DEPARTMENT OF SOENTIFIC AND

DSIR i',".') f0CFIRE F,,,:,' HORGANlZ;':", ,~~CE I ;,p,Y

No. n"ei( #J ~ 1()

INDUSTRIAL RESEARCH

AND

FIRE OFFICES' COMMITTEE

JOINT FIRE RESEARCH ORGANIZATION

FI RE RESEARCH NOTE

NO. 570

SOOT PRODUCTION BY DIFFUSION FLAMES

by

1. S. McLINTOCK

This rrzport has not belln publishoo andshoUld 00 considered as confidentialacIVonce information. No rrzW"ence should

00 maoo to it in any publicationwithout the written consent of theDirector of Firrz Research.

'October, 1964.

Fire ReslZarch Station.

Borrzham Wood.Herts.('phone ELStrrze 1341)

© BRE Trust (UK) Permission is granted for personal noncommercial research use. Citation of the work is allowed and encouraged.

F.R. Note No. 570

DEPARTMENT OF SCIENTIFIC AND INDUSTRIAL RESEARCH AND FIRE OFFICES' COMMI'h'EEJOINT FIRE RESEARCH ORGANIZATIlli

SOOT PRODUCTiON BY DIFFUSION FLAMES

by

I. S. McLintock

Part 1. Summary of Available Information andDiscussion of its Relevance

to Fire Problems

SECTION 1.Introduction

The production of smoke in fires constitutes a serious hazard. Smokecan drastically reduce visibility and has unpleasant physiological effeots.In addition~ the heat radiated by flames is related to the conoentration ofradiating solid particles within the flame~. It would be useful to have someinformation on those factors which might influence smoke produotion by. flames.

A major component of smoke from flames is soot, a solid material generallyof. high carbon corrtent, This material makes a large contribution towards' itsobsouring power. Information on the rate of soot production by diffusion flamesunder various combustion conditions would therefore be appropriate to aninvestigation of the smoke production problem, At present there is surprisinglylittle quantitative information on the rate of soot production, even under .laboratory conditions.

In enclosed compartments the composition of the atmosphere in whioh flamingcombuatdon is taking place will not,· in general, correspond to tha.t of air.The composition will be modified by the presence of combustion products anddecomposition products; the use of inerted atmospheres or'of chemioal'extinguishants to control the fire will cause further modification of theatmospheric composition. Smoke production might be expected to vary witn suohchanges in combustion conditions and it would therefore be useful to investigaterates of smoke production when different diluents are added to a flame.

e Measurement of sooting rates, rather than smoking rates, would probably beexperimentally simpler.

The literature on soot formation in flames is extensive and the resultsare often conflioting. In part this is due to the variety of systems whioh havebeen investigated. Referenoe to Appendix 2 and Seotion 3 of this report willshow that the oonfliot is partioularly pronounoed as far as the effeots ofadditives and diluents are conoerned. Most investigations have been oonoernedwith measurements of "smoke :;,oints", i.e. the oritioal fuel flow (diffusionflames) or mixture strength (premixed flames) at which soot just begins toleave the flame. Quantitat'1ve information is confined, for the most part, tomeasurements of oarbon inside the flame. In the present oontext we areooncerned, rather, with the quantities of unburnt oarbon whioh leave the flame.

It should be made olear at this point that soot produced by flames is notelementary oarbon. It may oontain hydrogen and oxygen in a~o~9 peroentages .inexoess of 3Q%, particularly under mild oombustion conditions(1)-(5J. Briefly,soot produced under mild oombustion conditions has the oharaoteristios of.apolybenzenoid hydrooarbon; it may contain volat~les of lower molecular weight,'

\some of which may possess oxygenated functional groups. Under stronger combustionconditions a soot of higher carbon content, and with interstitial hydrogen andoxygen set .in a strained graphite lattice, is formed. In this report the term"soot" is used to cover solid material given off by flames. The word "carbon"will be used to describe such material inside the flame, on the understanding thatsuch a term may not adequately describe its chemical nature.

It will be appropriate to begin with a brief mention of current ideas on themechanism of soot formation in flames. Relevant information on thaefactorswhich are known to influence soot production will then be summarised. Bothdiffusion and premixed flames will be discussed.

