July J, 1914.

THE COLLIERY GUARDIAN.

19

The Electrical Ignition of Gaseous Mixtures.

By Prof. W. M. THORNTON, D.Sc., D.Eng.

Abstracted from the Proceedings of the Royal Society.

The ignition of explosive gaseous mixtures for experi-
mental purposes is generally made by an electric spark
or train of sparks between fixed terminals. The tem-
perature of inflammation cannot in this case be
measured, and it is determined by that of a hot surface
in contact with the gas, or by calculation from its adia-
batic compression. The fact that there is a critical tem-
perature of ignition and that the velocity of an explosion
wave can be calculated from the thermal constants of
the gas and air, has led to the view that the process is
a thermal one throughout, with in general two stages,
a period of slow combustion and rise of temperature,
and the true explosion on this reaching a certain limiting
value. There is, however, a more intimate possible
cause of the division of the molecule of combustible gas
which precedes explosive combination. Recent work
on the ionisation of gases has made familiar the view
that a molecule can be ionised by corpuscular radiation,
and that by the gain or loss of such corpuscles the
nature of the molecule can be profoundly modified.
The present paper is an examination of certain typical
gases and vapours for the purpose of finding evidence of
the mechanism of the process by which the energy of the
source is transferred to the gas at the moment of
ignition.

An important series of observations on gaseous igni-
tions has been recently made by my colleague, Mr. J. R.
Thompson.* He found that it is possible to ignite a
cold explosive mixture by the incidence of X-rays on
a platinum surface in it, and that when the source of
ignition is a hot platinum wire an explosion is started
at that temperature at which ions are discharged from
the metal. These observations, if they do not decide
the ionic origin of gaseous explosions in general, prove
that ionisation and explosion are intimately connected.

Direct evidence of the kind of electrical action which
starts a gaseous explosion under normal conditions is
given by the difference between the igniting powers of
continuous and alternating current break-sparks. If
ignition were a simple thermal process, the igniting
power, measured by varying some non-electrical factor,
such as the percentage of combustible gas in the mixture,
and finding the corresponding variation of the least
igniting current, should be proportional to the square
of the current in the circuit, whether continuous or
alternating, other conditions being the same. But when
the current is alternating, the present experiments show
that the energy of the circuit required to ignite gas is
much greater than when the current is continuous, in
the case of methane 56 times greater at 200 volts; the
igniting effect is therefore not simply thermal. Again,
the current in a metallic circuit is proportional to the
number of free electrons passing per second. When the
current is broken, these are projected into the break-
spark by the electric gradient at the moment of break,
and a small volume of the gas in their immediate neigh-
bourhood is ionised by them. If the igniting power
of the current is ionic it should then be proportional to
the current, when this is unidirectional, and this found
to be the case in many gases over an important part of
their working range of inflammability. The differences
between continuous and alternating current are not only
in the magnitudes of the effects, but in the manner of
development of the curves in a series of related gases.
The general conclusion of a previous paperfi on this

Fig. 1.

3-0

2S

lOO' '	200	3:

.— IO PER. CENT

" sod....................'coo

Variation of least igniting con-
tinuous current with voltage on
circuit. Methane. Iron poles.

subject was that ignition by continuous current break-
sparks depends mostly upon the nature of the arc, that
by alternating current on the nature of the gas.

The duration of the arc at break, observed photo-
graphically by focussing it on a sensitive film attached
to a revolving drum, was found to depend, in the same
gaseous mixture, chiefly upon the voltage and the
material of the poles. The rate of break had a nearly
constant value of 12 cm. a second. Except at very slow
breaks, from 1 to 2 cm. a second, differences in the rate
of break by hand had no measurable influence upon the
current which, when broken, just ignited the mixture.
In every case the mixture before explosion was at
atmospheric pressure and in most cases at atmospheric
temperature. In the present work, iron rods were used
throughout having a diameter of OT in.

* “ Phys. Zeitschr.,” vol. 14, pp. 11-15, January 1, 1913,
and “ Science Abstr.,” No. 631, vol. 16, part 4, April 13,1913.

fi “ The Ignition of Coal Gas and Methane by Momentary
Electric Arcs,” W. M. Thornton, Inst. Mining Engineers
Trans., vol. 44, part 1, pp. 145-174, October 12, 1912.

