Chemical Monographs UC-NRIUF CHEMISTRY OF COMBUSTION J. NEWTON FRIEND D, Van Nostrand Company CHEMICAL MONOGRAPHS EDITED BY A. C. GUMMING, O.B.E., D.Sc., F.I.C. The Chemistry of Combustion CHEMICAL MONOGRAPHS EDITED BY A. C. GUMMING, O.B.E., D.Sc., F.I.C. THE progress of Chemistry is so rapid that it is becoming a matter of ever-increasing difficulty to keep abreast of the modern developments of the science. The volume of periodical literature is so enormous that few can hope to read, far less assimilate, all that is published. The preparation of summaries has therefore become a necessity, and has led to the publication of various well-known journals devoted to the abstraction of original papers. For obvious reasons, however, these do not fully supply the wants of advanced students and research workers, and it is now generally recognised that monographs on special subjects are also needed. This series of monographs is intended primarily for Advanced and Honours students. As each mono- graph is written by an author with special knowledge of the subject, and copious references are given, it is hoped that the series will prove useful also to those engaged in research. The following volumes are ready : THE ORGANOMETALLIC COMPOUNDS OF ZINC AND MAGNESIUM. By HENRY WEEN, M.A., D.Sc., Ph.D., Head of the Department of Pure and Applied Chemistry at the Muni- cipal Technical Institute, Belfast. THE CHEMISTRY OF DYEING. By JOHN KERFOOT WOOD, D.Sc., F.I.C., Lecturer on Physical Chemistry, College of Tech- nology, Manchester. THE CHEMISTRY OF KUBBER. By B. D. PORRITT, F.I.C., M.Sc. THE FIXATION OF ATMOSPHERIC NITROGEN. Second Edition. By JOSEPH KNOX, D.Sc., Lecturer on Chemistry, University of Glasgow. THE CHEMISTRY OF LINSEED OIL. By J. NEWTON FRIEND, D.Sc., Ph.D., F.I.C., Head of the Chemistry Department, Municipal Technical School, Birmingham. THE CHEMISTRY OF COAL. By JOHN BRAITHWAITE KOBERTSON, M.A., B.Sc., A.I.C., Lecturer on Chemistry at the South African School of Mines and Technology, Johannesburg. The Chemistry of Combustion BY J. NEWTON FRIEND D.Sc.(B'ham.), Ph.D.(Wiirz.), F.I.C. Carnegie Gold Medallist ; Head of the Chemistry Department, Municipal Technical School, Birmingham NEW YORK D. VAN NOSTRAND COMPANY EIGHT WARREN STREET 1922 $ \3> PREFACE THIS little book is the outcome of a series of lectures delivered to my senior students in the Chemical Department of the Birmingham Municipal Technical School during the Session 1920 to 1921. There appeared to be no small Text-book or Monograph to which the students could be referred dealing with the subject in its more modern aspect, and it was felt that the publication of my lectures might serve to fill, however imperfectly, a very obvious gap in our literature. My sincerest thanks are due to .Dr A. Parker, who has carefully read through the proof sheets and made many valuable suggestions. I am glad also to take this opportunity of thanking Miss Annie R. Russell, B.Sc., of this School, for assistance and advice. J. NEWTON FRIEND. November 1921. 490086 CONTENTS PAGE SECTION I. DEFINITIONS ..... 1 SECTION II. PHLOGISTON 9 SECTION III. THE COMBUSTION OF SOLID CARBON . 12 SECTION IV. FLAME . . . . .19 SECTION V. THE COMBUSTION OF GASEOUS HYDRO- CARBONS AND OTHER GASES . . .36 SECTION VI. IGNITION TEMPERATURES . . .47 SECTION VII. THE INFLAMMATION OF GASEOUS MIXTURES ...... 60 SECTION VIII. PROPAGATION OF FLAME IN GASEOUS MIXTURES ...... 69 SECTION IX. SURFACE COMBUSTION . . .89 BIBLIOGRAPHY AND NOTES . ' . .101 INDEX 109 vii THE CHEMISTRY OF COMBUSTION SECTION I. DEFINITIONS. AN old Spanish proverb states that "Where there is no hook to be sure there will hang no bacon." With equal truth it might be said that where there are no definitions to be sure there can be no science. For science is organised knowledge, and unless our terms are carefully defined our dissertations will inevitably be diffuse and obscure, if not indeed actually misleading. In no field of science is the need for a clear conception of the meaning of certain 'terms more necessary than in that now under discussion. The knowledge that certain substances will burn must be almost as ancient as man himself. Such terms as fire, flame, and combustion have for centuries been household expressions, and as such they have been frequently used to denote phenomena, which to the popular mind may appear like, but which in reality are widely separated from each other. In a critical discussion, therefore, of the various phases of combustion, it becomes necessary, if the A .2 , DEFINITIONS introduction of a large number of new terms is to be avoided, to give at the outset precise meanings to such popular terms as it is desired to employ. By combustion it is now usual to imply some form of chemical change accompanied by the evolu- tion of both heat and light. Thus, for example, the raising of platinum wire to redness through the agency ,of an electric current and the production of beautiful optical phenomena by passing electrical discharges through Geissler tubes are not examples of combustion, for although both heat and light are emitted, they are not accompanied by chemical change. On the other hand, combustion takes place when phosphorus burns in oxygen, for the enormous heat and dazzling light are accompanied by vigorous chemical change. Although in the large majority of cases combustion is the result of oxidation, it would be erroneous to suppose that combustion cannot also take place in the entire absence of oxygen whether free or com- bined. Thus if a small jet of burning coal-gas 10 is plunged into a jar of chlorine, combustion continues, although a marked change takes place in the attendant phenomena. The flame becomes deep orange in colour, and volumes of black soot are evolved ; the yellowish-green colour of the chlorine gradually disappears, giving place to steamy clouds of hydrochloric acid gas. A familiar lecture ex- periment is to plunge yellow phosphorus into bromine vapour when it readily burns yielding clouds of mixed bromides, whilst powdered arsenic dropped upon liquid bromine immediately inflames. A piece DEFINITIONS $ of copper foil burns brilliantly in sulphur vapour, and a rod of iron heated to redness and pressed against a piece of sulphur behaves similarly in the vapour thus produced, the molten ferrous sulphide falling in scintillating globules to the ground. These are, all of them, examples of true combustion, but of combustion in the entire absence of oxygen. As mentioned above, however, in the vast majority of cases combustion is accompanied by oxidation, mainly in consequence of the fact that the chemically active constituent of the atmosphere is oxygen. Had our atmosphere consisted of chlorine we should still have been familiar with combustion, but most examples would have been the result of chlorination. One of the chief factors which determines whether or not chemical change shall be accompanied by combustion is the magnitude of the heat evolution. On applying a light to a jet of hydrogen issuing from a tube into the air, the hydrogen readily burns, and if the surrounding air be replaced by oxygen, the temperature of the reaction rises rapidly, an intense heat being produced. The reaction is thus powerfully exothermic, the amount of heat evolved being as follows : (H 2 ) + (O) = (H 2 O) + 58,700 calories at room temperature. The physical state of the reacting masses is indicated by the brackets. Rounded brackets mean that the substances are present in the gaseous state ; square brackets [ ] indicate the solid state, whilst an entire absence of brackets indicates the liquid state. 4 DEFINITIONS The above equation tells us that gaseous hydrogen and oxygen unite to form water vapour, and that for every gram-molecule of hydrogen burned or of water produced 58,700 gram-calories of heat are evolved. If the water vapour is allowed to condense to liquid water, allowance must be made for the change of state involved which causes the evolution of still more heat the latent heat of steam. This must be added to the previous amount and our equation now^ becomes 1 (H 2 ) + (O) = H 2 O + 68,360 calories. The molecule of water is not bracketed as it is now assumed to be in the liquid state. But when hydrogen combines with oxygen, com- bustion does not always ensue. This has been known for many years. Hooke in 1803 observed that electrolytic gas confined in vessels over water very slowly combined at the room temperature, 2 the amount of combination being appreciable only after several months. As the temperature is raised, the rate of combination steadily increases 3 and proceeds with a measurable velocity at about 450 C. 4 There is still no sign of combustion, however, in the sense in which the term has been defined above, so that evidently the mere fact that a reaction is powerfully exothermic is not in itself sufficient guarantee that it shall be accompanied by combustion. The condi- tions obtaining at the time of experimentation must also be considered. It is immaterial in regard to the quantitative heat DEFINITIONS 5 evolution whether one element combines rapidly with another, or slowly. The same amount of heat per gram-molecule is evolved in either case, provided the initial and final phases are identical in the two sets of experiments. If the reaction proceeds very slowly the heat set free will ordinarily be dissipated with such relative rapidity that no sensible rise in temperature occurs, and the reaction, may even proceed to completion without the phenomena characteristic of combustion ever appearing. From the foregoing it will be clear that exothermic chemical reactions may take place with all degrees of velocity, ranging from that which proceeds so slowly as to be measurable only after many months up to that which is practically instantaneous or explosive. When a reaction occurs with a sensible rise in temperature, but unaccompanied by light, it is frequently termed slow combustion. The oxidation of impurities in the blood by air drawn into the lungs is a familiar example. In the majority of these cases the reaction does actually proceed slowly, and the term is unexceptionable. But such is not always the case. Every student of chemistry knows that nitric oxide, itself a colourless gas, is characterised by the readiness with which it combines with oxygen to yield rich brown fumes of nitrogen peroxide. The reaction is markedly exothermic, the heat evolution being as follows : 2(NO) + (O 2 ) = (N 2 OJ + 40,500 calories, and if suitable precautions are taken, a rise in temperature can readily be observed. This is an 6 DEFINITIONS example of "slow" combustion, as defined above, although the reaction is not in itself a " slow " one. Bearing in mind, however, the extended use of the prefix in this connection, the term slow combustion is a very convenient one to retain. Although most substances require to be raised in temperature before combustion can ensue, a number of reactions are known in which the in- gredients react vigorously with evolution of light and heat practically from the moment they are brought into contact with one another. Such re- actions are examples of spontaneous combustion. Thus sulphur spontaneously burns in fluorine even at -187C. forming the hexafluoride SF 6 . At ordinary temperatures both ammonia and nitric oxide inflame when brought into contact with fluorine, yellow phosphorus readily catches fire in chlorine, and powdered arsenic instantly inflames on the surface of liquid bromine. The pyrophoric metals become incandescent in moist air, whilst liquid phosphine, silico-ethane, and many organic substances imme- diately ignite in contact with oxygen or air. The temperature at which rapid combustion becomes independent of external supplies of heat is known as the ignition temperature. A mass of gas raised to incandescence by heat is termed a flame; this latter is produced only in those cases of combustion in which gases or vapours are present. Flame, however, does not always accom- pany rapid gaseous combustion, a striking exception being afforded by the rapid oxidation of hydrogen or coal-gas mixed with air on a catalysing surface, DEFINITIONS 7 such as that of platinised asbestos or porous firebrick. Such combustion is termed flameless or surface com- bustion, and is utilised commercially in a variety of ways. A substance undergoing slow combustion may exhibit pronounced luminosity or phosphorescence. This is not an exceptional phenomenon, as was formerly believed, but a natural prelude to rapid combustion, the appearance of flame being the cul- minating point of a series of changes, and coincident with the ignition temperature. 