14 WHEELER THE IGNITION OF GASES. 1V.-The Ignition of Gases. Part V. Ignition by Inductance &arks. Mixtures of the Para fins with Air. By RICHARD VERNON WHEELER. ONE object of this part of the research on the ignition of gases was to compare the relative ignitibilities of mixtures of methane and air by inductance sparks (low-tension “ break-flashes ” or momentary arcs) with the values obtained when capacity sparks (high-tension impulsive discharges) were used as described in J., 1920 117 903. For as a source of ignition of gaseous mixtures, an inductance spark (produced when an electric current in an inductive circuit is interrupted by the separation of metallic con-tacts) differs from a capacity spark mainly in its longer duration, the difference being sufficiently wide to make it of importance to discover whether inductance sparks can be regarded with capacity sparks as “ momentary ” sources of heat (see J.1924 125 1858), or whether they more nearly approach in character ‘‘ sustained ” sources such as heated surfaces (see ibid. p. 1869). The character of inductance sparks is most susceptible to change PART V. IGNITION BY INDUCTION SPARKS. 15 in the conditions under which they are produced so that consider-able variation can exist in their incendivity. Fig. 1 constructed mainly from oscillograph records represents the variations with time of current resistance voltage and total heat generated in an inductance spark-gap produced by the rapid separation of metallic contacts. As the area of contact anterior to rupture of the circuit, decreases the electric resistance a t that point increases and heat is generated.Eventually the last remaining points of contact become so hot that the metal volatilises and a t the actual moment of break of circuit a conducting band of metallic vapour is pro-duced. This band rapidly increases in length as the fracture is widened until the spark can no longer be maintained. With a spark of this general character just capable of igniting a given inflammable mixture ignition most probably occurs towards the end of its duration but the precise moment depends upon the exact character of the spark as determined by the rate of increase of resistance and of decay of current in it. For these reasons, in any attempt such as is made in this research to determine the relative ignitibilities of different gaseous mix-t'ures by means of inductance sparks of different intensities we must recog-nise not only the changes measured or deduced purposely made in the intensity but also the changes in the character of the sparks.These changes FIG. 1. e m *! in character may be either inadvertent (as when the condition of the metallic surfaces that are separated changes) or a necessary concomitant of 8 change in intensity (as when the inductance of the circuit is purposely altered). In the production of inductance sparks there are six chief vari-ables which can be divided into two groups according as they relate to ( a ) electrical or ( b ) mechanical conditions. I n the former class are (1) The self-inductance of the circuit ; (2) the impressed voltage ; and (3) the current flowing in the circuit before rupture.The latter class includes (4) The nature of the metal a t the spark gap; (5) the rate of break of circuit; and (6) the area of contact a t the moment of break. Each of these variables can be more or less effectively controlled independently and the influence of each can therefore be determined. The most difficult to control and of which to gauge the influence is the last-named and most of th 16 WHEELER THE IGNITION OF GASES. experimental difficulty of this work has been in maintaining con-stancy of this factor. For with the production of a spark there is a change in the condition of the surfaces at which it passes and with a readily oxidisable metal the change may be sufficient to alter considerably the area of contact available for successive sparks as was observed when the influence of different metals at the spark gap was studied.Platinum or gold surfaces were found to be least susceptible to change and the former have been used for the majority of the experiments. Three different types of apparatus each of which had its advantages for particular series of experiments have been used. These are described in the experimental portion of this paper and are referred to as A B and C. Electrical Conditions. With an inductive circuit carrying continuous current the energy that should theoretically appear in the break-flash is the amount of energy stored electromagnetically in the system and should therefore amount to &Li2.This expression does not however, take into account losses in the circuit or absorption at the sparking-points and although it might be permissible to express the relative incendivities of sparks produced under constant circuit conditions (with a given apparatus) by their energy values it would be mislead-ing to suggest a comparison of these values with others obtained under different circuit conditions and with a different apparatus for producing the sparks. The relative ignitibilities of different gaseous mixtures are therefore expressed in this paper simply by the values of the currents (in amperes) flowing iii the circuit at the moment of interruption which yielded an inductance-spark just capable of causing ignition. (1) The Inductance of the Circuit.-A number of inductances of known magnitudes were prepared consisting of coils of silk-covered copper wire wound on cores of wood so as to be of constant value a t all currents.These were introduced into the circuit from a battery of dry cells and the current a t 90 60 and 30 volts required for the ignition of different mixtures of methane and air by a break-flash at platinum contacts was determined using apparatus A, which enabled a rapid break of circuit to be obtained. A number of the results are shown graphically in Fig. 2 in which percentages of methane are plotted against igniting-currents each curve being for a given value of inductance and impressed voltage of the circuit. From the values used in the construction of these curves the rela-tionship between the igniting current for a given mixture and the inductance of the circuit can be determined.Thus Fig. 3 shows the relationship when mixtures cont,aining 6*0,7.0 and 8.0% of methan PART V. IGNITION BY INDUCTION SPARKS. 17 were used and the impressed voltage was 90 ; additional values used for these curves are the igniting-currents when the inductance was 0.00815 henry namely 1.52 1-18 and 0.94 amp. for the 6.0 7.0, and 8.0% mixtures respectively. The relationship between the values of L of 0.008 and 0.095 henry can be expressed by the equation ,W4 = A ; that is to say the energy required in the circuit before break (+Liz) to produce the igniting sparks was nearly constant. This result is deceptive however for in other series of experiments carried out wibh a different method of producing the sparks the value +Liz was by no means constant.