SECTION 2.Mechanism of soot formation in flames

Many theories of carbon formation have been proposed.discarded as untenable but even at present the problem hasReferences to work in this field are presented in Appendixthe many adequate reviews of the subject. '

Some have later beenby no me~s_been resolved.1., as are' references to

It is generally agreed that the process of carbon formation in flames isextremely rapid; simple fuel molecules are converted to aggregates of severalthousand carbon atoms in times of the order of milliseconds. Such a process isalmost certain to involve fr.ee radicals. Electron spin resonance measurements onsoot from diffusion f~s have,. in fact, demonstrated the presence of unpairedelectrons in the soott'1) U). This may be taken as circumstantiaL evidence for afree radical mechanism.

The process of carbon formation is now considered(16)(23) to involve two stages:

(i) Conversion of fuel to carbon nuclei~

(ii) Growth of carbon nuclei by a heterogeneous.process.

A third process:

(iii), Oxidation of carbon particles in a heterogeneous reaction with oxidantssuch as C02," ~O,. 02. and OR, competes with these two carbon formationreactions.

In this context diffusion and premixed flames differ in an important respect.In a premixed flame the oxidation processes (carbon removal) proceed concurrentlywith the carbon formation processes; in a diffusion flame carbon formation isgenerally well established before the competing reactions become significant. 0

Rence a diffusion flame can exhibit luminosity due to hot carbon particles wi.thinit, but yet produce no soot. On the other hand, in premixed flames the onset ofluminosity and soot production frequently coincide.

Whether a flame soots or not depends on the balance(22)(27)(31) between thesetwo competing processes. A diffusion flame will give off soot (or a premixedflame will show carbon luminosity) if the overall rate of carbon formation exceedsthat of carbon combustion.

Investigations have been carr}e~ 9U~ on the separate processes of carbopformation by hydrocarbon pyrolys:ill ~6)-t1 ) and of carbon particle combustion~17)-(21)•.Even under these circumstances the results are not s~raightforward. In a flamethe situation will be further complicated by such factors as flame size andstruature and the considerable temperature gradients which exist in the flame.

SECTION 3.Effect of additives

Published results on the effect af additives on soot formation show a large

- 2 -

measure of disagreement~ It has been mentioned earl}er that most work has beenconcerned with the effe9t of additives on smoke pointl26)(27). and luminosity(4)(14)(24)(25)(35). Milbergl 22) has measured sooting rates of acetylene-air diffusionflames at fuel flows beyond the smoke point,. but not the effect of additives onthese rates. The effect of additives on the rate of carbon formation withinnon-smoking diffusion flames has been me\lsured(23) and similar experiments havebeen carried out with premixed flames(27).

In the case of diffusion flames additives have generally been intfod~ced tothe fuel stream rather than to the air stream. The results of Creitz 36) are ofinterest here, although they refer to the extinction of diffusion flames byhalogenated methanes; the effectiveness of the extinguishant depended on whetherit was introduced to the fuel or the air stream. When large amounts of additiveare employed it is frequently termed a "diluent", especially if the oxygenconcentration is thereby depressed significantly.

The comple::;:U.v of, and conflict between,- the available results is summarisedin Appendix 2.

The mechanism by which "inert" additi ves such as nitrogen affect sootformation has not yet been adequately explained On~ obvious effect ofintroducing such an additive will be a lowering(4)(24) of fl~e temperature.This change in flame temperature has generally been discussedl4)l24) in relationto the carbon formation reaction rather than to the overall soot productionprocess. Such an approach would not appear to be adequate, since the deoreasein flame temperature will lower the rates of both carbon formation and combustion.In theory it would be possible to calculate which rate was decreased by thelarger amount and hence the effect (on sooting rate) of a known decrease in flametemperature. The situation is complicated in practice by at least two factors:

(i) Information ~ availabl~ ~g tte ~9tivatio~ energies of carbonformation 9Y pydr~9arppn pyrolysisl 6)l8) 16 l29)-l31) and of carboncombustionl17)l19;l20)l37;. But Tesner 20 has shown that the activationenergies of both types of reaction increase with decrease in temperature. Theeffect of temperature on rate will therefore be complex.

(ii) The presence of temperature gradients throughout the flame willaggravate the difficulties referred to above.

The calculation appears to be intractable at present. The effeot of flametemperature on carbon and soot formation has been examined experimentally. Theresults are summarised in Section 4.