5 Many substances are now known to exhibit phosphorescence, but the temperature of its appearance is so high or the temperature interval during which the phenomenon is perceptible is frequently so short as to be ordinarily overlooked. In the case of phosphorus it is par- ticularly well marked, the temperature intervals, namely, 7 to 60 C., including the ordinary range of atmospheric temperature, so that the phosphorescence could hardly be overlooked. Had an arctic climate prevailed, however, the phosphorescence would not have been so readily observed, for in igniting phos- phorus with a match or taper the phosphorescent temperature interval would be rapidly passed, as is the case under existing conditions with sulphur. Occasionally phosphorescence is referred to as de- graded combustion. This is not a happy term, how- ever, suggesting as it does that combustion is not complete. This may often be the case, but it is not universally so, sulphur being a case in point. This element phosphoresces direct to the dioxide, no incomplete oxidation products being obtained. 5 8 DEFINITIONS When a reaction proceeds with a rise of temperature and an ever-increasing velocity until a high maximum velocity is attained, an explosion or detonation results. Combustion is greatly facilitated by fineness of division. This is well illustrated by the pyrophoric metals. If yellow phosphorus is dissolved in carbon disulphide and the solution poured over some filter paper, the solvent rapidly evaporates leaving the phosphorus in a very fine state of subdivision, and inflammation rapidly takes place. Many explosions in factories, mines, etc., have been definitely traced to the presence of dust, the activity of which is largely a surface phenomenon akin to those now under discussion. The influence of dust may be illustrated in a harmless manner by introducing some very finely powdered coal or charcoal into a gas jar to the depth of about half an inch, and blowing oxygen into it from a glass tube reaching to the bottom. When the oxygen-enriched air in the jar is thick with dust, a light is applied with a long taper, and a flame flashes down the jar with explosive violence. SECTION II. PHLOGISTON. 1 FOR many centuries prior to the discovery of oxygen it had been assumed that all combustible bodies possessed a combustible principle the fire matter of the Greek philosophers. Owing to its ready com- bustibility, sulphur was for long regarded by European alchemists as the essential principle of combustion. Becher (1635-1682) christened this principle terra pinguis or "oily earth," and distin- guished it from sulphur which, however, he regarded as rich in this principle. When a substance burned, therefore, the oily matter escaped leaving an incom- bustible r ash or calx (from the Latin calx, lime). Stahl (1660-1'734) developed Becher's views, modified and extended them. He termed the combustible principle phlogiston from the Greek \oyifav, to ignite. When, therefore, a metal such as zinc burned in air, phlogiston escaped, the residual ash of dephlogisticated metal being termed zinc calx. Thus, zinc = zinc calx + phlogiston. The residual air (nitrogen), now saturated with phlogiston was termed phlogisticated air. Stahl appears to have regarded carbon as almost pure 10 PHLOGISTON phlogiston, and other chemists subsequently viewed hydrogen in the same light. So long as the adherents of the phlogistic theory were content to regard phlogiston as a principle only, and not as a material body, the theory was distinctly attractive, phlogiston being the prototype of what we now term the heat tone or heat of reaction. Such a theory, however, could not offer a complete explanation for many of the phenomena known, even in the times of Becher and Stahl, to be attendant upon combustion, but the materialisation of phlogiston, whilst removing some difficulties, introduced many others. Thus, for example, the identification of phlogiston with hydrogen enabled chemists to correlate the escape of this gas upon the solution of metals in acids with the combus- tion of the same metals in air; at the same time it introduced the very serious difficulty that no hydrogen could actually be detected in the air after the combustion of a metal, and no answer was forthcoming as to whither it had escaped. 2 The two chief difficulties which ultimately led to the overthrow of the phlogistic theory were the following : 1. It had been known for many centuries that ordinary combustion would not take place in the absence of air. Thus the Arab chemist Geber, in the eighth century, stated that the calcination of mercury must be carried out in open vessels, and Stahl himself was aware that even soot, which he regarded as almost pure phlogiston, would not burn out of contact with air. This was explained on the PHLOGISTON 11 assumption that the phlogiston could not leave a substance unless it had somewhere to go to. In other words, the air was believed to act as an absorbent for the phlogiston just as a sponge sucks up water or as charcoal adsorbs colouring matter from liquids and bad odours from gases. This ingenious explanation, however, takes no cognisance of the fact that the volume of the air actually becomes smaller during the calcination of metals. 2. In 1630, Jean Rey had, drawn attention to the fact that both lead and tin exhibit an increase in weight upon calcination in air. Now, according to the phlogistic theory, metal = calx + phlogiston, so that even if phlogiston was a non-material and imponderable principle, the metal could not weigh less than the calx ; whilst if phlogiston was material- ised into carbon, hydrogen, or any other substance, the metal must weigh proportionately more than its calx. This anomaly was realised by several chemists ; but mere recognition of a difficulty does not involve its solution, and it was not until the discovery of oxygen by the Birmingham divine, Priestley, 3 that Lavoisier was able to offer a more correct theory of calcination and aerial combustion. According to him, when a substance burns in air it combines with the oxygen to form an oxide^an explanation that is regarded as correct at the present time. SECTION III. THE COMBUSTION" OF SOLID CARBON. IT is a matter of common knowledge that when carbon is allowed to burn in excess of air or oxygen the only product is carbon dioxide ; whereas, if the supply of air is restricted, carbon monoxide appears in amounts varying according to the conditions. This is capable of explanation in one of three ways, namely : 1. It may be assumed that the first product of the interaction of carbon and oxygen is the dioxide, C0 2 , and that this, in contact with excess of carbon, is reduced to the monoxide, CO. Thus : (1) (2) C0 2 + C = 2CO. ^ After the overthrow of the phlogistic theory, and during the major part of last century, this was the commonly accepted theory. 1 2. In 1872 Sir Lowthian Bell, as the result of long continued study of the chemical processes involved in iron smelting, concluded that "carbon monoxide and not carbon dioxide is the chief, if not the exclusive and immediate action of the hot blast on the fuel " (coke). 12 COMBUSTION OF SOLID CARBON 13 According to this the presence of carbon dioxide is due to oxidation of the monoxide in excess of air, the gas thus being a secondary instead of a primary product. Thus : (1) 2C + O 2 = 2CO (2) 2CO + 2 = 2CO 2 . This second theory received support from a number of interesting observations. Thus in 1887, C. J. Baker 2 observed that carbon, which has been thoroughly dried by exposure to phosphorus pent- oxide, is allowed to absorb oxygen, dried in a similar manner at 12 C. and is then heated to 450 C. ; the evolved gas consists mainly of carbon monoxide. Having satisfied himself that, under the conditions of his experiments, carbon dioxide if initially formed could not have been reduced to the monoxide, Baker concluded that carbon is oxidised directly to carbon monoxide by the absorbed oxygen. These conclusions were supported the following year by H. B. Baker 3 who showed that : (i.) Thoroughly dry carbon dioxide is not reduced by dry carbon even at bright red heat. (ii.) Carbon monoxide is the main product of the combustion of carbon in dry oxygen. Thus, when oxygen which had been thoroughly dried by prolonged contact with phosphorus pent- oxide was passed over highly purified sugar charcoal at bright red heat, no visible combustion occurred. 14 COMBUSTION OF SOLID CARBON The gases passing over possessed the following composition : Oxygen . . . .58-1 per cent. Carbon monoxide . . 39-5 Carbon dioxide . . . 2 8 2 Bell's theory thus appeared to be well substantiated. 3. In 1913, however, a new complexion was put upon the whole problem by the extensive researches of Rhead and Wheeler. 4 These investigators found that charcoal which has been heated in a vacuum up to 950 C. and allowed to cool, readily absorbs or occludes appreciable quantities of oxygen at all lower temperatures. The amount of occluded oxygen increases with fall of temperature, and remains surprisingly constant for any given temperature for the particular specimen of charcoal employed. The rate of absorption of oxygen by exhausted charcoal is exceedingly rapid during the first fifteen seconds, after which a slow absorption continues over several hours. This is well illustrated in Fig. 1, which depicts the relative amounts of oxygen absorbed at various temperatures during short intervals of time. The type of curve is seen to be the same for each temperature, the main differences consisting not in the rates of fixation of the oxygen, but in the respective quantities of the gas. This rather suggests that the absorption is a purely f physical phenomenon. On the other hand, it is found that the oxygen clings so tenaciously to the carbon as to defy removal by the most complete exhaustion at any given temperature, provided that temperature COMBUSTION OF SOLID CARBON 15 remains constant. If the temperature is raised by a certain amount, a definite quantity of oxygen is removed in vacuo in the form of a mixture of carbon monoxide and dioxide until the saturation point of the charcoal at this new temperature is reached, after 10 TIME IN MINUTES. 