(In this con-METWPIE PER CENT FIG. 2. nexion see Morgan J. 1919 115 24). A simple law coiinect8ing i with L in ignition experiments could only be expected if sparks of the same charact,er could be produced wit'h different values of L, and it would appear not to be possible to vary the intensity of a spark (by varying the inductance) without altering its character ; for a t any particular instant during a break-flash when L is the inductance of the circuit T t,he resistance of the spark-gap R the resistance of the rest of the circuit and V the impressed voltage, thc value for the current is ( V - L d i / d t ) / ( r + R) whilst its limiting value a t thc outset is V/R.(2) The Impressed Voltage.-In general it can be stated that the amount of current in the circuit is of far greater importance tha 18 WHEELER THE IGNITION OF GASES. the impressed voltage as regards the igniting power of the flash produced on breaking the circuit. Especially is this so with highly self-inductive circuits. Thus an 8.0 yo methane-air mixture was ignited by the break-flash with a current of from 0.24 to 0.25. ampere under the conditions of the experiments at any voltage between 10 and 3Q the self-induction of the circuit being 0.095 henry. With higher circuit voltages however the igniting current decreases, as the results recorded in Fig. 2 show. When the contacts at which the break-flash is produced are together the voltage drop between them is zero; as soon as they are completely separated, the voltage between them is equal to the impressed voltage.During the separation of the contacts two actions are tending to make the arc between them persist namely the induced voltage which progressively diminishes and the impressed voltage which pro-gressively increases. If the impressed voltage is high i t will contribute materially to the maintenance of the arc. This effect is more marked when the rate of separation of the contacts is com-paratively slow so that the value of the induced voltage (Ldildt) is low. For example a series of experiments using apparatus B, in which massive electrodes of platinum are drawn apart slowly, gave the results shown in Table I. TABLE I. 0.095 henry.E.2I.I.P. (volts) ...... 25 40 60 70 80 90 110 140 Igniting current (amps.) ............ 1-18 1.00 0.80 0.73 0.66 0.60 0.50 0.38 It is evident that the energy of the sparks as deduced from the expression &Liz does not give a true measure of their incendivity ; as is also apparent from the fact that the igniting currents for the same mixture of methane and air (7.8%) as determined in the two apparatus A and B with the same circuit conditions before break (e.g. voltage 90 and inductance 0.095 henry) is markedly different. (3) The Current.-We have to consider the effect of using alternat-ing instead of continuous current. Thornton in his researches on the ignition of gaseous mixtures by inductance sparks (Proc. Roy. Xoc. 1914,90 A 272) has employed alternating currents at various frequencies and voltages and so far as the ignition of mixtures of methane and air is concerned has recorded that much larger currents are required to produce ignition than with direct current.For example under the conditions of his experiments the igniting current for a 9.5% methane-air mixture was 0.5 ampere with con-tinuous current at 200 volts (inductance of circuit not stated), Ignition of a 7-Sy0 Methane-Air Mixture. Inductance of Circui PART V. IGNITION BY INDUCTION SPARKS. 19 whilst with alternating cilrreizt a t 200 volts and 100 periods per second i t was 20 amperes root-mean-square (r.m.s.) value (or 28 amperes crest value) although the arrangement of the resistance and inductance of the circuit is stated to have been the same as that med for the experiments with continuous currents.Similar wide differences with alternating current a t lower voltages and frequencies are recorde I by Thornton. To check this remarkable result duplicate series of experiments werc mads with apparatus -4 using ( a ) continuous current at 35 volts with 0.086 henrF inductance and ( b ) alternating current at 23 volts r.m.s. value (33.3 volts crest value) with 0.098 henry inductance 50 periods per second. The break-flashes were produced between contacts of platinum. I n the experiments with alternating current the procedure was to produce a series of 50 sparks (with a given current) at 5 seconds’ interval in a charge of the mixture of methane and air undergoing test. If no ignition occurred a fresh charge of mixture was admitted to the explosion vessel and a further 50 sparks were passed with the current value increased by 0.01 ampere.This process was repeated until ignition occurred, wiien the current value was reduced by 0.01 ampere and two or three hundred sparks mere passed in several charges of the mixture to ensure that the least igniting current had been determined. The reason for this procedure which was unnecessary with continu-ous current (although i t wts followed in several of the trials in order to make the comparison exact) was that i t was impossible to arrange that the break-flash should be produced always when the current mas a t the crest of its cycle. Ignition was in fact, more “ difficult ” with alternating current because the production of a spark at the crest value of the current was a matter of chance.The lowest current at which ignition could be obtained was not, however much different from that required when continuous current was used. Typical results are recorded in Table 11. TABLE 11. Ignition of Methane-Air Mixtures using (a) Continuous and (b) Alternating Czcrrent (50-). Voltage 33. Inductance 0.095 henry . Methane yo ............ (3.15 7-10 7.GO S.00 8-50 9.00 9.60 10.20 10.90 Igniting current (amps.). Continuous 0.43 0.30 0.26 0.24 0.24 0.25 0.28 0.32 0.42 Alternating (crest values) ............... 0.49 0-36 0.30 0.2G 0.24 0.26 0.26 0.30 0.44 It was anticipated that under certain conditions rather more current would be required in the break-flash to ignite a give 20 WHEELER THE IGNITION OF GASES.mixture when the source of supply was alternating than when i t was continuous for the reason that the rapidly changing value of the former might shorten appreciably the duration of the transient arc and the fact that the values for the igniting currents of the mixtures containing the lower percentages of methane are rather higher with alternating than with continuous current is probably due to this effect. Mechanical Conditions. (4) The Nature of the Metal at the Spark-gap.-Since an inductance spark is of the nature of a transient arc current being conducted across the gap through the vapour of the metal i t would seem prob-able that the lower the volatility of the metal conductor the lower would be the igniting current for a given mixture other conditions being constant for less energy would be expended in forming a path for the current or that path might remain open during a longer int,erval of time.Several series of experiments to determine this question were made using apparatus C which was designed to allow of the ready interchange of contacts of different metals whilst preserving as nearly as possible all other experimental conditions constant. The results are summarised in Table 111. Mixtures of methane and air containing between 8.35 and 8.55% of methane were used and the inductance of the circuit was 0.03175 henry. TABLE 111. Igniting Currents with Contacts of Diferent Metals. (Mixtures of Methane and A i r ) . Metal. Cadmium ............ Zinc ..................Aluminium ......... Silver .................. Tin ..................... Copper ............... Nickel. ................. Iron .................. Platinum ............ Gold .................. Boiling point. 778" 91s 1 so0 1955 2270 2310 2330 ? 2450 2450 ? 2530 7 At 80 volts. 0-34 0.44 0.66 0-63 0.58 0.65 0.86 ---Igniting current. Ampere. First series. Second -'-= series. At 100 At 120 At 120 volts. volts. volts. - 0.22 0.23 0-26 0.23 0.25 - - 0.30 0-4 1 0.38 0.32 0.53 0.45 -- 0.49 0.38 0-58 0.55 0.39 0.52 0.49 0.42 0.56 0.48 0.48 0.59 0.50 0.34 The determinations presented considerable difficulty for not only was i t necessary to ensure that the area of contact between the poles 'at the moment of separation was the same in parallel experiments with different metals (a matter requiring fine adjust-ment of the apparatus) but with all the metals except platinu PART V.IGNITION BY INDUCTION SPARKS. 21 and gold the product’ion of a single spark sufficed to oxidise tJhe surfaces to a greater or less degree (thus altering the area of metallic contact) so that in most instances i t was necessary to repolish the surf aces between each break-flash. I n Table 111 the metals have been arranged in order of their boiling points (as recorded in Kaye and Laby’s ( ( Physical and Chemical Constants,” 1821) and it is clear that there is a close relationship between those values and the ( ( igniting currents ” under standard electrical conditions of a given mixture of methane and air.The energy available a t break is utilised mainly iii pro-ducing an arc of volatilised metal and a given quantity of energy presumably produces an arc of short duration if the metal has a high boiling point and an arc of longer duration when the boiling point is relatively low. The duration of the break-flashes with the metals that gave the highest and the lowest results for the igniting currents (gold platinum zinc and cadmium) were determined by photographing them 011 a rapidly revolving plate. The results are recorded in Table IF7. TABLE IV. -Th:~ai:*on of Break-flushes that Causes Ignition of an 8.5% dfethane-Air Mixture. Relative Igniting-current at 120 volts. Duration of Break-31ctal. Ampere. flash. Second.Cndniium ..................... 0.23 0.00321 Z i no ........................... 0.25 0.00234 Platinum ..................... 0.48 0~00081 Gold ........................... 0.50 0~00070 Thus under standard coiiditions a break-flash between cadminm surfaces with a current of 0.23 ampere flowing in the circuit before interruption lasts four times as long as one between platinum surfaces with a current of 0.48 ampere a fact which no doubt accounts mainly if not entirely for both sparks having the same incendivity although the amounts of energy in the circuit a t their iiioments of formation are so different. I n this connexion reference may be made to determinations by v. Lsiig (?Vied. Ann. 1887 31 384) of the minimum arcing potential using poles of different metals for although his results refer to maintained arcs which the break-flashes are not they give a measure of the degree of ease with which such arcs can be produced.The values depended essentially upon the distance apart of the poles and the current flowing in the circuit and could be expressed by a formula p = a + bli in which p is the observed P.D. in volts between the poles I is the distance apart of the poles in mm. and i is t,he current in amperes b being a constant (independent of th 22 WHEELER THE IGNITION OF GASES. current) and a the E.2II.F. required to maintain the arc. Taking the somewhat arbitrary values of 0-5 ampere for i and 0-5 mm. for I as lying within the range of the experiments recorded in Table 111, v. Lang’s determinations were Cadmium 10.9 ; zinc 20.0 ; silver, 20-0; copper 24.0; iron 25-2; nickel 26.4 and platinum 27.8 volts showing that the ease with which the arc is maintained is directly connected with the volatility of the metal.( 5 ) The Rate of Break of Circuit.-Since the incendivity of the break-flashes depends in part on the inductance voltage and since the magnitude of the inductance voltage depends on the product of the coefficient of self-induction 1; and the rate of change di/dt of the current in the circuit i t follows that the rate of break of the circuit-the speed at which the metallic contacts are separated-affects the incendivity of break-flashes produced under otherwise identical conditions. This is demonstrated qualitatively by experiments made with apparatus C using platinum electrodes, which showed that the igniting-current for a given mixture of methane and air was 0.24 ampere when the rate of break was “ rapid ” and 0.60 ampere when it was “ slow,” other conditions remaining constant.(6) The Area of Contact at the Moment of Break-Since the pro-duction of a break-flash depends essentially on the temporary provision of a path for the current through a band of metallic vapour i t follows that the incendivity of a spark produced under otherwise identical conditions will be affected by changes in the area of metallic contact at the moment of break; for the smaller the area of contact at the instant of rupture the more readily will the mass of metal that then remains to form a conductor be turned into vapour that can continue the conduction and the smaller the volume of vapour thus produced the greater will be the amount of energy concentrated in it and the greater in consequence will be its incendivity.Experiments (made with apparatus C) using pole pieces of platinum of different cross-sectional area with their surfaces carefully polished and aligned showed that a lower igniting-current was required for a given mixture of methane and air (under otherwise identical electrical conditions) the smaller the area of the poles. Further information on the effect of the area of contact at the moment of break was obtained when the poles were made of a readily oxidisable metal such as zinc for then unless the surfaces were repolished between each spark the igniting current regularly decreased to a minimum (at which sparking ceased) as oxidation proceeded ; presumably because the coating of oxide gradually reduced the area of metallic contact PART V.IGNITION BY INDUCTION SPARKS. 