The higher the voltage the less the current required
for ignition. The curves obtained, of which figd is an
example, closely resemble the characteristics of steady
arcs, which are known to be maintained by ionised vapour
between the poles, and they have the same form at all
percentages within the inflammable limits.

Relation of Current to Percentage of Combustible Gas in
Mixture.

The results in the preceding paragraph are of the same
type for all the gases examined. We now consider the
influence of varying the proportion of vapour in a series
of gases, the chemical constitutions of which are closely
related.

Paraffin Series.—The lower limit of inflammability is
very sharply marked in all gaseous mixtures. Many have
been carefully examined by Le Chatelier and Boudouard,*
and the paraffin series more accurately by Burgess and
Wheeler.fi I am informed by Dr. Wheeler that he has
found the upper limits to be equally sharp, and am
permitted to quote the following values for them,
expressed in percentage volume of combustible gas in
the mixture with air :—

Table I.

Lower limit

Methane ..... 5*60

Ethane.......	3 TO

Propane ..... 2’17

Butane.......	1’65

Pentane ..... 1'35

Upper limit.

14’80

10-70

7’35

5*70

4-50

Between these limits there has not been hitherto any
means of observing directly the inflammability of the
mixture, other than by its velocity of explosion. In
fig. 2 are given curves of igniting current for the paraffin
gases by the break-spark method, from which it will be
seen that the change of susceptibility to ignition varies
with the percentage of gas in the same manner for all the
gases examined. The curves have each three distinct
portions, that which gives the peculiar form being a
straight line passing through, or nearly through, zero,
and expressing the fact that over the greater part of
the range of inflammability the least igniting current
is very Mostly proportioned to the number of molecules
of combustible gas in unit volume of the mixture. At
the lower limit there is a steep branch indicating that

AMPERES
4-0 ---

Fig. 2.

Least igniting currents for the paraffin gases in air.
Continuous current. Iron poles. 100 volts.

3»O

20

I’O

•4

•6

•8

•8

•8

•4

•2

•2

•2

•6

o

2. 3

5 6 7	8 q IO II 11 13 14- IS 16

PERCENT OF COMBUSTIBLE GAS IN AIR BY VOLUME'

the transition from the most inflammable mixture to
that which cannot be ignited is sudden. The upper
limit is approached rather more gradually. The fact
that the projection of the straight base lines passes
through zero proves that the igniting current is propor-
tional to the absolute number of molecules, not merely
to the change of percentage.

The limits of inflammability found agree well with
those given above, but no attempt was made to find the
limits within a small percentage. The run of the curves
from the observed points is a sufficiently close check
for our purpose. The tendency of the break-spark igni-
tion would appear to be to lower both the upper and
lower limits. The most remarkable feature of the family
of curves in fig. 2 is that the same igniting current, that
is the same energy, is required for each most inflam-
mable mixture, in spite of the great difference in the
calorific values of methane and pentane. The next is
that there is no square law, except at the limiting mix-
tures. The third that the most inflammable mixture is
not always that in the proportions required for perfect
combustion, but a mixture in which the air is in excess.

The more symmetrically the molecule is bonded the

* Le Chatelier et Boudouard. “ Sur les Limites d’lnflam-
mabilite des Vapeurs,” ‘Comptes Rendus,’ vol. 126, p. 1510
(1897).

fi “ The Lower Limit of Inflammation of Mixtures of the
Paraffin Hydrocarbons with Air,” by M. J. Burgess and
R. V. Wheeler, ‘Chem. Soc. Journ.,’ vol. 99, p. 2013 (1911).

closer the coincidence, as shown in Table II.’, n^ is
the number of hydrogen atoms in a molecule of com-
bustible gas, nc + H the total atoms.

Table II.—Ignition by Continuous Current.

Gas.	Most inflammable mixture, pm.  Per cent.	Mixture for perfect com- bustion to CO2 Per cent.	Pm x n^. pm x +	
Methane .	8’0	9’5	.. 32’0 ...	40’0
Ethane ....	5’1	.	5’5	. 30’6 ...	40’8
Propane,...	3’9	3’96	. 31’2 ...	42’9
Butane ....	3’1	3’07	. 31’0 ...	43’4
Pentane....	2’5	2’51	. 30’0 ...	42’5

Influence of Number of Atoms in the Molecule upon Igniting
Current.