20 FIG. 1. Absorption of Oxygen by exhausted Charcoal (Rhead and Wheeler). which no further gas is removable without another rise in temperature. Thus, if the temperature were raised from 300 C. to 350 C. the amount of oxygen liberated in vacuo as oxides of carbon would correspond to the fall in concentration in the charcoal from A to B. If this is a case of purely physical absorption it 16 COMBUSTION OF SOLID CARBON is surprising that a reduction in pressure should alone be insufficient to remove either oxygen or its oxidation products, the monoxide- and dioxide' of carbon, at temperatures below 900 C. 5 Again, one would have expected that admission of either of the oxides of carbon to exhausted carbon at tempera- tures sufficiently low to avoid chemical action, for example, 250 C., would result in a partial absorp- tion of some of them. This, however, has been demonstrated not to be the case. Rhead and Wheeler therefore conclude that the first product of the combustion of carbon is a loosely formed complex which may be regarded as an unstable compound of carbon and oxygen of an unknown formula, C^O^. This is presumed to be formed by the " fixation " of such oxygen molecules as come into collision with the carbon. There are ^ good grounds for believing that the carbon molecule is extremely complex in structure, 6 and the authors suggest that during oxidation the oxygen molecule may actually penetrate into the carbon molecule, a rearrangement of the atoms taking place. However this may be, it is sufficient to assume that the oxygen molecule is temporarily fixed. The repeated fixation of oxygen molecules, however, causes the evolution of a considerable amount of heat, so that some of the molecules eventually acquire sufficient energy to seize hold of a carbon atom and depart with it as carbon dioxide. Some of them become torn apart in the process and leave the carbon molecule as carbon monoxide. The formation and partial decomposition of the COMBUSTION IN COKE-FIRED FURNACE 17 intermediate compound, C x O y , continues until the carbon becomes "saturated" with oxygen, after which there is alternate formation and decomposition of the complex. Each oxygen molecule that impinges on the carbon liberates so much energy upon fixation that the equivalent amount of monoxide or dioxide is liberated by disruption of a certain quantity of C x O y formed from previous oxygen molecules. This attractive theory, whilst not definitely proven, appears to fit in extremely well with known facts, and is not at variance with any of the arguments brought forward in support of either of the two previous theories. Composition of the Complex, C x O y . The quantity of oxygen absorbed by a sample of charcoal at 300 C. in one experiment amounted to 0-16 gram per 12 grams of the latter, so that the empirical composi- tion of the complex would be represented by the formula, C 100 O. On the other hand, the relative proportions of carbon dioxide and monoxide evolved on raising the temperature of saturated charcoal is found to vary with the initial temperature, so that it would appear impossible 7 from available data to determine the actual values for x and y. Combustion in a Coke-fired Furnace. When a current of ordinary air is injected into a furnace on to a mass of incandescent carbon (coke), the complex C X O initially formed is rapidly decom- posed yielding a mixture of the two oxides of carbon. B 18 COMBUSTION OF SOLID CARBON These tend, also rapidly, to reach a condition of equilibrium as represented by the equation 2CO^ C + CO 2 . Accordingly, if the time required to attain to this equilibrium at high temperatures is less than that taken by the gaseous phase to traverse the fuel bed, given constant pressure and temperature, the ratio C0 2 /CO, as found in the emergent gases, should remain constant. If the coke is at a high temperature, namely, not less than 1200 C., then approximately 99-94 per cent, of the gaseous mixture consists of carbon monoxide, and the reaction may be considered as taking place according to the equation 2C + O 2 = 2CO + 58,000 calories, although, theoretically, no matter how high the temperature attained, a small but definite proportion of carbon dioxide must always be present to maintain equilibrium. SECTION IV. FLAME. A FLAME has already been defined as a mass of gas raised to incandescence by heat. According to this definition it is possible for us to have flame without combustion, and this may be experimentally realised during the passage of electricity through rarefied gases. Such is the exception and not the rule, however, and our concern is with those flames that are a manifestation of combustion. Perhaps the most satisfactory method of studying flames is to take a few typical examples and examine them in detail. The Candle Flame. The candle flame is a never-failing source of interest. The fuel is solid, but is gradually liquefied by the heat generated when once combustion has been started, and lies in a cup-shaped hollow at the foot of the wick. By capillary attraction some of this molten fuel is continually being drawn up the wick, and, reaching a much hotter zone, is not only vaporised but ignited. If the candle flame is blown out, and the whitish vapours of paraffin escaping from the wick are brought immediately 19 20 FLAME into contact with a lighted taper (and thus not given time to cool and condense), a flame from the latter instantly darts across the intervening space to the wick, and the candle relights. The flame consists of gases which, at the tem- perature prevailing, are lighter than the surrounding air, so that the flame rises. This is a fortunate NON LUMINOUS MANTLE LUMINOUS ZONE NON LUMINOUS INNER ZONE BLUE ZONE FIG. 2. circumstance, although in a sense it is purely an accident. One result is that as the flame burns, cool air is drawn up from below, as shown in Fig. 2, and keeps the outer edge of the wax cool. It thus enables the molten fuel to lie in a cup-shaped re- ceptacle of solid fuel without overflowing, as already mentioned. If the wick is not central, or if the candle is not cylindrical, or finally, if the flame is THE CANDLE FLAME 21 exposed to a draught, this balance is disturbed and may result in molten wax overflowing and running down the sides where it congeals in characteristic streaks. As a rule, therefore, ornate candles are not as efficient as the plain, cylindrical ones. Examina- tion of the candle flame itself reveals a number of interesting points. 1 It is possible to distinguish four parts in the flame, namely : (1) The inner, non-luminous zone (see Fig. 2) which consists essentially of vaporised FIG. 3. paraffin wax. It has not had a chance to burn as yet, for appreciable amounts of air have not penetrated so far. It represents the first stage in the gasification of the wax. If a narrow glass tube is inserted in the flame as shown in Fig. 3 with its lower end 22 FLAME in the non-luminous zone, whitish gases are seen to ascend the tube and to escape at the top. They closely resemble the vapours obtained when the candle flame is suddenly blown out, and consist of vaporised paraffin mixed with some products of combustion. If the tube is warm and not too long the gases, upon ignition, will continue to burn at the top, affording a replica of the candle flame below. The temperature near the apex of this zone approaches 1000 C. (2) Surrounding the inner, non-luminous zone is a luminous portion known as the luminous mantle or zone. Here the temperature ranges from 1000 to 1300 C. Chemical change has now set in, to be completed in (3) The non-luminous outer mantle, where the carbon and hydrogen are completely oxidised in excess of air to carbon dioxide and water. This is the hottest part of the flame and it is in this mantle that the wick, already y partially carbonised, oxidises away com- pletely and thus does not require to be snuffed, as it would do if it went straight up into the main portion of the luminous mantle, where it would interfere with the normal burning of the candle. (4) A small blue zone is usually distinguishable just beneath the wick. CAUSES OF LUMINOSITY 23 Causes of Luminosity. The question now arises as to what actually causes the luminosity of the zone immediately within the non-luminous outer mantle of the candle flame. Before attempting to draw any general conclusion, let us examine a few of the more prominent features of the luminous and non-luminous portions of the flame. (1) On introducing a cold surface of porcelain into the outer non-lum.inous mantle no visible change occurs, but if once this mantle is penetrated and the luminous zone allowed to impinge upon the cold porcelain, the latter immediately becomes coated with a deposit of soot. There is thus a distinct difference between the two portions of the flame, the luminous portion behaving exactly as if it contained in suspension small particles of soot or carbon at white heat. On the other hand, a precisely similar deposit would be obtained by the decom- position of heated, dense hydrocarbons under like treatment. Hence this experimental result is capable of a double interpretation. (2) If a jet of hydrogen issuing from a glass tube is ignited, the flame is at first almost colourless, but as the tube becomes warm the hydrogen flame becomes yellow in conse- quence of sodium escaping from the glass. If, however, the hydrogen be made to issue 24 FLAME from a platinum tube and to burn in a pure, dust-free air, the flame is invisible and colourless. It may, however, be rendered luminous in at least two ways, namely, either by the introduction of solids, or simply by the purely physical act of in- . creasing the pressure. The increased luminosity consequent upon introduction of solids into the flame is utilised commercially in the incandescent gas mantles, the introduction of which entirely revolutionised the methods of illuminating private and public buildings. On the other hand, the fact that mere increase of pressure on the hydrogen flame will increase its luminosity shows that solids are not essential to produce this effect. (3) When the light from vapours and gases is passed through a spectroscope, a line spectrum is obtained. By increasing the pressure, however, the width of these lines increases, until ultimately a continuous spectrum is obtained, similar to that result- ing from an incandescent liquid or solid. A candle flame likewise shows a continuous spectrum, so that its illuminating con- stituents must either be solids, liquids, or highly dense vapours. x Conclusion. It is now evident that the early y theory of Davy (1815), namely, that the luminosity of a candle flame is due to the separation and raising THE COAL-GAS F^AME 25 to white heat of solid particles of carbon by incom- plete combustion, is not the only explanation that can be offered. For many years Davy's theory was generally accepted, but during the latter half of last century opinion inclined towards Frankland's view (1867) that the luminosity is due to the presence of dense gases 2 produced in the flame as the result of incomplete combustion; and con- sideration of the above-mentioned observations shows that this theory also is quite in accordance with known facts. It may well be that a complete explanation may include both of these views, and it is by no means impossible that the luminescent substances are in the colloidal state. 20 The Coal-Gas Flame. When coal-gas issuing from a jet is ignited, the resulting flame exhibits the same three prominent zones that characterise the candle flame. A cold porcelain basin becomes covered with soot in the luminous zone, where combustion is partial only, but undergoes no apparent change in the outer mantle. The inner, non-luminous zone consists merely of unburnt gas, and can be abstracted with a small tube precisely as in the case of the candle. Since the luminous zone owes its light-giving power to the presence of solids or of dense hydro- carbons at a high temperature formed through partial combustion, the question arises as to what would happen if fresh air could be injected straight into this portion of the flame. It is to be expected that combustion would by this means be greatly acceler- 26 FLAME ated and the life-period of any luminous particles rendered so short that the luminosity of the flame would be greatly reduced, if not, indeed, made to disappear entirely. .AIR GAS FIG. 4. A Model Bunsen Burner. Such in general is the case, as is well illustrated by the Bunsen burner, a useful form of which, for demonstration purposes, consists of a short piece of combustion tubing, some 6 or 8 in. in length, and a piece of narrower tubing bent at right angles and connected with the gas supply (Fig. 4). THE COAL-GAS FLAME 27 A cursory examination of the flame shows that it consists of two parts only, namely, an inner zone of unburnt gas and an outer, also non-luminous mantle in which vigorous combustion is taking place. The inner portion is cool, so cool that a match head may be thrust into it and kept there unburned, whilst the wooden stem ignites at the junction of the inner and outer zones. A piece of paper held momentarily in a plane at right angles to the mouth of the burner at the level A becomes charred in a ring, the inner portion remaining unaltered. At B, above the apex of the inner zone, the paper is charred in a disc. Similar results are obtained with wire gauze, a red hot ring being formed at A and a red hot disc at B. The question of the various temperatures attained in a Bunsen flame has been the subject of some little discussion, and published figures show a remarkably wide variation. The data shown in Fig. 5 are probably fairly accurate. 