23 From the fact that so many factors each having considerable influence on the character of the sparks have to be taken into account it will be realised that repetition of the results of appar-ently parallel experiments was by no means easy to obtain. No success attended experiments in which the break of circuit at which the flash occurred was made by hand; it was only by rendering all possible operations mechanical and automatic that any degree of consistency in the igniting currents could be secured and during the course of each series of experiments repeated checks had to be carried out with a standard mixture (8.5% of methane in air) under standard conditions to ensure that no unnoticed change had taken place in the condition of the contacts.From this study of the electrical and mechanical conditions necessary to produce inductance sparks of uniform character itl appeared that the optimum conditions were obtained if (a) The battery voltage was low and the inductance fairly high so as to ensure that the sparks should be maintained primarily by the inductance voltage; ( b ) the metal contacts a t which the sparks were formed were not readily oxidised; (c) the rate of separation of the contacts was rapid ; and ( d ) the area of contact a t the moment of break was small. Comparative series of experiments were made with mixtures of each of the paraffins methane ethane propane, butane and pentane with air using continuous current a t 30 volts with 0.093 henry inductance.The sparks were formed lietween contacts of platinum in apparatus A which was judged to provide the optimum mechanical conditions for the production of sparks of uniform character. The curves relating percentage of inflammable gas in the mixtures with air to " igniting-currents " were similar with each hydrocarbon, to those obtained when secondary discharges (capacity sparks) were used as the means of ignition (see J. 1924 125 1860) save that in each instance differentiation between the more readily ignitible mixtures was not so marked. The same differences in the degree of ignitibility of the paraffins was also observed but again the differences were not so marked.The essential data are given in Table V. The general result of these experiments is to show that so far as the paraffin hydrocarbons are concerned inductance sparks can be considered similar in effect to capacity sparks as means of ignition despite the wide difference there is in the two types both as regards duration and volume. The lolrger duration and larger volume of the inductance sparks apparently have the effect of mask-ing small differences in the ignitibility of those mixtures that are most readily ignited 24 WHEELER THE IGNITION OF GASES. TABLE V. Mixtures of the Parasns with Air. Mixtures Relative most readily igniting Mixtures most ignited by currents. readily ignited Combustible secondary Secondary by inductance gas.discharge. discharge. sparks. \ 2 Y - (J. 1924 125 1863.)* Ra.nge. Mean. Per cent. Ampere. Per cent. Methane ......... 8-3 0.59 7.8-9.0 8.4 Ethans ......... 6-7 0.47 6.0-6.8 6.4 n-Butane ...... 4.2 0.48 3.8-4.4 4.1 n-Pentane ...... 4.0 0.52 3.6-4.2 3-9 n-Propane ...... 5.1 0.36 4-8-5.4 6-1 Relative igniting currents. Inductranee sparks. Ampere. 0.24 0.15 0.12 0.15 0-23 * Currents in the primary circuit (see J. 1920 117 903.) E X P E R I M E N T A L . Apparatus A (Fig. 4). A brass rod passing through the side of a spherical glass vessel of 100 C.C. capacity carried at its end a pointed strip of platinum A to form one of the electrical contacts at which the inductance spark should be produced.The other contact was a platinum rod B. This rod was mounted on a glass support which passed through the ground-glass bearing C and could be caused to revolve by means of the pulley D driven by an electric motor. The glass support was hollow so as to enable electrical connexion to be established (by means of a copper wire E passing through it) between short pieces of thick platinum wire fused into either end. The upper platinum wire carried the contact rod B, and the lower wire dipped into a mercury-cup F whence the electric circuit could be completed. The rod was revolved a t such a speed as to make contact every 5 seconds with the strip which was bent at, an angle (in a manner not apparent from Fig. 4) so that the rod as i t revolved remained in contact with i t during about half a second and then released it suddenly forming a quick break of circuit.Apparatus E (Fig. 5).-The poles were cones of platinum fused into hollow glass supports which passed through ground glass bearings on opposite sides of a glass globe of 100 C.C. capacity. One support A was held by light springs (which allowed it a small amount of movement) so that the platinum pole was normally at the centre of the globe. The support B was attached to a strong spring C which could pull i t half-way through the bearing. A revolving cam (not shown in the diagram) acting on the rod D, pushed this support against the pull of the spring so that its platinum pole made contact periodically with that of A. Electrical con-nexions were made through copper wires passing within the hollo PART V.IGSITION BY INDUCTION SPARKS. 25 supports as in apparatus A. The cam was revolved by an electric motor at such a speed that contact between the poles was made and broken every five seconds and the arrangement was such that contact was maintained during half a second. Apparatus C (Fig. 6).-This apparatus was similar in design to one used by Thornton (The Electrician Sept. 8th 1916). A small solenoid A was supported within a cylindrical explosion vessel of glass of 100 C.C. capacity. Its plunger carried one of the poles a rod 0.5 mm. in diameter. The other pole 1.5 mm. in dia-meter was carried on a fixed support B. The pole pieces were F I G . 4. F I G . 6. F I G . 3 . removable so that different metals could be used.When the solenoid was out of action its plunger dropped so that the end surface of the pole attached to i t rested on that of the fixed pole but on passing an electric current through the coils the plunger was rapidly with-drawn so that a quick break of circuit occurred a t the surfaces of the poles. Make and break of electric circuit in the solenoid (the electrical connesions to which are not shown in the diagram) were made automatically so that the poles were separated every five seconds. Each apparatus could be evacuated so as to eiiable gaseous mixtures of known composition to be introduced. The mixiure 26 BLAIR AND LEDBURY THE PARTIAL FORMALDEHXDE VAPOUR were stored in glass gas-holders over glycerol and water and were analysed before use. I n each series of experiments the electric circuit included a Post Office resistance box (the coils of which were non-inductively wound) to enable small changes of current to be made a measured inductance and an ammeter which was short-circuited when the break-flashes were produced.Except during the experiments with alternating current the source of supply was a battery of dry cells. These experiments were carried out during the years 1914 to 1916. I was assisted throughout by Mr. W. Mason whilst the determinations of the duration of break-flashes with different metals were made by Mr. W. Shepherd to both of whom I am greatly indebted. EXPEILIMENTAL STATION, ESKMEALS CUMBERLAND. [Received November 3rd 1924. 