron inclusion within the soot particles exists, a suitable model accounting for optical inhomogeneity is required. In the present flame, it is expected that simultaneous particle coagulation and soot surface growth will lead to a random assemblage of iron oxide dispersed in a soot matrix. The optical characteristics of such a system are described by the Maxwell-Garnett model, 19 which combines the dielectric function E of a material dispersed in a matrix with dielectric function em, to yield the effective dielectric function E~-c of the mixture, through the relation + 2~m + 2X(e eM-c

growth rate (cm3 soot/cm z soot sec) may be determined from the rate of change of the soot volume fraction, and the total soot surface area per unit volume. 21 The particle residence times were calculated from the soot particle velocity profile, which was measured using an optical technique. 2~'z3 The specific surface growth rates are presented in Fig. 3 for all flames and iron seeding rates.

Discussion

Soot Particle Parameters:

~m)

= ~,.

1009

.

(1)

E + 2Era - - X(E -- era)

The parameter x represents the volume percentage of the dispersed material, taken to be the identified iron species, FeeO3. The soot matrix refractive index is taken to be that of the unseeded flames. The refractive index of FezO3 was measured using an ellipsometry techniquez~ on reagent grade compressed pellets, and found to be n~ = 1.64 - 0.3i at 488 nm. The percentage of iron oxide within the soot particles was determined by first calculating the volume fraction of iron oxide (cm3 Fe2Oa/em3 gas) at each position, based on an iron mass balance. The volume percentage of Fe203 was then calculated from the volume fraction of Fe203 divided by the total soot volume fraction. In the Rate 1 seeded flame, the maximum percentage of Fe2Oa is 19.1 percent, which decays within 6 mm to an average value of 8.0 percent. In the Rate 2 seeded flame, the maximum percentage of Fe203 is 33.9 percent, which decreases within 6 mm to an average value 14.8 percent. Utilizing the volume percentages of Fe203 and the refractive indices of soot and Fe2Oa, the effective refractive indices for the seeded flames were calculated at each position. The effective index value rfi = 1.55 - 0.37i represents the greatest change over the index of pure soot. The average increase in n for flames seeded at Rates 1 and 2 is +1.3 percent, while the average decrease in k is -3.3 percent. Utilizing the constant index in the unseeded flames, and the effective indices in the seeded flames, the soot particle radii, number densities, volume fractions, and surface areas were inferred. Representative values are presented in Table II for Flame 1, unseeded and seeded at Rate 2. The soot volume fraction and surface area profiles corresponding to all flames and seeding rates are presented in Figures 1 and 2.

Specific Surface Growth Rates: The effects of iron addition on soot growth processes were also assessed. The soot specific surface

As may be seen in Table II and Fig. 1, the soot yields increase in the iron seeded flames. This trend is consistent with the results of previous investigations in similar flame systems, 1~ but differs from the soot suppressing characteristics of iron observed in practical combustors. 24 Laboratory premixed flames provide a planar, one-dimensional flame which allows precise evaluation of soot formation and growth processes. However, commercial eombustors possess a final oxygen-rich soot oxidation or "burnout" region which subsequently destroys much of the soot mass formed. As will be discussed below, the results of the present study may be used to assess the soot suppressing mechanisms of iron in practical flame systems. The soot particle diameters are least affected by iron addition, increasing by an average of +1.9 percent for all seeding rates, whereas the particle number densities increase with iron addition by an average of + 16.4 percent for all flames. It is noted that the changes are slightly greater in the leaner flame. The combined effect is an average increase in soot volume fractions of +22.9 percent in the seeded flames. The increase in volume fraction is greater in the leaner flames, and is enhanced with increasing iron addition rates. Iron addition also results in an increase in soot surface area for all flames and iron addition rates studied, with an average increase of +20.5 percent. While the total inferred soot yields are significantly enhanced with iron addition, the soot surface growth rates are clearly less affected. In the region between 3 and 12 mm above the burner surface, which accounts for nearly 90 percent of the total soot formed, the growth rates for the unseeded and seeded flames differ by an average of 2.9 percent for all flames. It is noted that the growth rates in the seeded flames are markedly less than in the unseeded flames at 3 mm only, which is in agreement with initial decreases in growth rates in seeded flames reported by Ritrievi and co-workers. 10 The overall increase in soot loading with the addition of iron may be explained by the simultaneous increase in soot surface area at all locations. The initial increase in soot surface area may be at-

1010

SOOT

TABLE II Experimental results for Flame 1, unseeded and seeded with iron pentacarbonyl 0.32 percent by weight iron to fuel (Rate 2) Height

Particle Diameter

(ram)