3 If, whilst the flame is burning, the gas supply be gradually reduced, the flame becomes smaller and smaller, begins to flicker and darts down the tube, and we find an ordinary gas flame now burning at the end of the bent tube. This is called striking back. The mixture of air and coal-gas passing up the tube of a Bunsen flame in normal operation is explosive and the flame tends to strike back. Opposing that action is the cooling due to the tube and also the velocity of the mixed gases up the tube. By reducing this latter, however, a point is 28 FLAME reached when the threatened explosion does take place and the flame rushes back but obviously only so far as the explosive mixture extends, namely to the supply tube. -I550C. -IOOOC. 300C. 4* FIG. 5. The Temperature of the Bunsen Flame. SmiihelU Separator* By an ingenious arrange- ment due to Smithells, it is possible to catch the flame in the act of striking back. Two pieces of THE COAL-GAS FLAME combustion tubing are taken, about 8 and 10 in. in length respectively. The longer piece has an external diameter of about 0-2 in. less than the shorter, and is attached to the chimney of a Bunsen burner (Fig. 6). COMPLETE COMBUSTION CO,CO a ,H a .H 2 PARTIAL COMBUSTION BUNSEN CHIMNEY FIG. 6. Smithells' Separator. By wrapping a little paper or a rubber ring round the narrower tube the outer one may be loosely attached concentrically. On lighting the gas at the top a normal Bunsen flame is produced. On reducing the pressure of the gas, the flame is seen to flicker as before and part passes down the outer tube until it reaches the inner one, up which the gases are passing with a higher velocity in consequence of its narrower diameter. Here the flame stops. We 30 FLAME have now apparently two flames. In the lower one partial combustion takes place, and a mixture of water, hydrogen, and the oxides of carbon passes to the top where combustion to water and carbon dioxide is completed. If a piece of wire gauze is gradually brought down on to a Bunsen flame from above, it will be seen that the flame appears to be pressed down, and may even be extinguished if the gauze is brought as low as the top of the chimney. This is because the wire mesh- work rapidly conducts the heat away from the flame, --,- -*- -^ - GAUZE FIG. 7. cooling the combustible gases on their passage through the gauze to a temperature below their ignition point. If the gauze is held stationary across the flame for a few seconds, it becomes red hot, as we have already seen, and presently the gas above ignites. The heated wires do not now reduce the temperature of the penetrating gases. If, on the other hand, the wire gauze is laid on the top of the Bunsen chimney, and the gas above it ignited, the gauze may be raised some distance carrying the flame with it. As the heated gases tend to rise and are already above the gauze, the latter keeps cool and there is little tendency for the INFLUENCE OF PRESSURE 31 flame to travel down. This fact is made use of in the household incandescent burners, some gauze invariably covering the pipe through which the air and gas are supplied, in order to prevent striking back. This is the principle of the Davy Safety Lamp, for use in mines and other places where the escape of combustible gases renders it dangerous to carry naked lights. Numerous modifications have been introduced from time to time in the lamp, but the principle remains the same. An oil fed wick is surrounded by a glass cylinder, the upper part of which is closed by a cylinder of wire gauze through which fresh air for combustion passes inwards and burned air outwards through the upper portions. Should any marsh gas or fire-damp be present it exerts a peculiar influence on the flame. Small quantities cause a flickering, whilst with larger quantities the flame becomes increasingly elongated. The combustion, however, will not ordinarily pass from the inside of the lamp through the wire gauze, so that the danger of explosions is greatly minimised. Influence of Pressure on Luminosity. In 1859 Frankland 2 burned six candles at Chamounix and found a loss in weight of 94 grams per candle per hour. The same candles were burned on the summit of Mount Blanc and were found to lose 9-2 grams on the average. Hence the difference in pressure had not materially affected the rate of combustion. Frankland observed, however, that the FLAME inner, non-luminous zone of the flame on the mountain top was larger than in the valley and the luminosity appreciably less. On returning to England he carried out a series of photometric measurements on the influence of pressure on the luminosity of a candle flame, and was able to deduce the following law: The diminution of illuminating power is directly proportional to the diminution of atmospheric pressure. For every fall of 1 in. of the mercury barometer, the luminosity falls by 5-1 per cent. Hence, taking that at London as the standard, namely 100, at Miinchen it would be 91, and at Mexico 61-5. Frankland next showed that the law was trust- worthy even up to a pressure of three atmospheres, but at still higher pressures the luminosity rapidly increased, an observation attributable to less complete combustion. His results were as follows : Pressure in Atmospheres. Observed Luminosity. Calculated Luminosity. 1 2 3 100 263-7 406 100 253 406 4 959 559 Even an alcohol flame becomes very luminous under a pressure of four atmospheres. INFLUENCE OF TEMPERATURE 33 Influence of Temperature on Luminosity. As a general rule, increase of temperature causes a simultaneous increase in the luminosity. If a flame is cooled by bringing near to it a block of cold metal, a distinct reduction in luminosity may be made apparent. On the other hand, by warming coal-gas as it passes through the chimney of a Bunsen burner, the normally non-luminous flame may be made decidedly luminous. Causes of the Decreased Luminosity of the Bunsen Flame. The question now arises as to why the Bunsen flame should be non-luminous. This arises from several causes. 5 First, the injection of oxygen into the heart of the flame causes the rapid, complete combustion of the coal-gas as already indicated. This, however, is only part of the explanation. Second, the dilution of the flame with atmospheric nitrogen tends to prevent the formation of inter- mediate luminous products by increasing the tempera- ture necessary to bring about the necessary partial decomposition of the hydrocarbons. Third, the air introduced into the flame is cold, and thus tends to reduce the effective heating influence brought about by the oxidation consequent upon the admission of atmospheric oxygen. FLAME Reciprocal Combustion. It must not be overlooked that a flame such as we have been considering is simply a boundary between the combustible gases where chemical com- bination is proceeding rapidly. It is usual to regard FIG. 8. Coal-Gas burning in Air, and Air burning in Coal-Gas. coal-gas as combustible, but this is simply a matter of convenience, for coal-gas is not combustible in so far as nitrogen is concerned, and oxygen is combustible where coal-gas is concerned. That a o flame of coal-gas burning in air is simply the inverse of air burning in coal-gas may be very effectively shown by the experiment illustrated in Fig. 8. The coal-gas supply is turned on full and ignited RECIPROCAL COMBUSTION 35 at the top of the lamp glass, which may be protected by wire gauze at A to prevent cracking. The gas is turned down slightly to create a draught up B and a lighted taper is carefully passed up. A flame appears at C as soon as the gas is reached air burning in coal-gas. 6 SECTION V. THE COMBUSTION OF GASEOUS HYDROCARBONS AND OTHER GASES. THE problems connected with the combustion of gaseous hydrocarbons have been the subject of considerable controversy. During the greater part ^ of last century the theory of preferential combustion was widely accepted, according to which there is competition between the different constituents of the burning gases for the oxygen of the air. If, therefore, a hydrocarbon gas undergoes partial com- bustion in a limited supply of air, the " most favoured" element will tend to burn first, leaving the remainder to oxidise as best it may. In this way the luminosity of a hydrocarbon flame, such as that of ethylene or acetylene, received explana- tion. 1 The gas is first decomposed into hydrogen y and carbon at the high temperature of the flame. The hydrogen being under these conditions the favoured element rapidly burns to steam, whilst the less favoured carbon remains suspended in the flame in a white hot condition, thereby rendering it luminous. Ultimately the carbon itself burns, and the process of combustion is thus completed. COMBUSTION OF HYDROCARBONS 37 On the other hand Dalton 2 had found, in the early years of last century, that marsh gas, exploded with its own volume of oxygen, yields equal volumes of steam, carbon monoxide, and hydrogen. Thus : Similarly, ethylene yields carbon monoxide and hydrogen : C 2 H 4 These results received support from the researches of Kersten 3 in 1861, and, in later years, of many other investigators. They are of particular interest in that they prove hydrogen to be no longer the " favoured " element under explosive conditions. Kersten 4 sought to explain the phenomena on the assumption that when the hydrocarbon has been decomposed by the heat of the flame into hydrogen and carbon, the latter is preferentially oxidised to carbon monoxide, after which any excess of oxygen distributes itself between this gas and the hydrogen. Assuming the theory of preferential combustion to be true, the question at once arises as to what factors determine whether or not a given element shall be the more " favoured." Clearly the chemical nature of the element cannot be the sole deciding factor, otherwise it is not clear why hydrogen should in certain cases be more favoured than carbon, and in other cases less so. Although the theory of preferential combustion has not been definitely disproved, a more satisfactory 38 COMBUSTION OF HYDROCARBONS explanation of the foregoing and of other phenomena of combustion is afforded by what may be termed the Association theory, according to which the oxygen of the air first combines with the hydro- carbons forming more or less unstable hydroxylated products which ultimately, in a sufficiency of air or oxygen, decompose to carbon dioxide and water. These conclusions have been arrived at by Bone and his collaborators mainly as the result of an extensive series of researches on the slow combustion of methane at temperatures ranging from 300 C. to 510 C., and receive ample support from the behaviour of mixtures of oxygen and other hydrocarbon gases under analogous conditions. The initial experiments 5 were conducted with various mixtures of gases contained in sealed bulbs of boro-silicate glass at temperatures between 300 C. and 400 C., but as the volume of gases capable of being dealt with in this manner was limited, namely about 70 c.c., and as, moreover, such a method was not adapted for the detection and isolation of transient intermediate products, later experiments 6 were conducted with the aid of a different type of apparatus. This con- sisted of a combustion tube filled with fragments of ignited porous porcelain through which some 1200 c.c. of the reacting gases were continuously circulated by x means of a Sprengel pump which worked auto- matically. The experiments were protracted, in some cases extending over several weeks, during which times the following facts were definitely established : 1. None of the reactions a to f indicated below COMBUSTION OF HYDROCARBONS 39 takes place to any appreciable extent at temperatures of 400 C. downwards. (a) C+H 9 O (6) (c) O/) 2CO + O 9 - 2CO 9 0) Also, the following pairs of gases were found to have no appreciable mutual action at or below 400 C,, namely, CH 4 +C0 2 , CH 4 + H 9 0, and CO + H 8 . This greatly simplifies the study for, when once the methane has begun to oxidise, an indefinite number of secondary reactions might be expected to take place if the foregoing reactions were capable of proceeding with an appreciable velocity. 