14 WHEELER THE IGNITION OF GASES. 1V.-The Ignition of Gases.Part V. Ignition by Inductance &arks. Mixtures of the Para fins with Air. By RICHARD VERNON WHEELER. ONE object of this part of the research on the ignition of gases was to compare the relative ignitibilities of mixtures of methane and air by inductance sparks (low-tension “ break-flashes ” or momentary arcs) with the values obtained when capacity sparks (high-tension impulsive discharges) were used as described in J., 1920 117 903. For as a source of ignition of gaseous mixtures, an inductance spark (produced when an electric current in an inductive circuit is interrupted by the separation of metallic con-tacts) differs from a capacity spark mainly in its longer duration, the difference being sufficiently wide to make it of importance to discover whether inductance sparks can be regarded with capacity sparks as “ momentary ” sources of heat (see J.1924 125 1858), or whether they more nearly approach in character ‘‘ sustained ” sources such as heated surfaces (see ibid. p. 1869). The character of inductance sparks is most susceptible to change PART V. IGNITION BY INDUCTION SPARKS. 15 in the conditions under which they are produced so that consider-able variation can exist in their incendivity. Fig. 1 constructed mainly from oscillograph records represents the variations with time of current resistance voltage and total heat generated in an inductance spark-gap produced by the rapid separation of metallic contacts. As the area of contact anterior to rupture of the circuit, decreases the electric resistance a t that point increases and heat is generated.Eventually the last remaining points of contact become so hot that the metal volatilises and a t the actual moment of break of circuit a conducting band of metallic vapour is pro-duced. This band rapidly increases in length as the fracture is widened until the spark can no longer be maintained. With a spark of this general character just capable of igniting a given inflammable mixture ignition most probably occurs towards the end of its duration but the precise moment depends upon the exact character of the spark as determined by the rate of increase of resistance and of decay of current in it. For these reasons, in any attempt such as is made in this research to determine the relative ignitibilities of different gaseous mix-t'ures by means of inductance sparks of different intensities we must recog-nise not only the changes measured or deduced purposely made in the intensity but also the changes in the character of the sparks.These changes FIG. 1. e m *! in character may be either inadvertent (as when the condition of the metallic surfaces that are separated changes) or a necessary concomitant of 8 change in intensity (as when the inductance of the circuit is purposely altered). In the production of inductance sparks there are six chief vari-ables which can be divided into two groups according as they relate to ( a ) electrical or ( b ) mechanical conditions. I n the former class are (1) The self-inductance of the circuit ; (2) the impressed voltage ; and (3) the current flowing in the circuit before rupture.The latter class includes (4) The nature of the metal a t the spark gap; (5) the rate of break of circuit; and (6) the area of contact a t the moment of break. Each of these variables can be more or less effectively controlled independently and the influence of each can therefore be determined. The most difficult to control and of which to gauge the influence is the last-named and most of th 16 WHEELER THE IGNITION OF GASES. experimental difficulty of this work has been in maintaining con-stancy of this factor. For with the production of a spark there is a change in the condition of the surfaces at which it passes and with a readily oxidisable metal the change may be sufficient to alter considerably the area of contact available for successive sparks as was observed when the influence of different metals at the spark gap was studied.Platinum or gold surfaces were found to be least susceptible to change and the former have been used for the majority of the experiments. Three different types of apparatus each of which had its advantages for particular series of experiments have been used. These are described in the experimental portion of this paper and are referred to as A B and C. Electrical Conditions. With an inductive circuit carrying continuous current the energy that should theoretically appear in the break-flash is the amount of energy stored electromagnetically in the system and should therefore amount to &Li2. This expression does not however, take into account losses in the circuit or absorption at the sparking-points and although it might be permissible to express the relative incendivities of sparks produced under constant circuit conditions (with a given apparatus) by their energy values it would be mislead-ing to suggest a comparison of these values with others obtained under different circuit conditions and with a different apparatus for producing the sparks.The relative ignitibilities of different gaseous mixtures are therefore expressed in this paper simply by the values of the currents (in amperes) flowing iii the circuit at the moment of interruption which yielded an inductance-spark just capable of causing ignition. (1) The Inductance of the Circuit.-A number of inductances of known magnitudes were prepared consisting of coils of silk-covered copper wire wound on cores of wood so as to be of constant value a t all currents.These were introduced into the circuit from a battery of dry cells and the current a t 90 60 and 30 volts required for the ignition of different mixtures of methane and air by a break-flash at platinum contacts was determined using apparatus A, which enabled a rapid break of circuit to be obtained. A number of the results are shown graphically in Fig. 2 in which percentages of methane are plotted against igniting-currents each curve being for a given value of inductance and impressed voltage of the circuit. From the values used in the construction of these curves the rela-tionship between the igniting current for a given mixture and the inductance of the circuit can be determined.Thus Fig. 3 shows the relationship when mixtures cont,aining 6*0,7.0 and 8.0% of methan PART V. IGNITION BY INDUCTION SPARKS. 17 were used and the impressed voltage was 90 ; additional values used for these curves are the igniting-currents when the inductance was 0.00815 henry namely 1.52 1-18 and 0.94 amp. for the 6.0 7.0, and 8.0% mixtures respectively. The relationship between the values of L of 0.008 and 0.095 henry can be expressed by the equation ,W4 = A ; that is to say the energy required in the circuit before break (+Liz) to produce the igniting sparks was nearly constant. This result is deceptive however for in other series of experiments carried out wibh a different method of producing the sparks the value +Liz was by no means constant.