In any case, however, several results(4)(24) indicate that the effect of"inert"' diluents on carbon or soot formation cannot be explained purely on thebasis of t~ei~ addition resulting in a decrease in flame temperature. Streetand Thomasl24; have discussed the si~atton in terms of a decrease in oxygenconcentration. Arthur has suggestedl 32) that hydrogen atoms play an importantrole in c~rbon formation and that additive moleoules suppress carbon format}o~

by actingl32) as third bodies in hydrogen atom recombinations. Later workl4)by Arthur and Napier shows that, although hydrogen atoms may be important speciesin the reactions which lead to carbon formation, the situation cannot beadequately explained on this basis.

It will be useful here to speculate briefly on further possible explanationsof the effect of inert additives. It has been mentioned that carbon formationin flames is almost certain to involve a free radical mechanism. The freeradicals involved are likely to be those encountered in typical hydrocarbonpyrolyses as well as large polybenzenoid precursors of soot particles. It seemsreasonable that in such a system inert molecules such as nitrogen may alter theradical concentrations by acting as second or third bodies in collisionalprocesses. In this role they would playa direct physico-chemical part in thecarbon formation process.

- 3 -

~ome information is ~vailable(5)(33)(34) on the concentrations of nitrogenand combustion produc~s (C02,· H20. CO) within diffusion flames. Ni~rogenconcentrations inside a flame may be as high as 80%, even when it has not beenadded as a d.i.Luerrt; It is therefore suggested that nitrogen molecules mayinfluence the rate of carbon formation in all diffusion flames burning in air.Nitrog~n introduced as a diluent could reinforce this effect. Molecules ofcombustion products might be expected to behave similarly, but they mightinflu~nce carbon formation by additional, more obviously chemical, effects.

Other additives mentioned in Appendix 2 cannot be classed as inert.Various chemical explanations of their effect on soot formation have beensugg~sted. For example, the promoting effect of S03 could be a result ofincreased radical formation following the formation of peroxides between fuelmolecules and S03(35). It is considered(35) that S02 may suppress sootformation by reacting with solid carbon as well as by reacting with freeradicals involved in the carbon formation process. Other additives may ~c

considered to. operate as sources of free radicals; appropriate refe!:ences tohalogens and halogenated compounds are given in AppendiX 2. Additives suchas CO2, H2O and CO may alter flame chemistry by upsetting equilibrium reactionsin the flame, and C02 is involved further in the carbon combustion reactions.

The introduction of large amounts of diluent will decrease the oxygenconcentration in the flame. This might be expected(37) to lower th? r~te ofcarbon combustion. Small amounts of oxygen are known to accelerate~28) therate of hydrocarbon pyrolysis. Pyrolysis of hydrocarbons is involved inseveral of the current mechanisms of carbon formation and it is thereforepossible that changes in oxygen concentration will affect the rate of carbonformation as well as that of carbon combustion.

SECTION 4.Effect of some other variables

(a) Air flow. Smoke points of diffusion flames increase with increase inair flow (for a given burner diameter) up to a critical air flow beyond whiyhincreases in air flow have no further effect in suppressing smoke formationl26).The value(~~)this critical air flow is specific for a given system. Milberghas shown that rates of soot production for acetylene-air diffusion flamesdecrease with increase in air~fuel ratio~

(b) .Temperature. Schalla and McDonald report(26) that inorease in the flametemperature of diffusion flames either increased smoke points monotonously orelse lead to an initial deorease in smoke points followed by a steady increase.The fuel used determined which of the two effects was observed. Their methodof altering flame temperature; however, was to add varying amounts of nitrogenor argon to the oxidant flow. In the light of the discussion in Section 3above it is possible that they thereby altered other parameters as well asflame temperature.

According to Street and Thomas(24) increase in the flame temperature ofpre~xed flames decreased their luminosity and therefore decreased carbonformation.

(c) Pressure. Soot production increases with pressure(22)(26)(27) for bothdiffusion and premixed flames. In any particular case there is a criticalpressure below which soot formation does not occur.

SECTION 5.Conclusions

There are several gaps in our knowledge of the processes of soot formationin diffusion flames. In particular, information on the variation in sootingrate with combustion conditions is very meagre. Further information of this·

- 4 -

kind, with particular reference to the part played by atmospheric composition,should prove useful 'in a study of the problem of smoke production by fires. Asystematic study, on a laboratory scale, of the effeot of :fuel type" diluentsand temperature on soot production in flames would be appropriate.