2. Between 300 and 400 C. methane is oxidised with comparative rapidity by oxygen gas. When insufficient oxygen is present to completely oxidise the methane, the final products are water, carbon monoxide, and carbon dioxide. Neithe'r free hydrogen nor free carbon is produced in detectable quantities. Since, if once formed, it would be impossible for them to be oxidised away in accordance with schemes 6, c, or /, it follows that their detection and isolation would be an easy matter, and hence it may be postulated that under normal conditions of slow combustion the methane is not first dissoci- ated into carbon and hydrogen. It seems equally clear, moreover, that the carbon monoxide and water formed are two of the primary disintegration products 40 COMBUSTION OF HYDROCARBONS of the partial oxidation of the methane molecule at these temperatures as these are too low for reaction / to take place. 3. A large proportion of carbon dioxide is frequently formed, sometimes almost equal in volume to the carbon monoxide produced. Since the conditions preclude all possibility of its forma- tion by oxidation of the monoxide either directly with oxygen or through interaction with steam (reactions d and e above), it would appear to be a disintegration product of some more complex oxygenated molecule. 4. Formaldehyde, H.CHO, is formed during the slow oxidation of methane at 450 to 500 C. and can be detected as a transient intermediate product. Since it is not produced when mixtures of moist carbon monoxide and hydrogen are continuously circulated for two days over a hot surface of porous porcelain at 460 to 480 C., 6 ' 7 it seems reasonable to suppose that it is one of the products of the slow oxidation of methane, and is not produced by minor secondary reactions. When heated, in the absence of air or oxygen, it readily decomposes into hydrogen and carbon monoxide : H.CHO = H 2 + CO, but there is no evidence that the reaction is revers- ible, although under the influence of a silent electric discharge formaldehyde may be obtained from hydrogen and carbon monoxide. 8 In the presence of air it yields carbon dioxide and water. Piecing all this evidence together, Bone and COMBUSTION OF METHANE 41 Wheeler in 1903 came to the conclusion that the slow combustion of methane takes place in several stages involving the formation and subsequent decomposition of formaldehyde, with the final pro- duction of carbon dioxide and water if sufficient oxygen is present. Armstrong, 9 however, suggested that the real initial product is not formaldehyde but methyl alcohol, which rapidly decomposes to formaldehyde and steam. This view is accepted by Bone. 10 These changes may be represented schematically as follows : H H H H.C. H - > H.C. OH -+ H.C. OH H Methane. H Methyl alcohol. OH Hypothetical dihydroxy. methane. H + H 2 O Formaldehyde + Steam. OH OH > H . C : O > HO . C : O Formic acid. Carbonic acid. CO 2 + H 2 O The methyl alcohol, which, on account of its easy oxidation cannot be experimentally detected, is assumed to undergo hydroxylation to hypothetical dihydroxymethane. This instantly decomposes to formaldehyde, which, with its next hydroxylation product, namely formic acid, can easily be detected amongst the products of the slow oxidation of methane. The final stage of the oxidation is reached with . the hydroxylation of formic acid to carbonic 42 COMBUSTION OF HYDROCARBONS acid, which immediately undergoes fission into carbon dioxide and water. With ethane, C 2 H 6 , the mechanism of slow com- bustion is believed to proceed as follows u : CH CH CH Q CH OH CHo CH Q . CHO Lq V-/J.J.O V^AJlo V/** > _ CH(OH) 2 H 2 Ethane. Ethyl [Dihydroxy Acetaldehyde alcohol. ethane.] + Steam. H.CHO > + >HCOOH >CO(OH) 2 CO + H 2 O Formic Carbonic Formaldehyde, carbon monoxide, Y C0 2 + H 2 The presence of ethyl alcohol as the primary product of oxidation of ethane has not been deter- mined experimentally, for alcohol is, under these conditions oxidised far more rapidly than ethane itself. Ethyl alcohol has, however, been detected among the products of the interaction of ethane and ozone at 100 C., and it seems highly probable, there- fore, that ethyl alcohol is really the primary product of oxidation as indicated above. 12 Hydrogen, methane, and ethylene are sometimes 13 found amongst the products of oxidation, without, however, any carbon being liberated. Their appear- ance is due to the purely thermal decomposition of ethane, formaldehyde, and acetaldehyde. 14 Thus : H.CHO - H + CO CH 3 .CHO = CH 4 COMBUSTION OF CARBON MONOXIDE 43 In the case of ethylene, 15 the reactions suggested are as follows : CH 2 CH 2 CHOH -> || -> || --> 2H.CHO CH 2 CH(OH) CHOH Formaldehyde. Ethylene. Vinyl Dihydroxy alcohol. ethylene (hypothetical). --> H . COOH > CO(OH) 2 Formic Carbonic acid. acid. C0 2 +H 2 Vinyl alcohol could not, of course, be experi- mentally detected among the products. Acetylene appears to react as follows l6 : CH C(OH) C(OH) H.CHO CH CH C(OH) CO Acetylene. Formaldehyde + Carbon monoxide. -> H.COOH -^ CO(OH) 2 ' > CO 2 + H 2 O Formic Carbonic acid. acid. It seems reasonable to conclude that during rapid combustion of the hydrocarbon gases, closely similar reactions obtain. Combustion of Carbon Monoxide. Lavoisier demonstrated the general principle that the increase in weight of the products of combustion 44 COMBUSTION OF HYDROCARBONS in air is equal to the weight of oxygen consumed. From this date until 1880, chemists regarded the combustion of carbon monoxide as a simple straight- forward chemical change, represented by the equation 2CO + O 2 = 2CO 2 . In his address to the Chemical Section of the British Association in 1880, however, Dixon 17 pointed out that dry carbon monoxide and dry oxygen do not unite. A trace of water must be present. Neither dry carbon dioxide, nitrogen, nor cyanogen was found to have any effect, but such gases as hydrogen, hydrogen sulphide, ether vapour, ethylene, ammonia, etc., act in a similar manner to water vapour. In other words, all substances that will form steam under the conditions of the experi- ment are capable of determining the combustion. Dixon, therefore, suggested that " in the ordinary combustion of carbonic oxide," the steam present acts the part of a " carrier of oxygen " by undergoing reductions and successive re-formations : (1) CO + H 2 _ CO, + H 2 (2) 2 Traube 18 disputed the generalisation. He pointed out that the reaction is reversible, and at the temperature of the electric spark proceeds from right to left. He therefore COMBUSTION OF CARBON MONOXIDE 45 argued that Dixon's hypothesis must be incorrect. 19 Having observed that hydrogen peroxide is obtained when moist carbon monoxide is burned, Traube concluded that the reactions were as follows : CO 2 + H 2 O 2 It has also been suggested 20 that percarbonic acid, H 2 C 2 O 6 , is formed as an intermediate product, and this is supported by the fact that a flame of carbon monoxide impinging on water containing a little potassium hydroxide and cobalt chloride gives a precipitate identical with that obtained with the same mixture on addition of potassium percarbonate. But this test alone is not conclusive since potassium hydrogen carbonate gives the same precipitate with hydrogen peroxide in the presence of cobalt chloride. 21 Bone, 24 in discussing the influence of hydrogen upon the rate of explosion of mixtures of carbon monoxide and oxygen, suggests that carbon monoxide is in- capable of combining directly with molecular oxygen. The hydrogen serves to induce the formation of either nascent oxygen or ''activated" steam molecules, both of which can oxidise the monoxide. It is assumed that oxygen molecules unite in the flame each with four atoms of hydrogen to yield the un- stable dihydrol complex H \ / H >o=o/ H/ \H which instantly dissociates into a mixture of free hydrogen molecules, nascent oxygen, and nascent or 46 COMBUSTION OF HYDROCARBONS "activated" steam. The carbon monoxide is then oxidised as follows : CO + : O = CO 2 CO + :OH 2 = CO 2 + H 2 Bone does not apply this to a direct consideration of the influence of steam upon the oxidation of carbon monoxide, but since steam dissociates at the tempera- ture of flames into hydrogen and oxygen it is easy to see how the above reactions could take place and the oxidation of carbon monoxide be thereby facilitated. The Combustion of Cyanogen. Cyanogen burns in a Smithells' separator with characteristic beauty. 22 The inner flame is red, the cyanogen burning to carbon monoxide. Thus : C 2 N 2 + 2 = 2CO + N 2 . The outer flame is blue, the carbon monoxide now undergoing further oxidation to the dioxide. The Combustion of Hydrogen. It is a remarkable fact that although hydrogen and oxygen unite with explosive violence when a spark is passed into the mixed gases, yet the pure gases dried over phosphorus pentoxide do not explode, even on heating to redness. 23 Even at the melting-point of silver (960 C.) no combination takes place. This is a typical example of the curious fact that, as the methods of preparing pure substances become increasingly refined, the number of reactions known to occur between the impure reagents, but not between them or only with excessive slowness when in a state of high purity, is steadily increasing. SECTION VI. IGNITION TEMPERATURES. THE ignition temperature has already been defined (see p. 6) as that temperature at which rapid com- bustion becomes independent of external supplies of heat. The temperature of ignition is the temperature at which the heat evolved by the reaction just equals and therefore counterbalances the loss of heat consequent upon radiation, etc. A clear idea l may be obtained by imagining a combustible mixture of gases to issue from an orifice into an inert atmosphere. 2 If the orifice is sur- rounded by a ring of platinum wire which is being raised in temperature by passage of an electric current, a flame will gradually make its appearance. If, as soon as this is observed, the heating of the wire by the electric current be discontinued, the flame will disappear. It is not self-supporting, but depends upon the accessory supply of heat from the electrically heated wire. If now we raise the ring to a still higher temperature than before, a brighter flame is obtained in consequence of the increased rate of chemical action ; and finally a temperature is reached at which the flame will continue to burn without the aid of any heat from the wire, so that the current may be cut off. This is the temperature of ignition. 47 48 IGNITION TEMPERATURES Attention has already been directed to the phenomena of spontaneous combustion. When, for example, liquid phosphoretted hydrogen, P 9 H 4 , or zinc ethyl, Zn(C 2 H 5 ) 2 , is allowed to come into contact with air, it spontaneously bursts into flame. The ignition temperatures of these bodies lies below the usual temperature of the atmosphere. The ignition temperature of pure gaseous phosphoretted hydrogen, PH 3 , lies below 100C., and a sample of the gas prepared by the action of alcoholic potash upon phosphorus if allowed to escape from a jet, may be ignited by causing it to impinge upon a test-tube of boiling water. The vapour of carbon disulphide ignites at a slightly higher temperature, namely, at about 120 C. If, therefore, a hot glass rod is intro- duced into a beaker containing a few drops of carbon disulphide, the vapour immediately inflames. Hydrogen ignites in air more readily than coal- gas, and this is readily shown by allowing the latter to impinge upon a warmed bundle of platinised asbestos, when the latter becomes red hot, but does not usually effect the ignition of the gas. 13 If, how- ever, the coal-gas be replaced by hydrogen, the latter immediately inflames. The earliest systematic attempts to determine the ignition temperatures of various gases were those of Davy, 3 who observed that hydrogen could be inflamed at the "lowest visible heat of iron," that is approximately 500 C. Ethylene and carbon monoxide inflamed at red heat (probably c. 700 C.), whilst methane required contact with an iron rod in brilliant combustion. Modern methods of deter- IGNITION TEMPERATURES 49 mining ignition temperatures may be divided into five main groups, namely : I. The combustible mixture of gases is contained in glass bulbs and raised in temperature by immersing in a heated bath. 4 Under these conditions the gases are fired under pressures usually somewhat greater than atmospheric, and in contact with a relatively large superficial area of glass which may act catalytically upon the reaction. This method has also been adopted for slow combustion experiments, but, in consequence of the catalytic action just referred to, different in- vestigators have frequently obtained widely different/ results. Thus, for example, with electrolytic gas, / Meyer and Raum 4 made the following observations : / ^ Temperature. 