(nm) Unseeded

Rate 2

Unseeded

Rate 2

Unseeded

Rate 2

Unseeded

Rate 2

8.83 13.9 22.8 29.8 35.0 38.9 42.9 45.7 48.2

9.82 14.6 23.7 30.9 35.9 39.7 43.5 45.3 48.0

184.0 70.5 22.4 12.5 8.78 7.08 5.38 4.79 4.31

191.0 78.3 25.8 14.4 10.4 8.43 6.59 6.22 5.30

O.085 0.125 0.178 0.221 0.252 0.278 0.284 0.306 0.323

0.121 0.163 0.229 0.284 0.320 0.351 0.363 0.386 0.391

0.513 0.484 0.418 0.398 0.386 0.383 0.355 0.358 0.359

0.658 0.598 0.517 0.492 0.478 0.474 0.447 0.457 0.436

3 4 6 8 10 12 14 16 18

1.0

~

,

t,')

E 0.8

,

,

i

,

,

i

0

Flame1,

0

Fia~ 1, Fe RaIB 1

A

Flare 1, Fe Rale2

X

F'l~w 2, Un~md~l

+

F I ~ 2, Fe Paw 2

,

i

,

,

,

~

i

,

,

,

1.2

,

E 0

+

4`

o X

X

x X

r

+

O

0.8

I

'

'

'

'

I

'

,

0

0

F~

0

i

I

I

J

5

I

I

I

I

I

'

'

'

Flame 1, Fe Rate 2

X

Flame 2, ~ s ~ o d

X

4`

FI~

2, F~ P~te 2

~

+

4-

+

+

X

X

X

•

A

+

+

X

X

0

0

0 :3

I

'

~, Unst~ldsd

4-

O

0.4

0

0

0 0

u)

> 0.0

I

Flame 1, Fe Ra~ 1

o o

0

0

o

E

O O

O


~

i

4` X

0.6

X

u. 02 E

,

6

E

X

x

+

+ x

,

1.0

§

'~ 0.4 C: 9

,

03

+

E 0,6

i

,@ +

O O u)

Surface Area (cm2/cm 3 gas)

Volume Fraction (10-~cma/cma gas)

Number Density (109/cm 3 gas)

t

I

I

i

I

10 15 HeightAboveBurner(mm)

J

I

I

I

0.2

20

FIG. 1. Inferred soot particle volume fraction profiles for unseeded and iron pentacarbonyl seeded flames.

I

0

I

I

I

I

I

I

I

I

[

i

,

,

t

I

5 10 15 HeightAboveBurner(mm)

,

,

I

I

20

FIG. 2. Inferred soot particle surface area profiles for unseeded and iron pentacarbonyl seeded flames.

Chemical Analysis: tributed to the additional surface area available for deposition of soot, provided by the iron oxide nuclei. Since more soot surface area is available from the beginning for surface growth in the seeded flames, more soot will be produced even though the growth rates are approximately equal to those in the unseeded flames. This observation is in agreement with results reported by Harris and Weiner, 21 in which soot yields for a variety of flame conditions correlated well with growth rates and surface area profiles.

X-ray photoelectron spectroscopy was used to identify the iron species present within the soot particle samples. Rigorous particle sampling and analysis procedures were followed to ensure that the identified iron species were representative of the species present within the particles in the actual flames. The XPS analysis indicated the existence of Fe203 as the only significant iron species in samples from throughout the flame. Ritrievi and coworkers ]~ used Mossbauer spectroscopy to identify the chemical species of iron within sampled soot

ROLE OF IRON ADDITIVES IN FLAMES 75.0

[

t

i

=

I

i

t

t

t

[

t

i

t

t

[

i

~

I

0

F I ~ 1, Unseeded

0

Flame 1, Fe Rate 1

The results of the present study suggest that iron may affect soot emissions primarily in the soot burnout zone as follows. The oxide FezO3 is known to catalyze]2'25 the oxidation of carbon by oxygen given by the reactions

i

o,I

E o

60.0

tO

X ~

E 0

0

+ 0

o. 45.0

+ X

,,,

~

X

Flame 2, U.seeded

+

Fla~ 2, Fe Raffi 2

+

0

), Fe L~te 2

1 Csoli d +

|

0

n-

30.0

o

•

0

+ x

0

~ ,

0.0

0

,

,,

I

5

,

2

C~olid + 02--> C02.

+

15.0

,

,

,

I

,

,

(3)

- 0 2 -----> C O

§ 6

0 |

1011

,

,

~ I

10 15 HeightAboveBurner(mm)

,

0

•

,~,

20

FIG. 3. Inferred soot specific surface growth rate profiles for unseeded and iron pentacarbonyl seeded flames. particles and reported the existence of metallic iron only. However, the samples were collected on a plate mounted at the end of the postflame zone, and therefore may not be indicative of the iron species present in soot particles within the actual flame.