To complete the investigation it would be necessary to study the effect ofthese factors on soot" or smoke, production during smouldering combustion.

References

TESNER. 8th SymP. (International) on Combustion. p.801.

SINGER and GRUMER. U. S. Bureau of Mines. Rep. of Investigations6007 (1962).

GOLOVINA and KHAUSTOVICH. ibid, p.784.

MILBERG. J.Chem.Phys• .§2, 578 (1959).

COMERFORD. Fuel l.2..o 333 (1956).

PARKER and HOTTEL. Ind. Engng. Chem, g,§" 1334 (1936).

J. Chem. Phys , l.2..o' 264 (1961).

8th SymP. (International) on Combustion; p.807.

7th Symp , (International) on Combustion, p.546.

8th Symp. (International) on Combustion, p.627.

BRADLEY and KISTIAKOWSKY.

TESNER.

TESNER.

TESNER.

STREET and THOMAS. Fuel llt". 4 (1955).

BONHOEFFER and HARTECK. Z.Phys.Chem.1.22. 64 (1928).

SCHALLA and McDONALD. 5th SymP. (International) on Combustion, p.316.

FENIMORE et ale 6th SymP. (International) on Combustion, p.242.

PORTER. AGARD Combustion Researches and Reviews, Butterworths 1955, p.108.4th Symp. (International) on Combustion, p.•248.

DATSCHEFSKI, Ph",D" Thesis. University of Sheffield (1962).

WICKE.' 5th Symp. (International) on Combustion, p.245.

ATEN and GREENE. Disc" Farad.. Soc. 22, 162 (1956).

STEHLING et ale 6th Symp. (International) on Combustion, p.247.

SILCOCKS. Proc , Roy. Soc.~, 411 (1957).

KOKlJRIN, Zh. Prikl. Khim. b 1784 (1963), cf. C.A • .22., 1507d.

TESNER and YECHEISTOVA, Dokl. Akad. Nauk, SSSR• .§l. 1029 (1952).

COMERFORD. Fuel2b 67 (1953)

THOMAS. Comb. and Flame f, 46 (1962).

THORP et ale Fuel llt" Sl (1955) •

RAY and LONG. Comb. and Flame.§., 139 (1964).

ARTHUR and NAPIER. 5th Symp. (International) on Combustion, p.303.

(1)

(2)

.(3)

.; (4)

(5)

(6)

(7)

(8)

(9)

(10)

( 11)

( 12)

(13)

( 14)

( 15)

(16 )

(17)

(18 )

(19 )

(20)

(21 )

(22)

(23)

(24)

(25)

(26)

(27)

- 5 -

(28) VOEVODSKY. TransvFarad.Soo., 22.> 65 (1959)v

(29) HOOKER. 7th SymP. (International) on Combustion, p.949.

(30) MINKOFF et ale J. Appv Ohem, L 406 (1957)v

(31) MILLIKAN. J. PhysvChem• .§§., 794 (1962).

(32) ARTHUR. Nature.1.22.. 557 (1950).

(33) GORDON et ale 7th Symp" (International) on Combustion, p.317.

(34) BURKE and SCHUMANN. Ind. Engv Chem, 20, 998 (1928).

(35) GAYDON and WHITTINGHAM, ?roo. Roy. Sao. 189A, 314 (1947).

(36) CREITZ. Journal of Researoh~ Nat. Bureau of Standards 65A, 389 (1961). ~

(37) LEE et ale Comb. and Flame.2. 137 (1962).

,/

- 6 -

APPENDIX I

Bibliography of mechanism of carbon and soot formation

oj

Referenoe

Gaydon and Wolfhard, "Flames",. Chapman and Hall, 2nd Ed" 1960.

Ch.VIII.

Minkoff and Tipper, "Chemistry ofCombustion React.Lons":Butterworths 1962·, pp.343..,7.

Singer and Grumer, Bureau of MinesRep. of Investigations 6007 (1962)and 7th Symp. (International) onCOnibustion; p.559

Porter, AGARD Combustion Researchesand Reviews" Butterworths 1955,p.108~ 4th Symp. (International)on Combustion, p•.24B.

Datschefski, Ph.D. Thesis, Univ. ofSheffield, 1962.

Gill. Rev.No.182. British CoallJ'J:;ilis. Res ; Aasoc ; XII" No.'12,Part II (1958).

Street and Thomas, Fuel ~' 4 (1955).