100 C . . No water detectable after 218 days. 300 . . Water detected in 65 days. .350 . . Water detected in 5 days. 448 . . Perceptible action, but very slow. On the other hand, Bone and Wheeler 4 have - succeeded in keeping electrolytic gas ^in glass bulbs at 400 C. for seven days without observing any combination of the gases, although in some of their experiments the presence of water could be detected. It would thus appear that 400 C. is the border-line for the slow combustion of hydrogen and oxygen under these conditions within finite time. II. A second method consists in passing a stream v of the combustible mixture of gases through a tube which is gradually heated up until the gases ignite. 5 This method has the advantage of yielding results at the normal pressure of the atmosphere, but is still D 50 IGNITION TEMPERATURES liable to be influenced by the catalytic action of the walls of the tube. Further, the gases are not at rest, but in motion. III. A particularly valuable method is that devised by Dixon 6 in 1903 and shown diagrammatically in Fig. 9. f v PORCELAIN TUBE / ELECTRICALLY '1 /> HEATED '/ V 1 fl % 'J A I c / 7 xx ^A V /. \ T * '/ '/ 7. / y t / 7, i I V / / KS- r* . "5 > . FIG. 9. Dixon's Apparatus for determining Ignition Temperatures of Gases. The principle of this method consists in heating the combustible gases separately to such a tempera- ture that they immediately inflame upon coming into contact. The combustible gas was admitted to the apparatus at A and the atmosphere of air or IGNITION TEMPERATURES 51 oxygen at B. During passage up the tubes the gases attained the temperature registered by the thermocouple in T and began to mix at C, from which orifice the combustible gas issued. The porcelain tube, D, 4-5 cms. in diameter, surrounded by platinum wire and asbestos packing, was electri- cally heated, the temperature being allowed to rise at the rate of 5 C. per minute, until the issuing gas at C. inflamed in contact with the air. This method, like the previous one, gives results at atmospheric pressure, but possesses the further advantages that the mixed gases are only momentarily in contact with a possible catalyst, namely, the small orifice at C, Experiment showed that, on working with different furnace tubes, a constant ignition-point was obtained when both the diameter of the outer tube and the rate of passage of the combustible gas through the orifice were made to exceed a certain minimum value. The following results were obtained.: Ignition Temperature, C. Oxygen. Air. Hydrogen 580 to 590 580 to 590 Carbon monoxide (moist) Methane 637 556 658 700 644 650 , 658 , 750 Ethane . 520 630 520 , 630 Propane 490 570 Ethylene 500 519 542 , 547 Acetylene 416 440 406 , 440 Cyanogen Hydrogen sul phid( i 803 220 818 235 850 346 , 862 , 379 Ammonia 700 860 52 IGNITION TEMPERATURES It will be observed that in the majority of cases the ignition temperature in air is practically identical with that in oxygen. IV. Adiabatic compression. 7 As is well known, when gases are compressed, heat is evolved, and if the rate of compression is sufficiently rapid to prevent the undue escape of heat, the gases will rise in temperature. By suitably arranging the apparatus the gases may be fired, and if the volume of the gases at the moment of firing can be determined, their temperature the ignition temperature under compression admits of calculation from the well- known equation T representing the absolute temperature, V the volume, and y the mean ratio of the specific heats of the gaseous mixture at constant pressure and volume. The essential features of the apparatus used by Dixon and Crofts 7 are shown in outline in Fig. 10. The body of the apparatus consisted of a steel cylinder 56 cms. in length and^ 11 cms. in diameter, with a cylindrical cavity 3-02 cms. in diameter through its axis. The lower end of this cavity was opened out and threaded to enable a plate to be inserted to close the bottom of the cylinder and to be kept in place by a powerful screw. The combust- ible mixture of gases was introduced into the apparatus through a tube A, and the upper end of the cylinder was closed with a plug attached to a piston, bearing a head, H, of great strength, upon IGNITION TEMPERATURES 53 which an iron block, weighing 76 kg. (2J cwts.), was allowed to fall from a height of about 1-5 metres. STEEL WASHERS FIG. 10. Dixon and Crofts' Apparatus for determining Ignition Temperatures by Compression. This sufficed to drive the plug well down the cylinder and so to compress the gases. The extent 54 IGNITION TEMPERATURES of compression was regulated by a series of steel washers which could be arranged to catch the piston head at any desired position during its descent. 8 The initial and final volumes being thus known, the temperature after compression was calculated by means of tl\e equation given above, y being taken as 1-4. The temperature of ignition was found by " trial and error," the compression being increased or decreased in successive trials until ignition just took place. The following results were obtained : Relative Volumes of the Gases. Ignition Temperature, C. 2H 2 + 2 2H 2 + 2O 9 2H 2 +8O 2 2H 2 +16O 2 2H 2 +32O 2 526 511 478 472 never exploded These results are not, of course, strictly comparable with those obtained by the previous method, owing to the high pressures developed, these being of the order of 30 atmospheres. The temperatures obtained are appreciably lower than that found for the ignition of hydrogen in oxygen by Method III. V. The Soap Bubble Method. This exceedingly ingenious device adopted by M'David 9 in 1917 consists in blowing soap bubbles with the combust- ible gaseous mixture and allowing them to impinge upon an electrically heated wire, the temperature of which is raised until the gases ignite. The FLASH-POINT 55 results obtained by this method are high and are regarded as not very reliable. 10 In the following table are given the results obtained by M'David, and these where possible are compared with those obtained by Dixon and Coward by Method III. Gaseous Mixture. M'David. Dixon and Coward. Hydrogen-air . 747 580 to 590 Carbon monoxide-air 931 637 658 Ethvlene-air 1000 500 519 Coal-gas-air 878 ... Benzene-air 1062 ... Ether-air . 1033 Petrol-air . 995 ... It will be seen that not only does a wide difference exist between the different sets of results, but the temperatures do not follow the same relative order. Thus, M'David finds the ethylene to have a higher ignition temperature than either hydrogen or carbon monoxide, whilst Dixon and Coward found it to be lower. In the table on pp. 56 and 57 are given the more important results obtained by the various methods described for the ignition temperatures of gases. The temperature at which the vapour of a liquid forms an inflammable mixture with air is frequently termed its flash-point, and its accurate determination is often a matter of considerable legal importance, more particularly in the case of low flash-point paraffin oils sold for domestic illuminating purposes. 56 IGNITION TEMPERATURES Gases. Ignition Temperature, C. Method. Authority and Date. Hydrogen-oxygen . 518 to 606 I Krause and Meyer (1891). 5*9 I Emich(1900). 650 to 730 II Meyer and Kreyer (1892). 620 680 II Meyer and Munch (1893). 550 II Mallard and Le Chatelier (1880). 653 to 710 II Bodenstein(1899). 674 II Mitscherlich (1893). 840 II Gautier and Helier (1896). 845 II Helier (1897). 580 to 590 III Dixon and Coward (1909). 2H 2 + 2 . 2H 2 + 2O, 540 514 IV IV Falk (1906). 99 99 2H 2 + 4O 2 530 IV 2H 2 + O 2 . 526 IV Dixon and Crofts (1914). 2H 2 + 2O 2 511 IV 99 99 2H 2 + 8O 2 478 IV 99 99 Hydrogen-air . 580 to 590 III Dixon and Coward (1909). 747 V M'David(1917). Methane-oxygen . 606 to 650 I Meyer and Freyer (1893). 650 730 II 99 99 656 678 II Meyer and Munch (1893). 556 700 III Dixon and Coward (1909). Methane air . 650 750 III 99 99 Ethane-oxygen 530 606 I Meyer and Freyer (1893). 606 650 11 99 99 605 622 Meyer and Munch (1893). 520 630 iii Dixon and Coward (1909). Ethane-air 520 630 in 99 99 Propane-oxygen 545 548 490 ,, 570 ii in Meyer and Munch (1893). Dixon and Coward (1909). Ethylene-oxygen 530 606 i Meyer and Freyer (1893). 606 650 ii 99 99 577 ,, 590 ii Meyer and Munch (1893). 500 519 in Dixon and Coward (1909). IGNITION TEMPERATURES 57 Gases. Ignition Temperature, Method. Authority and Date. Ethylene-air . 542 to 547 1000 Ill V Dixon and Coward (1909). M'David (1917). Acetylene-oxygen . 509 to 515 416 440 II III Meyer and Munch (1893). Dixon and Coward (1909). Acetylene-air . 406 ,, 440 III Propylene-oxygen . 497 511 II Meyer and Miinch (1893). Isobutane-oxygen . 545 550 II I s obuty len e-oxy ge n 537 548 II Coal-gas-oxygen 647 649 II > > Coal-gas-air 878 V M'David (1917). Benzene-air . 1062 V Ether-air 1033 V 9) Cyanogen-oxygen . 803 to 818 III Dixon and Coward (1909). Cyanogen-air . 850 862 III Carbon monoxide- oxygen 650 730 650 730 637 658 I II III Meyer and Freyer (1893). 5 Dixon and Coward (1909). Carbon monoxide- air 644 658 931 III V M'David (1917). Hydrogen sulphide- oxygen 250 to 270 315 320 220 235 I II III Meyer and Freyer (1893). Dixon'and Coward (1909). Hydrogen sulphide- air 346 379 III Ammonia-oxygen . 700 860 III > Hydrogen-chlorine . 240 270 430 440 I II Meyer and Freyer (1893). 58 IGNITION TEMPERATURES An elaborate and carefully standardised apparatus is then necessary. A determination of the flash-point of an oil is frequently desirable as a check upon its purity. Thus, for example, linseed oil has normally a flash- point of 250 C., but if adulterated with rosin oil, of usual flash-point 155 to 160C., the presence of this latter oil is readily detected in this way. This is a great advantage, because although rosin oil is denser than linseed, the density of the. mixture could easily be " doctored " by addition of some other lighter ingredient which would render gravity determinations useless. A convenient rough method of determining the flash-point consists in heating the oil in the inner pot of a glue pot, the outer one serving as an air- bath. A thermometer is inserted in the oil and a small gas jet made from a mouth blow- pipe is brought close to but not actually touching the surface of, the oil. The temperature is taken at which a small blue flame is seen to flash for the first time across the oil. This is the flash-point. In the following table are given the approximate flash- points of several well-known liquids : Liquid. Flash-point, C. Linseed oil . 250 Rosin oil 155 to 160 Paraffin illuminating o Is 38 50 Turpentine . S5 40 Rosin spirit. 35 40 Naphtha . 16 21 Methylated spirit 14 16 IGNITION TEMPERATURES 59 The ignition temperatures of a few substances that are solid at the ordinary temperature have been determined, and a few of the more interesting data are given in the accompanying table, together with the melting- and boiling-points of the substances concerned n : Substance. Meltinp-point, C. Ignition Temperature, C. Carbon Diamond c. 3500 800 to 850 Graphite n 690 Amorphous 345 Phosphorus, Yellow Red . 