(4)

Since the FezO3 is present within the soot particles, where the oxidation of carbon is significant, 26 the seeded soot particles should be oxidized much faster through catalysis by Fe203. This mechanism may explain the overall soot reductions which are associated with iron addition in practical combustors, and accounts for the increased soot yields in the premixed flames of the present study which contain no soot oxidation zone. Soot Oxidation Model: A soot oxidation model was developed to predict the final soot yields based on the experimental results, and to demonstrate the proposed soot reduction mechanisms of the present study. The change in volume-fraction as a function of time may be related to the rate of soot oxidation through the relation

Mechanisms Affecting Soot Emissions: It has been proposed l'l~ that reactions such as

dfo dt

FeO + Csolia--->Fe + CO,

Of~ dN -

ON dt

Ofo Odrm +

.

.

.

.

Or,. dt

.

&A .

f5)

p

(2)

may be important in the reduction of soot emissions by metal additives. However, the insignificant effect of iron addition on soot surface growth rates coupled with the chemical state of iron remaining as Fe203 throughout the flame, suggest that direct oxidation of soot by iron oxides is unimportant in the present flames. Furthermore, in practical combustors the addition of iron has been reported to yield reductions in soot emissions of up to 90 percent. 24 However, with additive concentrations typically less than I percent by weight, there is insufficient metallic oxide present to yield significant effects throughloReaction (2), which has been previously noted. In the present study, the amount of Fe203 present in Flame 1 is approximately one order of magnitude less than that required to yield a 90 percent reduction in soot emissions via Reaction (2), assuming the reaction is 100 percent complete and that all the added iron is available. Alternatively, it has been suggested that metal additives may accelerate the rate of soot oxidation. 1

The derivatives aft~ON and Of,/Orm are calculated from the volume fraction expression, 4

[15

\

3

(6)

f~ = -3 zrN e x p ~ - O{o)r ....

and dN/dt is calculated from a linear fit of the soot number density profile, N(t) = No - at, for residence times greater than 20 msec. The parameter p is the mass density of soot, A the soot surface area per unit volume, and d~ is the soot mass oxidation rate (g soot/cm 2 soot sec). Equation (5) reduces to the differential equation dr,. -

dt

arm

o) exp[--3.5~]

3(No - at)

p

-

-

,

(7)

in which a is the magnitude of the slope of the number density profile, and the initial condition is

1012

SOOT

.o

Conclusions

100

o

-

-

Unseeded,

No Ca1,~/sis

75 ~ ~10"1

/

/

.

so ~ o9. :~ 25 ~

0 t-

\

/

,~o 10-2

/

\

"

0

>10-3 0.0

~"

0

/ /

~ ~

, , , I , , , , I

1.0

. . . .

I

. . . .

I

2.0 3.0 4.0 ResidenceTime (msec)

,

,

,

-25 5.0

FIe. 4. Simulated soot oxidation profiles for Flame 1, unseeded and seeded with iron pentacarbonyl 0.32 percent by weight iron to fuel (T = 1650 K, eoz = 0.5 atm.).

rm(t = O) = ro. The solution of the above equation is

rm(t) = (1 - a ~ - ' 3 [ 3Now o3 ~oo t) [ro - 4ap exp(-3.5O~o)J 3N~176 exp[-3.5o~o] ( 1 - ~ o t ) .

4ap

(8)

Using the final soot particle diameter and number density values at 18 mm above the burner surface as initial conditions, the soot radius, number density and volume fractions were calculated as a function of time, thereby simulating soot burnout. The soot oxidation rate d~ was estimated from the Nagel and Strickland-Constable correlations,z7 For the iron seeded flames, Fe203 catalysis of carbon oxidation was assessed by assuming a catalysis rate of 20 percent. While no quantitative information for Fe203 catalysis rates is available, catalysis by manganese, also a main transition metal with fuel additive characteristics similar to iron, of carbon oxidation has been reported to be on the order of 20 percent. 11 For the calculations, the oxygen partial pressure is 0,5 atm. and the temperature is 1650 K. The calculated soot volume fraction profiles and the corresponding soot reduction percentages, as determined by the difference between the flames, are presented in Fig. 4, for the unseeded and Rate 2 seeded flames. The simulated effects of Fe203 catalysis on soot oxidation are readily apparent, with a 75 percent reduction in soot volume realized in 5.0 msec for 20 percent catalysis.