(In this con-METWPIE PER CENT FIG. 2. nexion see Morgan J. 1919 115 24). A simple law coiinect8ing i with L in ignition experiments could only be expected if sparks of the same charact,er could be produced wit'h different values of L, and it would appear not to be possible to vary the intensity of a spark (by varying the inductance) without altering its character ; for a t any particular instant during a break-flash when L is the inductance of the circuit T t,he resistance of the spark-gap R the resistance of the rest of the circuit and V the impressed voltage, thc value for the current is ( V - L d i / d t ) / ( r + R) whilst its limiting value a t thc outset is V/R. (2) The Impressed Voltage.-In general it can be stated that the amount of current in the circuit is of far greater importance tha 18 WHEELER THE IGNITION OF GASES.the impressed voltage as regards the igniting power of the flash produced on breaking the circuit. Especially is this so with highly self-inductive circuits. Thus an 8.0 yo methane-air mixture was ignited by the break-flash with a current of from 0.24 to 0.25. ampere under the conditions of the experiments at any voltage between 10 and 3Q the self-induction of the circuit being 0.095 henry. With higher circuit voltages however the igniting current decreases, as the results recorded in Fig. 2 show. When the contacts at which the break-flash is produced are together the voltage drop between them is zero; as soon as they are completely separated, the voltage between them is equal to the impressed voltage.During the separation of the contacts two actions are tending to make the arc between them persist namely the induced voltage which progressively diminishes and the impressed voltage which pro-gressively increases. If the impressed voltage is high i t will contribute materially to the maintenance of the arc. This effect is more marked when the rate of separation of the contacts is com-paratively slow so that the value of the induced voltage (Ldildt) is low. For example a series of experiments using apparatus B, in which massive electrodes of platinum are drawn apart slowly, gave the results shown in Table I. TABLE I. 0.095 henry. E.2I.I.P. (volts) ...... 25 40 60 70 80 90 110 140 Igniting current (amps.) ............1-18 1.00 0.80 0.73 0.66 0.60 0.50 0.38 It is evident that the energy of the sparks as deduced from the expression &Liz does not give a true measure of their incendivity ; as is also apparent from the fact that the igniting currents for the same mixture of methane and air (7.8%) as determined in the two apparatus A and B with the same circuit conditions before break (e.g. voltage 90 and inductance 0.095 henry) is markedly different. (3) The Current.-We have to consider the effect of using alternat-ing instead of continuous current. Thornton in his researches on the ignition of gaseous mixtures by inductance sparks (Proc. Roy. Xoc. 1914,90 A 272) has employed alternating currents at various frequencies and voltages and so far as the ignition of mixtures of methane and air is concerned has recorded that much larger currents are required to produce ignition than with direct current.For example under the conditions of his experiments the igniting current for a 9.5% methane-air mixture was 0.5 ampere with con-tinuous current at 200 volts (inductance of circuit not stated), Ignition of a 7-Sy0 Methane-Air Mixture. Inductance of Circui PART V. IGNITION BY INDUCTION SPARKS. 19 whilst with alternating cilrreizt a t 200 volts and 100 periods per second i t was 20 amperes root-mean-square (r.m.s.) value (or 28 amperes crest value) although the arrangement of the resistance and inductance of the circuit is stated to have been the same as that med for the experiments with continuous currents.Similar wide differences with alternating current a t lower voltages and frequencies are recorde I by Thornton. To check this remarkable result duplicate series of experiments werc mads with apparatus -4 using ( a ) continuous current at 35 volts with 0.086 henrF inductance and ( b ) alternating current at 23 volts r.m.s. value (33.3 volts crest value) with 0.098 henry inductance 50 periods per second. The break-flashes were produced between contacts of platinum. I n the experiments with alternating current the procedure was to produce a series of 50 sparks (with a given current) at 5 seconds’ interval in a charge of the mixture of methane and air undergoing test. If no ignition occurred a fresh charge of mixture was admitted to the explosion vessel and a further 50 sparks were passed with the current value increased by 0.01 ampere.This process was repeated until ignition occurred, wiien the current value was reduced by 0.01 ampere and two or three hundred sparks mere passed in several charges of the mixture to ensure that the least igniting current had been determined. The reason for this procedure which was unnecessary with continu-ous current (although i t wts followed in several of the trials in order to make the comparison exact) was that i t was impossible to arrange that the break-flash should be produced always when the current mas a t the crest of its cycle. Ignition was in fact, more “ difficult ” with alternating current because the production of a spark at the crest value of the current was a matter of chance.The lowest current at which ignition could be obtained was not, however much different from that required when continuous current was used. Typical results are recorded in Table 11. TABLE 11. Ignition of Methane-Air Mixtures using (a) Continuous and (b) Alternating Czcrrent (50-). Voltage 33. Inductance 0.095 henry . Methane yo ............ (3.15 7-10 7.GO S.00 8-50 9.00 9.60 10.20 10.90 Igniting current (amps.). Continuous 0.43 0.30 0.26 0.24 0.24 0.25 0.28 0.32 0.42 Alternating (crest values) ............... 0.49 0-36 0.30 0.2G 0.24 0.26 0.26 0.30 0.44 It was anticipated that under certain conditions rather more current would be required in the break-flash to ignite a give 20 WHEELER THE IGNITION OF GASES.mixture when the source of supply was alternating than when i t was continuous for the reason that the rapidly changing value of the former might shorten appreciably the duration of the transient arc and the fact that the values for the igniting currents of the mixtures containing the lower percentages of methane are rather higher with alternating than with continuous current is probably due to this effect. Mechanical Conditions. (4) The Nature of the Metal at the Spark-gap.-Since an inductance spark is of the nature of a transient arc current being conducted across the gap through the vapour of the metal i t would seem prob-able that the lower the volatility of the metal conductor the lower would be the igniting current for a given mixture other conditions being constant for less energy would be expended in forming a path for the current or that path might remain open during a longer int,erval of time.Several series of experiments to determine this question were made using apparatus C which was designed to allow of the ready interchange of contacts of different metals whilst preserving as nearly as possible all other experimental conditions constant. The results are summarised in Table 111. Mixtures of methane and air containing between 8.35 and 8.55% of methane were used and the inductance of the circuit was 0.03175 henry. TABLE 111. Igniting Currents with Contacts of Diferent Metals. (Mixtures of Methane and A i r ) . Metal. Cadmium ............ Zinc .................. Aluminium ......... Silver ..................Tin ..................... Copper ............... Nickel. ................. Iron .................. Platinum ............ Gold .................. Boiling point. 778" 91s 1 so0 1955 2270 2310 2330 ? 2450 2450 ? 2530 7 At 80 volts. 0-34 0.44 0.66 0-63 0.58 0.65 0.86 ---Igniting current. Ampere. First series. Second -'-= series. At 100 At 120 At 120 volts. volts. volts. - 0.22 0.23 0-26 0.23 0.25 - - 0.30 0-4 1 0.38 0.32 0.53 0.45 -- 0.49 0.38 0-58 0.55 0.39 0.52 0.49 0.42 0.56 0.48 0.48 0.59 0.50 0.34 The determinations presented considerable difficulty for not only was i t necessary to ensure that the area of contact between the poles 'at the moment of separation was the same in parallel experiments with different metals (a matter requiring fine adjust-ment of the apparatus) but with all the metals except platinu PART V.IGNITION BY INDUCTION SPARKS. 21 and gold the product’ion of a single spark sufficed to oxidise tJhe surfaces to a greater or less degree (thus altering the area of metallic contact) so that in most instances i t was necessary to repolish the surf aces between each break-flash. I n Table 111 the metals have been arranged in order of their boiling points (as recorded in Kaye and Laby’s ( ( Physical and Chemical Constants,” 1821) and it is clear that there is a close relationship between those values and the ( ( igniting currents ” under standard electrical conditions of a given mixture of methane and air.The energy available a t break is utilised mainly iii pro-ducing an arc of volatilised metal and a given quantity of energy presumably produces an arc of short duration if the metal has a high boiling point and an arc of longer duration when the boiling point is relatively low. The duration of the break-flashes with the metals that gave the highest and the lowest results for the igniting currents (gold platinum zinc and cadmium) were determined by photographing them 011 a rapidly revolving plate. The results are recorded in Table IF7. TABLE IV. -Th:~ai:*on of Break-flushes that Causes Ignition of an 8.5% dfethane-Air Mixture. Relative Igniting-current at 120 volts. Duration of Break-31ctal. Ampere. flash. Second. Cndniium ..................... 0.23 0.00321 Z i no ...........................0.25 0.00234 Platinum ..................... 0.48 0~00081 Gold ........................... 0.50 0~00070 Thus under standard coiiditions a break-flash between cadminm surfaces with a current of 0.23 ampere flowing in the circuit before interruption lasts four times as long as one between platinum surfaces with a current of 0.48 ampere a fact which no doubt accounts mainly if not entirely for both sparks having the same incendivity although the amounts of energy in the circuit a t their iiioments of formation are so different. I n this connexion reference may be made to determinations by v. Lsiig (?Vied. Ann. 1887 31 384) of the minimum arcing potential using poles of different metals for although his results refer to maintained arcs which the break-flashes are not they give a measure of the degree of ease with which such arcs can be produced.The values depended essentially upon the distance apart of the poles and the current flowing in the circuit and could be expressed by a formula p = a + bli in which p is the observed P.D. in volts between the poles I is the distance apart of the poles in mm. and i is t,he current in amperes b being a constant (independent of th 22 WHEELER THE IGNITION OF GASES. current) and a the E.2II.F. required to maintain the arc. Taking the somewhat arbitrary values of 0-5 ampere for i and 0-5 mm. for I as lying within the range of the experiments recorded in Table 111, v. Lang’s determinations were Cadmium 10.9 ; zinc 20.0 ; silver, 20-0; copper 24.0; iron 25-2; nickel 26.4 and platinum 27.8 volts showing that the ease with which the arc is maintained is directly connected with the volatility of the metal.( 5 ) The Rate of Break of Circuit.-Since the incendivity of the break-flashes depends in part on the inductance voltage and since the magnitude of the inductance voltage depends on the product of the coefficient of self-induction 1; and the rate of change di/dt of the current in the circuit i t follows that the rate of break of the circuit-the speed at which the metallic contacts are separated-affects the incendivity of break-flashes produced under otherwise identical conditions. This is demonstrated qualitatively by experiments made with apparatus C using platinum electrodes, which showed that the igniting-current for a given mixture of methane and air was 0.24 ampere when the rate of break was “ rapid ” and 0.60 ampere when it was “ slow,” other conditions remaining constant.(6) The Area of Contact at the Moment of Break-Since the pro-duction of a break-flash depends essentially on the temporary provision of a path for the current through a band of metallic vapour i t follows that the incendivity of a spark produced under otherwise identical conditions will be affected by changes in the area of metallic contact at the moment of break; for the smaller the area of contact at the instant of rupture the more readily will the mass of metal that then remains to form a conductor be turned into vapour that can continue the conduction and the smaller the volume of vapour thus produced the greater will be the amount of energy concentrated in it and the greater in consequence will be its incendivity.Experiments (made with apparatus C) using pole pieces of platinum of different cross-sectional area with their surfaces carefully polished and aligned showed that a lower igniting-current was required for a given mixture of methane and air (under otherwise identical electrical conditions) the smaller the area of the poles. Further information on the effect of the area of contact at the moment of break was obtained when the poles were made of a readily oxidisable metal such as zinc for then unless the surfaces were repolished between each spark the igniting current regularly decreased to a minimum (at which sparking ceased) as oxidation proceeded ; presumably because the coating of oxide gradually reduced the area of metallic contact PART V.IGNITION BY INDUCTION SPARKS. 