44-5 c. 60 255 to 260 Sulphur in air 12 ,, in oxygen 12 119-2 255 257 to 264 SECTION VII. THE INFLAMMATION OF GASEOUS MIXTURES. UPON introducing a source of heat into a combustible mixture of gases, two conditions must be satisfied in order to ensure propagation of the flame throughout the mixture. (a) The initial source of heat must be of sufficient volume, intensity, and duration to raise the adjacent layer of combustible gases to at least their ignition temperature. (6) The heat resulting from the combustion of this layer must be sufficient in turn to raise the next adjacent layer to its temperature of ignition, and so on. Any excess of one of the constituent gases over that required for complete combustion will serve as a diluent, and if the excess is very great the heat absorbed in raising the temperature of the mixture may be so large as to prevent the attainment of the ignition temperature. The largest quantity of the diluent which may be present and yet allow a flame to be propagated from layer to layer throughout the mixture without the continued presence of the original source of heat, is a measure of the limit of inflammation. The majority of the researches on 60 COMBUSTION OF AMMONIA VAPOUR 61 this branch of our subject have been carried out with gases that are combustible in air or oxygen, and it is to these that our attention will now be directed. An interesting example is afforded by ammonia vapour which readily burns in oxygen but not in ordinary air. If a taper is applied to ammonia vapour as it issues from a jet, the characteristic livid OXYGEN FIG. 11. Combustion of Ammonia Vapour in Air enriched with Oxygen. flame of the gas is seen burning side by side with the flame of the taper. But combustion at once ceases when the taper is withdrawn. If, however, the jet is surrounded by oxygen or air enriched with this gas, the ammonia may be made to burn con- tinuously even after the removal of the taper. This is conveniently shown by means of the apparatus figured above. 62 INFLAMMATION OF GASEOUS MIXTURES Concentrated ammonia solution is gently warmed in the flask and the escaping vapour is surrounded by air enriched with oxygen which enters the circum- scribing cylinder and is distributed round the central tube by the glass wool C. If, before passage of oxygen, a lighted taper is applied, the ammonia vapour is seen to burn alongside of the flame of the taper, but immediately the taper is removed com- bustion ceases. On adding increasing quantities of oxygen, the ammonia flame grows stronger until eventually the taper may be removed and the ammonia continues to burn without requiring any external source of heat. The minimum quantity of oxygen required to maintain this condition is termed the lower oxygen limit of inflammation. By the lower limit of inflammation is understood "the smallest quantity of any combustible gas which, when mixed with air or oxygen, will admit of this self-propagation of flame. But there is also a higher limit of inflammation, for clearly if the combustible gas is in excess, the excess will function as a diluent, and the higher limit thus becomes the lower oxygen or atmosphere limit. ^ Some of the earliest work on the subject was that of Davy 1 in 1816, whose experiments indicated a lower limit of inflammation of fire-damp in air as one part in 16 or 17, that is between 6-3 and 6-7 per cent. Modern researches have been carried out in three different ways, namely : 1. The gases are ignited by passing an electric spark between terminals placed at the centre of a large glass globe containing the combustible LIMITS OF INFLAMMATION 63 mixture. This was the method adopted by Burgess and Wheeler 2 in 1911, their apparatus being shown in Fig. 12. The globe had a capacity of approximately 2 litres and was fitted with platinum electrodes which entered the globe through ground stoppers. The method of experiment consisted of " trial and error," the proportion of combustible gas being successively reduced or increased until two mixtures were obtained, FIG. 12. Apparatus used by Burgess and Wheeler (191 1). one of which just enabled the flame to be propagated whilst the other did not. The authors 3 found that the lower limit mixture could be distinguished with certainty from that containing a slightly insufficient proportion of combustible gas, because the first sparking produced in the former case inflammation of all the gas in the globe, and on further sparking no further sign of combustion was manifest. In the latter case, on the other hand, although the flame might appear to travel pretty well through the globe with the first spark, a " cap " would appear upon 64 INFLAMMATION OF GASEOUS MIXTURES sparking a second time, showing that the whole of the combustible gas had not disappeared. 2. The gases may be electrically ignited in a horizontal tube closed at both ends. For this purpose a tube about 6 cms. in diameter is con- venient, the flame steadily creeping along the upper walls in some such manner as that indicated in Fig. 13, as the lower limit is approached. 4 FIG. 13. 3. A vertical tube may be employed, closed at both ends and ignited electrically either at the top or at the bottom. Method. Lower Limit. Methane per cent. Higher Limit. Methane per cent. 1. Globe . 5'6 14-8 2. Horizontal tube . 5 '4 flame travels 14'3 only along top of tube. 5 '6 methane all combusted. 3. Vertical tube : (a) Top ignition 6-0 13-4 (ft) Bottom ignition . not< 5-4 not > 14-8 As is to be anticipated, these methods do not all give quite the same result. Owing to convection currents, a low value for the lower limit and a high value for the higher limit are obtained by bottom gnition in a vertical tube, whilst top ignition yields LIMITS OF INFLAMMATION 65 a high value for the lower limit and a low value for the higher limit. These differences are well illus- trated by the results obtained by Burgess and Wheeler 4 for mixtures of methane and air. The lower limit mixture depends upon a variety of factors, the more important of which are as follows : (a) The calorific power of the gas, which may be designated as C. (b) The relative volume and specific heat of the diluent. (c) The ignition temperature of the combustible (d) The pressure. Increase of pressure raises the higher and lower limits, 5 as is evident from the following data for methane in air: Pressure, Lower Limit. Higher Limit. cm. Mercury. Methane per cent. Methane per cent. 76 6'0 13-0 125 6-05 13-15 290 6-2 13-6 465 6-4 14-05 (e) The temperature. Rise of temperature tends to reduce the lower limit value, as theoreti- cal considerations would lead us to expect. This is evident from the data given below, which refer to methane in air. 6 Temperature, C. . Lower Limit. ) Methane per cent. / 20 5'80 175 5-25 237 4-75 312 4-30 555 690 3*40 3-00 E 66 INFLAMMATION OF GASEOUS MIXTURES (/) In the case of the inflammation of gaseous mixtures in horizontal or vertical tubes, the diameter of the tubes exerts an important influence if less than about 5 cms. This is well illustrated by the data obtained for acetone 7 in air. Lower Limit. Higher Limit. Per cent. Acetone. Per cent. Acetone. Diameter of Tube. Hori- zontal. Down- ward. Upward. Hori- zontal. Down- ward. Upward. cm. 2-5 2-40 2-75 2-30 6-7 6-5 7-5 5-0 2'25 2'40 2-20 9'3 8-3 9-5 10-0 2-20 2-35 2-15 9-5 8-5 9-7 It is of interest to inquire whether or not the lower limit concentration of a combustible gas in air can be calculated, making due allowance for the above factors. 8 Assuming d and e to represent the average pressure and temperature of the atmosphere, and either /, the diameter of the tubes, to remain constant, or the values for the lower limits in the globe experiments, only, to be considered, it is evident that factor a, the calorific power of the combustible gas is the most important of the remaining factors in the majority of cases. Writing L as the proportion of combustible gas in a lower limit mixture, it is to be anticipated that L should be some function of 1/C. If strict proportionality obtains, which is very possible for analogous gases such as those of the saturated hydrocarbon series, then L oc 1/C or L = K/C. Taking methane as the LIMITS OF INFLAMMATION 67 standard, for which L = 5-6 and C = 189-1, we arrive at the value K = LC = 1059. Using this value for K, L may be calculated for the other saturated hydrocarbon gases from their calorific values determined by Thomsen. This is done in the following table 8 : Gas. Calorific Value. L determined by the Globe Method. L calculated. Methane 189-1 5-60 standard Ethane . 336-6 3-10 3-15 Propane 484-2 2-17 2-19 w-Butane 631-7 1-65 1-68 n-Pentane 779-2 1-37 1-36 Isopentane 779-2 1-32 1-36 The agreement is surprisingly close ; but when this method of calculation is applied to other combustible gases, such as hydrogen 9 (L = 10-0), using the above value for K, serious discrepancies arise. But this is to be anticipated, for there is no reason why L should be directly proportional to C an assumption upon which the calculated value for K rests. A few experiments have been carried out with a view to determining the composition of the atmo- sphere that extinguishes flames. It appears that the extinction of a flame is not determined only by the proportion of the "inert" gas and oxygen in the surrounding atmosphere. The nature of the inert gas also plays an important part in determining the lower oxygen limit. Thus, fbr example, carbon dioxide is found to exert a more powerful extinctive effect than nitrogen. 68 INFLAMMATION OF GASEOUS MIXTURES The composition of the extinctive atmosphere produced by a candle flame is as follows 10 : Oxygen Nitrogen Carbon dioxide Per cent. 15 to 16 80 81 3 This closely corresponds to the average composition of the air expired by human beings, and may be breathed by most people without producing any distinctly noticeable ill-effect. Other results of interest are 11 : Combustible Substance. Residual Air contains. Per cent. Candle burned until extinguished -j 4 to 6 CO 2 13 ,, 15 oxygen Alcohol burning on cotton-wool to \ extinction / 6-5 CO 2 11 oxygen Wood-charcoal glowing to extinc- \ tion / 8CO 2 9 oxygen Sulphur burning to extinction 13'5 oxygen Decrease of pressure raises the lower oxygen limit, as is evident from the following data 12 : Combustible Substance. Extinction percentage of Oxygen. Total Pressure of Gases. Ethyl alcohol burning from \ asbestos wick / 15-1 19-0 mm. 736-7 129 Candle . . { 16-1 19-9 736-7 91 SECTION VIII. PROPAGATION OF FLAME IN GASEOUS MIXTURES. A THOROUGH knowledge of the rate of propagation of flame in mixtures of air and various inflammable gases, especially hydrocarbons, is eminently desirable in view of the ever-present danger of serious fires and explosions in coal-mines. When an explosive mixture of gases contained in a horizontal tube closed at one end is ignited at the open end, it is observed l that : /^?f? i 1. A flame travels a certain distance along the tube with a uniform and relatively slow velocity. In the case of acetylene and air this distance is small; with methane and air it is relatively long. 2. Vibrations are gradually initiated which become increasingly intense, the flame moving backwards and forwards with oscil- lations of ever-increasing amplitude. 3. The flame either dies out or the remainder of the gas in the tube detonates, initiating what is termed the explosion wave. 70 PROPAGATION OF FLAME It is thus evident that a flame may be propagated throughout a gaseous mixture in two ways, namely : A. With a relatively slow motion, characterised by its uniformity, and B. With a rapid explosion wave, usually accom- panied by detonation. It is proposed to deal briefly with each of these in this section. A. Uniform Slow Movement in the Propagation of Flame in Gaseous Mixtures. The conditions most favourable for obtaining and preserving the initial uniform movement of flames are given 2 as follows : 1. The inflammable mixture should be contained in a long, straight, horizontal tube open at one end and closed at the other. 