The effects of iron addition in hydrocarbon flames were investigated using in-situ light scattering techniques and X-ray photoelectron spectroscopy (XPS). Tile soot particle optical inhomogeneity resulting from inclusion of iron species was accounted for using the Maxwell-Garnett model. The effects of iron addition on soot particle inception and growth proeesses, and the evolution of iron species throughout the seeded flames were evaluated. The findings of the present study are summarized as follows: 1) Iron will nucleate as iron oxide particles prior to soot inception, and will provide sites for soot inception and deposition. Subsequently, the iron oxide nuclei are included within the soot particles. 2) XPS analysis revealed that the iron oxide Fe203 is the only significant iron species present within the sampled soot particles, corresponding to the range of particle residence times considered. 3) Iron addition had no significant effect on soot surface growth rates, with the exception of a decrease in growth rates in the first few milliseconds following soot inception. 4) The existence of the iron in the form of Fe.~O3 throughout the flame coupled with the insignificant changes in soot growth rates, suggests that direct soot oxidation by Fe203 may not be a significant mechanism in the flame studied. 5) Increased soot volume fractions with iron addition in premixed flames may be attributed to increased initial soot surface area resulting from the available iron oxide nuclei. 6) Overall reductions of soot emissions in practical combustors with iron addition may be realized through FezO3 catalysis of carbon oxidation to CO and COz in the soot burnout zone,

Acknowledgements This work was supported in part by the National Science Foundation through grant CBT 8820480. The authors would like to acknowledge Dean Thomas Lester and Professor K. Saito of the University of Kentucky, and Dr. H. Woo of Exxon Research and Development Laboratories in Baton Rouge, Louisiana, for useful discussions and assistance with the XPS analysis,

REFERENCES 1. COTTON, D. n., FRISWELL, N. J. AND JENKINS, D. R.: Combust. Flame 17, 87 (1971).

ROLE OF IRON ADDITIVES IN FLAMES 2. BULEWICZ, E. M., EVANS, D. G. AND PADLEY, P. J.: Fifteenth Symposium (International) on Combustion, p. 1461, The Combustion Institute, 1975. 3. FEUGIER, A.: Adv. Chem. Series, Evap. Cornbust. Fuels 166, 178 (1978). 4. HAYNES, B. S., JANDER, H. AND WAGNER, H. GG. : Seventeenth Symposium (International) on Combustion, p. 1365, The Combustion Institute, 1979. 5. BONCZYK, P. A.: Combust. Sci. Tech. 59, 143 (1988). 6. F1NFER, E. Z.: J. Air Pollution Control Assoc. 17, 43 (1967). 7. SALOOIA, K. C.: J. Institute Fuel 45, 37 (1972). 8. HOWARD, J. B. AND KAUSCH, W. J.: Prog. Energy Combust. Sci. 6, 263 (1980). 9. MITCHELL, J. B. A.: Combust. Flame 86, 179 (1991). 10. RITRIEVI, K. E., LONGWELL, J. P. AND SAROEIM, A. F.: Combust. Flame 70, 17 (1987). 11. FENIMORE, C. P. AND JONES, G. W.: J. Phys. Chem. 71, 593 (1967). 12. MICHEL, A., BERT, M., VAN HOANG, T., BUSSlERE, P. AND GUYOT, A.: J. Appl. Polymer Sci. 28, 1573 (1983). 13. MANNING, M. P. AND REID, R. C.: Ind. Eng. Chem. Proc. Des. Dev. 16, 358 (1977).

1013

14. FREUND, H. J. AND BAUER, S. H,: J. Phys. Chem. 81, 994 (1977). 15. CHARALAMPOPOULOS,T.: Rev. Sci. Instr. 58, 1638 (1987). 16. ALLEN, G. C., CURTIS, M. T., HOOPER, A. J. AND TUCKER, P. M.: J. Chem. Soc. (Dalton) 14, 1525 (1974). 17. ESPENSHEID, W. F., KERKER, M. AND MATIJEVlE, E.: J. Phys. Chem. 68, 3093 (1964). 18. CHARALAMPOPOULOS,T. AND CHANG, U.: Combust. Sci. Tech. 59, 401 (1988). 19. N1KLASSON,G. A., GRANQVIST, C. G. AND HUNDERI, O.: Appl. Optics 20, 26 (1981). 20. STAGG, B. J.: Ph.D. Dissertation, Louisiana State University (1992). 21. HARRIS, S. J. AND WEINER, A. M.: Combust. Sci. Tech. 32, 267 (1983). 22. DASGH, C. J.: Twentieth Symposium (International) on Combustion, p. 1231, The Combustion Institute, 1985. 23. LAWTON, S. A.: Combust. Sci. Tech. 57, 163 (1988). 24. SHAYESON, M. W.: SAE Paper # 670866 (1967). 25. MCKEE, D. W.: Chemistry and Physics of Carbon 16, 2 (1981). 26. DONNET, J. B.: Carbon 20, 267 (1982). 27. NAGLE, J. AND STRICKLAND-CONSTABLE, R. F.: Proc. Fifth Conference on Carbon 1, 154 (1962).