Panel discussiclh 5th Symp.Q:nternational) on Combuabdon;p •.801.

Schalla and Macdonald, ibid,. p.3t6

Arthur and Napier ibid, p.303 •.

Notes

Review

Review

Review. Results on rich flames.' Roleof CO in carbon formation.

Review of role of hydrocarbon pyrolysisin carbon formation. Hydrocarbons~~ C2~

Review. Results on soot growth in afuel-rich turbulent flow system.

Review.

Review. Results for premixed flames;effeot of temperature, O2concentration, additives.

Review

Review. Results on diffusion flames;effect of ~ enrichment, argon~

temperature, pressure etc.

Diffusion flames; normal and reversedflames; effect of additives;analysis of smoke deposits.

Diffusion flames; analysis of smokedeposits.

Review. Results for diffusion flames;effect of air flow and additives;analysis of smoke deposits.·

Diffusion flames; formation of carbonand polycyclic hydrocarbons.

Thorp et ale Fuel ~ si (1955).

Ray and Lon~. Comb. and Flame .§.,139 (1964).

Arthur et ale Comb.' and Flame b}267 (1958).Nature, ill. 372 (1952)

Garner et al, Fuel ,g,. 116 (1953) Diffusion flames;deposits; dianeformation.

analysis of smoketheory of carbon

Milbev~. J" Phys , Chem• .22.,578 l1959).

- 7 -

C2 H2 - air diffusion flames; smokingrates; effeot of· pressure and fuel/airratio; analysis of smoke deposits.

Referenoe

Tesner et ala 8th Symp~ (Inter~national) on Combustion, p..s01"

Smith and Gordon. J .i Phys,,' Chem",£Q, 759, (1956).

Gordon et al Q 7th Symp~

(International) on Combustion,P0317.

Notes

Diffusion flames'; effect of additiveson carbon concentration, dispersionand surfaoe area within the flame.

Diffusion flames. 'C H;~ CHydrocarbons _..--- 6 -"b s

--~ C2 H2 --+ Cs

Diffusion flames. Hydrocarbons

~C6H6~Cs

Fenimore et alo(International)p.242 •

6th Symp.on Combustion,

Premixed flames. Hydrocarbon~C2 H2 - Cs Carbon oxidation

reactions; effect of additives.

Cole and' Minkoff. Proc; Roy. Scc, ,~ 280 (1957).

Thomas. J. Chem.Phy s., 20. 899 (1952)

Parker and Wolfhard. J. Chem" Soc s,1.22Q,' 2038

Sweitzer and Heller. Rubber World,'1l.!I: 855 (1956).

Gaydon and Wolfhard. ProoJRoy.Soo.,20M, 570 (1950).

Behrena., 4th SymP. (International),on Combuab i on; pp.252 and 358.

Evidence ,against C2 H2 theory.

Hydrocarbons _ chain polymers~ Cs•

Mechanism of polymerisation followedby graphitisation.

Carbon formation via polybenzenoids.

Premixed flames" Role of C2 and ofpolymerisation and unsaturation incarbon formation.

Role of the reaction 2 CO"'= Cs + C02.in carbon formation.

Stehling et al"(International)p.774'.

8th Symp.on Combustion.

Mechanisms of carbon nucleation incarbon formation.

Clarke et al" J. Inst" Petrol.b 627 (1946).

Millikanv J" Phys.Chem. 66, 794(1962).

Wolfhard and Parker" Proo.Phys.Soo.,.§2!. 2. (1952).

Diffusion flames; smoking tendenciesof hydrocarbons.

Non equilibrium carbon formation inpremixed flames. (C2 H4 -" air).C2 ~ ---+ C2 H2 - Cs' Activationenergies • cr , Kydd" Comb. and Flame,l> 133 (1959).

Speotroscopic studies of diffusionflames ..

Norrish et al~ Proo.Roy.216A. 165 (1953)227A, 423 (1954)

Soc. Stoichiometry of carbon formation inhydrocarbon explosions.

Payne and Weinberg. Proc..Roy.Soc.,25QA. 316 (1959).

Ibicuru and Gaydon. Comb. andFlame.§., 51 (1 964) •

- 8 -

Effect of electric fields on carbonformation.

Spectroscopic study of effect ofinhibitors on a counterflowdiffusion flame.