It is safe to say on the electronic theory of conduc-
tion in metals that since the minimum igniting current
is the same in every gas of the paraffin series, the same
number of electrons must be discharged into the mix-
ture. The position of each curve in the figure depends
on the slope of the base line, that is on the ratio I/p,
where I is the igniting current and p the percentage of
gas in the mixture. This slope is not proportional to
the molecular weight or to the heat of combustion. If
the first step in electric ignition is a direct ionic attack
upon the atoms in a molecule, there should be a definite
relationship between the igniting current, considered as
a source of ions, and the number of atoms either in the
molecule of combustible gas or in unit volume. The
ratio I/p is found to be nearly proportional to the number
of hydrogen atoms in the molecule, and since hydrogen
is the most easily ionised of all gaseous molecules it is
at least conceivable that if the effect is due to ionisation
the ignition of the complex molecule of combustible gas
is started by the ionisation of its hydrogen atoms, and
that in its disintegration the first compounds formed are
those of hydrogen and oxygen. This is in agreement
with Prof. H. B. Dixon’s conclusion* that it is during
the “ preflame ” period of ignition which precedes the
true explosion that water vapour influences the rate of
propagation. Amy “ preferential ” action of hydrogen
and oxygen prior to the formation of carbon products
will be exceedingly rapid, and is probably beyond
analysis.

The outstanding result in the continuous current igni-
tion of the paraffin series is that each gas has the same
minimum igniting current. The mixtures at which this
occurs are in most cases near those for perfect combus-
tion, and though it is clear that combustion is an
important factor even in the initial stage, yet there is
evidence that the influence of the discharge from the
spark, whatever its nature may be, acts directly upon
the atoms in a molecule rather than upon the molecules
as a whole.

The curves for methyl and ethyl alcohol are the same
in type as for the paraffins, having a straight base
inclined towards the origin. The curve for benzene
C6H6 differs in important features from the preceding.
It has no straight part and is nearly symmetrical within
the limits of 1’24 and 6’37 per cent, found by Sorelfi
for pure benzene. The most sensitive mixture is about
4’8 per cent.; that for perfect combustion 2’73 per cent.
There is clearly no relation between the latter and the
shape of the curve, but the point of combustion to car-
bon monoxide is 4’47 per cent. The same form and
position of the benzene curve is found when the current
is alternating as when continuous. The evidence is
therefore strong that in so far as the magnitude of the
igniting current is an indication of the atomic changes
occurring at the instant of explosion, carbon monoxide
is the chief and first product of combustion, and the
formation of carbon dioxide is in this case a much slower
process than in lighter gases. Dixon has shown that
combustion to carbon monoxide followed by burning to
carbon dioxide is characteristic of the explosion of many
gases. The evidence given here from the circumstances
of electrical ignition shows that it is much easier to
ignite benzene vapour when it burns to carbon monoxide
than to carbon dioxide. Disulphide of carbon is always
found in commercial benzol and possibly plays an impor-
tant part in the ignition of rich mixtures, for the vapour
is inflammable within wide limits of mixture with a:r.

Hydrogen and Carbon Monoxide.

In the ignition of hydrogen by continuous current
break-sparks the current is found to be the same in all
strengths of mixture except towards the limits of inflam-
mability. Carbon monoxide differs from hydrogen
in that the continuous current curve is slightly inclined,
and the alternating current curve horizontal.

Influence of Circuit Voltage on Form of Curves.

All the previous work has been done with a circuit
voltage of 100. The influence of varying the voltage
is to change the slope of the base in a regular and rapid
manner, fig. 3.

Voltage ...... 50	...	75	...	100	...	125

Slope of base	... 0’472	...	0’222	...	0’119	...	0’0468

The variation is approximately as the cube of the voltage
and is such that at voltages over 100 the igniting current
is nearly the same over the whole range of inflamma-
bility, a point of importance in the working of internal
combustion engines. The circuit voltage has no influ-
ence on the position of the point of greatest inflamma-
bility.

** Time of Explosion.”

The paraffin molecule is in general heavy and complex,
and a number of oxygen molecules sufficient to form
the products of combustion must surround it at the
moment of disintegration. This group of gas and oxygen

* “The Initiation and Propagation of Explosions.” ‘Chem.
Soc. Journ.,’ vol. 99. pp. 589-599. April 1911.

t “ Carburetting and Combustion in Alcohol Engines,”
by E. Sorel. Trans. Woodward and Preston, p. 53.