COMMENTS A. F. Sorofim, Massachasetts Institute of Technology, USA. Iron is an effective oxidation catalyst and the question of its use for suppressing soot emission is how it becomes incorporated into soot in the locally fuel rich regimes of a combustor. The authors present an impressive methodology for measuring by a combination of light scattering and absorption the role of iron additives on particle growth in a sooting premixed flame. Their assumption that the iron was present in the flame as Fe203 is questionable since both equilibrium calculations and Mossbauer spectroscopy on soot extracted from such flames (Ritrievi et al, l~ suggest that the iron is most probably present in it's elemental form. The authors may wish to examine this alternative state of the iron in a reinterpretation of their optical measurements.

Dr. P. Cadman, University of Wales, U.K. XPS detects the species in the first few A, of surface only. This layer of oxide on the iron particles can be formed or modified by transfer of your sample to the spectrometer via atmospheric conditions. I do not think you can be sure what iron species are produced in the flame from the XPS results.

Alan S. Feitelberg, General Electric Co., USA. X-ray photoelectron spectroscopy (XPES) probes only the surface of a sample. Couldn't the observed Fe203 be an artifact, resulting from oxidation of Fe species on the surface when the sample was exposed to air?

Author's Reply. The comments are consistent with the view that finely divided iron is completely oxidized on air exposure 1. However, the following points need to be made: (i) The sampled particles do not consist of pure iron but instead the iron is dispersed throughout the soot matrix (ii) The samples were kept under vacuum conditions for a short period of time at temperatures less than 25~ C prior to XPS analysis. (iii) It has been reported z that X-ray analysis of iron powders subsequent to exposure to oxygen at 25~ C revealed only metal lines. On the other hand, X-ray analysis of samples exposed to 40 ~ C showed ferrous oxide, magnetite and metal. As noted in the text in this study the only species detected using the X-ray analysis was Fe203. Given that proper precautions were taken to prevent sample oxidation

1014

SOOT

during the sampling process the results of the present study suggest that the state of the iron species in the flames investigated is in the form of FezOa. Furthermore, it should be noted that irrespective of the state of the iron species in the flame the soot particles are not optically homogeneous due to the dispersion of the iron species in the soot matrix. The scattering/absorption data were analyzed utilizing the refractive index of the Fe and FeO as well as that of FezOa. The calculated differences for all inferred soot parameters were less than 10% when compared with the values reported in the paper. Nevertheless it should be pointed out that an insitu method would be preferable to determine the state of the metal species in this type of system. Work is underway in this laboratory to develop such a technique.

very good approximation for the effective refractive index.

~:.Author's Reply. It is possible that for Rayleigh size particles the volume weighted average refractive index can be a good approximation for the effective refractive index. However, it should be noted that in both cases accurate knowledge of the refractive indices of both phases (soot + metal species) ~ is required. Furthermore, for large agglomerated soot structures the volume weighted average refractive index may possess significant uncertainties. REFERENCE 1. CHARALAMPOPOULOS, T. T., HAHN, D. W. AND CHANG, H.: Role of metal additives in light seattering from flame particulates Applied Optics, Vol. 31 (1992).

REFERENCES 1. SEWELL, P. B. AND COHEN, M.: The oxidation of iron single crystals around 200 ~ Journal of Electrochemical Society, p. 501 (1964). 2. DAVIS, P. E., EVANS, U. R., F.R.S. AND AGAR, J. N.: The oxidation of iron at 175 to 350 ~ Proc. Royal Society of London A 225, 443 (1954).

A. S. Feitelberg, General Electric Co., USA. How does the use of an effective refractive index compare with the use of volume weighted average refractive index? For Rayleigh-size particles, a volume weighted average refractive index is often a

A. D'Alessio, University of Naples, ltaly. Would you think that to measure scattering at different wavelengths will be useful for evaluating the complex refractive index of the coated particles? Author's Reply. Measurement of the spectral variations of the scattering at different wavelengths will be helpful for the inference of the complex refractive index of the coated particles. However, accurate determination of the index of the coated particles requires knowledge of the index of the dominant component of the inhomogeneous soot particles as well as the nature of the agglomerated structures.