23 From the fact that so many factors each having considerable influence on the character of the sparks have to be taken into account it will be realised that repetition of the results of appar-ently parallel experiments was by no means easy to obtain. No success attended experiments in which the break of circuit at which the flash occurred was made by hand; it was only by rendering all possible operations mechanical and automatic that any degree of consistency in the igniting currents could be secured and during the course of each series of experiments repeated checks had to be carried out with a standard mixture (8.5% of methane in air) under standard conditions to ensure that no unnoticed change had taken place in the condition of the contacts.From this study of the electrical and mechanical conditions necessary to produce inductance sparks of uniform character itl appeared that the optimum conditions were obtained if (a) The battery voltage was low and the inductance fairly high so as to ensure that the sparks should be maintained primarily by the inductance voltage; ( b ) the metal contacts a t which the sparks were formed were not readily oxidised; (c) the rate of separation of the contacts was rapid ; and ( d ) the area of contact a t the moment of break was small. Comparative series of experiments were made with mixtures of each of the paraffins methane ethane propane, butane and pentane with air using continuous current a t 30 volts with 0.093 henry inductance.The sparks were formed lietween contacts of platinum in apparatus A which was judged to provide the optimum mechanical conditions for the production of sparks of uniform character. The curves relating percentage of inflammable gas in the mixtures with air to " igniting-currents " were similar with each hydrocarbon, to those obtained when secondary discharges (capacity sparks) were used as the means of ignition (see J. 1924 125 1860) save that in each instance differentiation between the more readily ignitible mixtures was not so marked. The same differences in the degree of ignitibility of the paraffins was also observed but again the differences were not so marked.The essential data are given in Table V. The general result of these experiments is to show that so far as the paraffin hydrocarbons are concerned inductance sparks can be considered similar in effect to capacity sparks as means of ignition despite the wide difference there is in the two types both as regards duration and volume. The lolrger duration and larger volume of the inductance sparks apparently have the effect of mask-ing small differences in the ignitibility of those mixtures that are most readily ignited 24 WHEELER THE IGNITION OF GASES. TABLE V. Mixtures of the Parasns with Air. Mixtures Relative most readily igniting Mixtures most ignited by currents. readily ignited Combustible secondary Secondary by inductance gas.discharge. discharge. sparks. \ 2 Y - (J. 1924 125 1863.)* Ra.nge. Mean. Per cent. Ampere. Per cent. Methane ......... 8-3 0.59 7.8-9.0 8.4 Ethans ......... 6-7 0.47 6.0-6.8 6.4 n-Butane ...... 4.2 0.48 3.8-4.4 4.1 n-Pentane ...... 4.0 0.52 3.6-4.2 3-9 n-Propane ...... 5.1 0.36 4-8-5.4 6-1 Relative igniting currents. Inductranee sparks. Ampere. 0.24 0.15 0.12 0.15 0-23 * Currents in the primary circuit (see J. 1920 117 903.) E X P E R I M E N T A L . Apparatus A (Fig. 4). A brass rod passing through the side of a spherical glass vessel of 100 C.C. capacity carried at its end a pointed strip of platinum A to form one of the electrical contacts at which the inductance spark should be produced. The other contact was a platinum rod B.This rod was mounted on a glass support which passed through the ground-glass bearing C and could be caused to revolve by means of the pulley D driven by an electric motor. The glass support was hollow so as to enable electrical connexion to be established (by means of a copper wire E passing through it) between short pieces of thick platinum wire fused into either end. The upper platinum wire carried the contact rod B, and the lower wire dipped into a mercury-cup F whence the electric circuit could be completed. The rod was revolved a t such a speed as to make contact every 5 seconds with the strip which was bent at, an angle (in a manner not apparent from Fig. 4) so that the rod as i t revolved remained in contact with i t during about half a second and then released it suddenly forming a quick break of circuit.Apparatus E (Fig. 5).-The poles were cones of platinum fused into hollow glass supports which passed through ground glass bearings on opposite sides of a glass globe of 100 C.C. capacity. One support A was held by light springs (which allowed it a small amount of movement) so that the platinum pole was normally at the centre of the globe. The support B was attached to a strong spring C which could pull i t half-way through the bearing. A revolving cam (not shown in the diagram) acting on the rod D, pushed this support against the pull of the spring so that its platinum pole made contact periodically with that of A. Electrical con-nexions were made through copper wires passing within the hollo PART V.IGSITION BY INDUCTION SPARKS. 25 supports as in apparatus A. The cam was revolved by an electric motor at such a speed that contact between the poles was made and broken every five seconds and the arrangement was such that contact was maintained during half a second. Apparatus C (Fig. 6).-This apparatus was similar in design to one used by Thornton (The Electrician Sept. 8th 1916). A small solenoid A was supported within a cylindrical explosion vessel of glass of 100 C.C. capacity. Its plunger carried one of the poles a rod 0.5 mm. in diameter. The other pole 1.5 mm. in dia-meter was carried on a fixed support B. The pole pieces were F I G . 4. F I G . 6. F I G . 3 . removable so that different metals could be used. When the solenoid was out of action its plunger dropped so that the end surface of the pole attached to i t rested on that of the fixed pole but on passing an electric current through the coils the plunger was rapidly with-drawn so that a quick break of circuit occurred a t the surfaces of the poles. Make and break of electric circuit in the solenoid (the electrical connesions to which are not shown in the diagram) were made automatically so that the poles were separated every five seconds. Each apparatus could be evacuated so as to eiiable gaseous mixtures of known composition to be introduced. The mixiure 26 BLAIR AND LEDBURY THE PARTIAL FORMALDEHXDE VAPOUR were stored in glass gas-holders over glycerol and water and were analysed before use. I n each series of experiments the electric circuit included a Post Office resistance box (the coils of which were non-inductively wound) to enable small changes of current to be made a measured inductance and an ammeter which was short-circuited when the break-flashes were produced. Except during the experiments with alternating current the source of supply was a battery of dry cells. These experiments were carried out during the years 1914 to 1916. I was assisted throughout by Mr. W. Mason whilst the determinations of the duration of break-flashes with different metals were made by Mr. W. Shepherd to both of whom I am greatly indebted. EXPEILIMENTAL STATION, ESKMEALS CUMBERLAND. [Received November 3rd 1924.