2. Ignition should be effected at the open end of the tube by a source of heat not greatly exceeding in temperature the ignition temperature of the mixture, and not productive of mechanical disturbance of the mixture. The speed of the uniform movement then depends upon four factors, to wit : (a) The diameter of the tube. (b) The material of which the tube is made. (This is negligible for tubes above a certain small diameter.) (c) The source of ignition. (d) The composition of the gaseous mixture. PROPAGATION OF FLAME 71 Given constancy of the first three factors, the flame speed of a given mixture may be regarded as a definite physical constant for that mixture. Method of Experiment. The method adopted by .Wheeler 3 and his co-workers for recording the speed of the flames and the general mode of procedure may be briefly summarised as follows : Tubes of different materials, but frequently of glass, measur- ing several metres in length, of diameter ranging from 2 mm. 4 to 9.6-5 cms. 5 (internal measure- ment) and open at both ends, are fixed horizontally in a straight line. The ends are flanged and ground to receive flanged end- pieces which are held in position by metal clips. Each end-piece is fitted with a wide-bore three- way tap. Glass-covered platinum electrodes reaching to the centre of the tube, leaving a spark gap of 3 mm. are fused 4 cms. from one end (E in the figure). Screen UJ< , 4 ._ Apparatus used by Wheeler and his co- workers - wires of metallic copper, 6 0-025 mm. in diameter, are threaded vertically across the tube through fine 72 PROPAGATION OF FLAME holes bored into the walls and afterwards sealed with adhesive material. In order to avoid any irregularity in the speed of the flame consequent upon the impetus that might be given by the igniting spark, the first screen wire, W 15 is fixed some 40 cms. from E, and other screen wires, W 2 , etc., are fixed 50, 100, 200, 300, and 400 cms. respectively from the first. It is convenient to employ an electrical method of recording the time of passage of flame along the , tube. An electric current is passed through the screen wires raising them to nearly red heat, so that they rapidly melt as soon as the flames touches them. The current is thus instantly interrupted and the time automatically recorded by the chronometer. This method appears to yield very uniform results. In beginning an experiment, the two end taps are opened and a rapid current of the desired mixture of gases passed through. The taps are now closed, and the left hand end-piece, A, is removed by sliding it downwards very gently in such a manner as not to disturb the quiescent gaseous mixture in the tube. A spark is passed through E, whereby the mixture is ignited at the now open end of the tube. In experimenting with gases yielding flames of sufficient actinic power to affect a photographic plate, photographic methods may be employed, a revolving drum bearing the film. In the case of the combustion of carbon disulphide with oxygen or nitric oxide the flames are highly actinic, and in consequence readily photographed. 7 It is possible also to obtain photographs in the case pf acetylene and air, and in this latter case it is a FLAME SPEED OF METHANE 73 particular advantage, as the uniform speed soon gives place to the explosion wave, and may only cover a matter of 20 cms. in the tube, so that the percentage error in the screen fusion method is abnormally great. By photography a permanent record is obtained, which can be examined at leisure, 100 17-5 CM. CM. 1-35 CM. 12 METHANE PER CENT FIG. 15. Velocity of Flame in various percentages of Methane Air Mixtures in Tubes of different Diameters (Parker). the rate of motion being readily determined by markings made on the photographic film with a tuning-fork. The size of the tube exerts an important influence upon the rate of propagation of the flame. 8 In the case of mixtures of methane and air, for example, 74 PROPAGATION OF FLAME if the diameter of the tube is small, say, of the order of 2-5 cms., the velocity of the flame is greatly retarded by the cooling effect of the walls. Where, on the other hand, the diameter is increased above 10 cms. convection currents tend to unduly accelerate the velocity of the flame, which now exhibits a 5 9 30-5 DIAMETER OF TUBE (CMS) 96-5 FIG. 16. Influence of Diameter of Tube upon Flame Speed (Mason and Wheeler). . turbulent front. The flame appears to possess a V swirling motion in a direction nearly perpendicular to the direction of translation of the flame front. This is caused by the rapid movement of the hot gases from below upwards by convection. In tubes of from 5 to 9 cms. in diameter this rapid movement is suppressed, although the shape of the flame front FLAME SPEEDS IN NARROW TUBES 75 indicates the existence of a definite movement of the hottest gases towards the upper part of the tube. Some of the results obtained by Parker and by Mason and Wheeler are given in Figs. 15 and 16, in the latter case the diameter of the tube being plotted against the flame speed. It will be observed that in the neighbourhood of -GAS FIG. 17. 9 cms. there is a slowing up in the influence of diameter upon flame speed. But this is only temporary, for further increase in the diameter shows that the maximum has not been reached even at 96-5 cms. There is, on the other hand, a lower limit to the size of the tube that will allow a flame to traverse it. If the tube is diminished in size, a point is reached at which the flame will traverse only a few centimetres, and with further reduction the flame 76 PROPAGATION OF FLAME will not pass along at all. The material of the tube X O in these cases plays an important part, metallic tubes being better conductors of heat, cool the flame, and prevent the passage of the flame more readily than glass. This may be conveniently demonstrated by connecting a slanting tube, as shown in Fig. 17, with the gas supply, and igniting the gas as it issues from the narrow tube at A. The cork at B is then removed and the gas simultaneously cut off. The flame, if tube A is not too small, will pass down and a slight explosion take place in C. By employing different tubes at A the influence of size and material can easily be demonstrated. This, of course, is the principle of the Davy Safety Lamp, to which reference has already been made, for wire gauze may be regarded as a series of thin slices of narrow tubes joined together transversely. Variation of Gaseous Mixture. When ignited under precisely similar experimental conditions, the velocity of the uniform movement of the flame rises as the percentage of combustible gas increases above its lower limit value, a maximum velocity being ultimately attained, after which further increase of the combustible gas effects a reduction in velocity until the higher limit concentration value is reached, when the flame ceases to pass. This is well illustrated by the curves in Fig. 18, which depict the results of several series of experi- FLAME SPEEDS OF METHANE 77 ments carried out with varying mixtures of methane, oxygen, and a neutral gas nitrogen to serve as diluent. 9 It will be observed that by increasing the proportion of oxygen not only does the concentration of methane at the lower limit fall slightly, but there is a very toor AiR 15-05 5 . 10 VOLUME OF C IN 100 VOLUMES OF 2 AND N 2 15 FIG. 18. Flame Speeds of Mixtures of Methane in Oxygen and Nitrogen (Mason and Wheeler). great increase in the upper limit concentration, pro- vided the proportion of oxygen in the oxygen- nitrogen mixture is less than about 25 per cent. With oxygen - nitrogen mixtures containing more than 25 per cent, oxygen, the lower limit of methane rises slightly. 100 78 PROPAGATION OF FLAME Further, the methane concentration yielding the maximum flame speed rises with the proportion of oxygen, and yields a greatly enhanced flame speed. Similar results are obtained with mixtures of methane and air, the latter being enriched with oxygen. 10 \IOO/0 0; 5 30 60 VOLUME OF CH 4 IN 100 VOLUMES OF 2 AND N 2 FIG. 19. Flame Speeds of Methane in Oxygen-enriched Air (Pay man). These results may be compared, in so far as their general characteristics are concerned, with those given in Fig. 18 for the speed of flames of methane mixed with atmospheres less rich in oxygen than ordinary air. 11 The curve showing the results for methane in pure FLAME SPEEDS OF GASES 79 oxygen is interesting. The maximum flame speed is 5500 cms. per second more than fifty times the l^ maximum speed attained in air. Further, the maximum speed in oxygen occurs with the mixture containing the two gases in the \^~ proportions necessary for complete combustion, namely, one volume of methane to two of oxygen. This is interesting in view of the observation that when a detonation wave is set up in mixtures of methane and oxygen, its speed is greatest when the ^ two gases are present in equal proportions, namely, CH 4 +0 2 . In the case of hydrogen and air, the measurement is less easy inasmuch as the period of uniform move- ment does not, in the majority of mixtures, extend over so great a distance as 1 metre. In contra- distinction to mixtures of methane and air, the maximum speed of propagation of the flame does not occur with the mixture containing hydrogen and oxygen in combining proportions, namely, 29-5 per cent, of hydrogen, but with mixtures over the range 38 to 45 per cent, of hydrogen. 12 In Fig. 20 are shown diagrammatically the relative speeds attained on firing mixtures of three saturated hydrocarbons with varying proportions of air, 13 and for the sake of ready reference, the limit and maximum flame speeds for various gases mixed with air are given in the accompanying table. It will be observed that the limit speeds, for both the lower and higher limit concentration values tend to approach the same value (approximately 20 cms. per second) for all the gases concerned. 80 PROPAGATION OF FLAME Further, the maximum speeds of the hydrocarbon gases are practically the same, namely, about 82 cms. per second, with the sole exception of methane. The value for this last gas is about 67 cms. per second. It is rather remarkable to note that in each instance the mixture possessing the maximum flame speed should contain more combustible gas than 100 " U 67 o 5 10 15 PER CENT COMBUSTIBLE GAS FIG. 20. Flame Speeds of Hydrocarbon Gases (Payman). is required for complete combustion, save again in the case of methane. The ratio oxygen required for complete combustion oxygen yielding maximum flame speed is approximately 1 to 0-85 by volume. 14 In the case of methane, the maximum speed occurs with the proportion required for complete combustion. FLAME SPEEDS IN AIR 81 Flame Propagation in Air Mixtures of Various Combustible Gases. r 1 na Lower Limit. Maximum Flame Speed. Upper Limit. uas. Gas. Flame Speed. Gas. Flame Speed. Gas. P'lame Speed. Per cent. (cm. sec.). Per cent. (cm. sec.). Per cent. (cm. sec.). CH 4 is 5'80 23-3 9-52 66-6 13-35 19'1 C 2 H 6 3-30 18-1 6-53 85-6 10-60 19-7 C 3 H 8 w 2-37 20-8 4-71 82-1 7-30 20-3 C!H o 1-95 20-1 3-66 82-6 6-53 20-3 C 5 H 12 1*61 20-2 2-92 83-0 5-40 20-2 H 2 . , 15 CH 4 + H 2 6-19 6-03 10-0 15-0 36 -3 12 14-93 503-0 12 135-3 71-39 20-80 50-0 24-3 CO . 15 - 16 16-29 19-5 44-84 60-1 71-19 19-4 CH 4 + CO 15 9-45 21-9 15-95 91-3 21-55 19-8 CO + H 2 . 15 9-25 18-2 45-92 315-2 71-34 44-4 C 2 H 2 . 3-45 41-0 8-9 282-0 16'0 68-0 Experiments with gases other than hydrocarbons, such, for example, as hydrogen and carbon monoxide, or with mixtures of these with methane, do not manifest the same regularities as the* simple hydro- carbon gases. 15 This will be evident from a glance at the curves given in Fig. 21. Although in most cases the lower limit mixtures yield flame speeds not widely removed from 20 cms. per second, in the case of hydrogen the lower limit speed is only 10 cms. Considerable variation is manifest in the upper limit speeds, that of hydrogen being 50 cms., and that of a mixture of hydrogen and carbon monoxide in equal proportions by volume being 44-4 cms. per second. (See Table.) F 82 PROPAGATION OF FLAME The most pronounced speed variation, however, occurs with the various maxima. Thus, the maximum speed attained in a mixture of hydrogen and air is nearly eight times the maximum attained in methane and air. 18 500