Narasimhan and Foster. 10th s.YmP­(International) on Combustion,Paper 132•

applicationto 2CO~ Cs +

...

Ferguson;(1957).

Reference

Comb.' and Flame -l> 431

Notes\

Isotopic tracer studies;to C2 H2 mechanism andC02 machand sm.,

Rate of Boot growth in turbulent flowwith combustion products and methane.

- 9 -

,i

'J

.APPENDIX II

Effect of additives on carbon and soot formation

Additive System Effect Referenceor diluent

N2 Diffusion flames Suppression (a),(b).(d),(e)

Diffusion flames; Suppression (butcarbon within flame dispersion increased) (c)

Hydrocarbon pyrolysis Suppression (f)

Diffusion flames Promotion (t)

Premixed flames Promotion .2!:nil , (g),(h)

A and He Premixed flames Suppression (g)

H:2 Diffusion flames; Suppression (but(c)carbon within flame dispersion increased)

Hydrocarbon pyrolysis Suppression (f)

Premixed flames Suppression (g), (h)

Premixed flames Promotion (i)(addition of H atoms)

Diffusion flames Suppression (d) -

CO2 Diffusion flames Suppression (a),(b),(e),(d)

Premixed flames Suppression (g) ~

Premixed flames Nil (h)

CO Diffusion flames Suppression (b),(d)

Premixed flames Promotion ( j)Premixed flames Suppression (j),(h)Premixed flames Nil (g)

S02 Diffusion flames Nil (suppression in (h)very rich flames)

Diffusion flames Suppression (b).(s)Premixed flames Suppression (k)

S03 Premixed flames Promotion (g),(k),(h)

Diffusion flames Suppression? ( j)

CS2 Premixed flames Nil (or slight promotion (h)

H:2S Premixed flames Suppression (k)

- 10 -

APPENDIX II (Cont.d )

Additive System Effect Referenceor diluent

~O Premixed flames Suppression (g)

Premixed flames Nil (h)

Diffusion flames Suppression (d)

Halogens Premixed flames Promotion (h)and Premixed flames Promotion (also for (i)halog$Jll;8d (addition of other atoms andcompotnls halogen atoms) radicals)

Diffusion flames Promotion (l),(n),(s)

Premixed flames Very small effect (k)

Argon Diffusion flames. Results complex (m)See reference

NO, N20 Diffusion flames Promotion or nil (p)

N02 Diffusion flames Suppression (q)(coal)

Others (c)'(y),(h),(q ,(r)

- 11 -

,

(a)

(b)

(c)

(d)

(e)

i . (f)

:i (g)

(h)

(i)

(j)

(k)

(1)

(m)

(n)

(p)

(q)

(r)

(8)

(t)

Reference to Appendix II

Bonhoeffer and Harteck. Z.Phys~Chem.1l2" 64 (1928).

Arthur and Napier.· 5th Symp. (International) on Combustion, p.303.

Tesner et ala 8th Symp. (International) on Combustion, p.801.

Herbst. Brennstoff-Chem• .!l!l:, 1 (1963) cf. C.A. 2§., 7761a.

Iyengar et al. J. Sci.Industr.Res• .1.ib 455 (1952).

Tesner and Rafal'kes, Dokl., Akad , Nauk. SSSR~ 821 (1952) cr , C.A.!J:lJ 3671 e.

Fenimore et ala 6th Symp. (International) on Combustion. p.242.

Street and Thomas. Fuel~. 4 (1955).

Sacks and Ziebell, Nat.,Aero.Estab., Lab.Rep. LR-30, June 195.2.

Gaydon and WOlfhard, "Flames", Chapman and Hall, 2nd Ed., p.186.

Gaydon and Whittingham, Proc.Roy.Soc. 189A, 314 (1947).

Ibicuru and Gaydon. Comb. and Flame, §., 51. (1964).

Schalla and Mcdonald. 5th Symp. (International) on Combustion, p.316.

Ray and Long. Comb. and Flame §., 139 (1964).

Wolfhard and Parker, Proc.Phys.Soc • .§.2!,. 2 (1952).

Billington, British Coal Utilis.Res.!ss., Information Circular No.64,Sept. 1952.

Long. Comb. and Flame &, 50 (1962).

Cole and Minkoff. Proe.Roy.Soe.~ 280 (1957).

Singer and Grumer.· U.S. Bureau of Mines Report of Investigations 6007(1962).

- 12 -