Previous work has demonstrated that knock in the spark-ignition engine is a phenomenon confined to the last part of the charge to burn, and that it is the chemical reactions in this ‘ end-gas ’ which determine whether or not knock will occur. The purpose of this paper is to try to elucidate the nature of these reactions and to discover some of the critical chemical factors controlling the occur- rence of knock. This has been done in three main ways: (
a
) the sequence of chemical reactions occurring in the ‘end-gas’ prior to knock has been followed by means of a specially designed electromagnetic gas-sampling valve which can abstract samples of this gas at various stages in the combustion cycle, (
b
) the effect of the addition of various substances to the cylinder charge on the knock-limited compression ratio of the engine has been determined and (
c
) a study has been made in a motored engine of the limits of cool and hot flame formation. Ricardo E. 6 variable-compression engines were used for all these investigations. Preliminary tests with small quantities of additives, such as azomethane, alkyl nitrites and nitrates, carbon tetrachloride and chloropicrin, demonstrated that substances which would be expected to give rise to free radicals in the engine cylinder are strong pro-knocks. This and the anti-knock action of minute quantities of additives such as lead tetraethyl show the chain nature of the reactions leading to knock. Experiments are described to demonstrate that these reactions are substantially independent of the nature of the cylinder walls. Qualitative sampling tests indicated that knock depends not so much on the formation of some new substance which was not present under non-knocking conditions but rather on the attainment of a critical rate of formation of products present under both conditions. Further experiments with additives in which various intermediate products of reaction, such as aldehydes, nitrogen peroxide and organic peroxides, were tested, showed that with normal fuels the organic peroxides were the only products which were strongly pro-knock and formaldehyde the only product which had an anti-knock effect. This led to the subsequent sampling investigation being concentrated largely on the estimation of these two products in the ‘end-gases’. A method for analyzing for organic peroxides in small concentration and in the presence of other reaction products, such as nitrogen peroxide, was developed. Use of this method with a higher paraffinic fuel showed that peroxide formation was two-stage in nature. The first stage culminated in the formation of a point of inflexion or small peak in the curve of peroxide concentration about 1 ° after top dead centre. In the second stage the curve normally rose again to a higher peak at about 7° 1. Lead tetraethyl had a much greater depressing effect on the peak at 7° L compared with that at 1° 1. The mixture strength giving maximum knock was the same as that giving maximum peroxide concentration. Analyses were also made for aldehydes, shown to be mainly formaldehyde. The main growth of the formaldehyde formation took place in the second of the two stages of peroxide formation. A much smaller quantity of peroxide appeared to be formed when methane was the fuel, and this was eliminated when lead tetraethyl was added to suppress the knock. Benzene formed no peroxides, and none was detected with the alkyl benzenes up to cumene. This difference between benzene and methane on the one hand and the higher paraffinic fuels on the other in regard to peroxide formation was paralleled by the different effect of additives on the two classes of fuel. Formaldehyde was an anti-knock in the latter class and a pro-knock in the former class; acetaldehyde and benzaldehyde were ineffective in the latter but strongly pro-knock in the former; nitrogen peroxide has only a slight pro-knock effect in the latter but was strongly pro knock in the former. Experiments with the higher paraffinic fuels on a motored engine showed that if conditions of temperature and pressure were such as to simulate those obtaining in the ‘end-gas’ prior to knock, cool flames were formed at a time in the cycle approximately corresponding to the point of inflexion of the peroxide sampling curves. No cool flames were detected with benzene or methane. Bright blue flames were observed near the ignition point over certain ranges of mixture strength with benzene and methane as well as with the higher paraffins. Non-engine work has shown the existence of two main types of combustion, namely, so-called ‘low’- and ‘ high ’-temperature types. Of these the first is associated with peroxide and cool-flame formation, ignition taking place by a two-stage process, whereas ignition of the second type is a single-stage process and not usually associated with cool flames or peroxide formation. From the evidence of (1) the peroxide analyses, (2) the experiments with additives and (3) the experiments with a motored engine, it is clear that normal fuels knock by the ‘low’-temperature process but that benzene and methane knock by the ‘ high’-temperature process. Experiments in which formaldehyde was used as a fuel and the effect of various additives on its knock limit determined, showed that its oxidation is best classified as of the ‘ high ’-temperature type and yielded further understanding of the nature of the reactions leading to knock. The effect of hydrogen as an additive with other fuels and the effect of various additives on its own highest useful compression ratio was determined. No peroxides were detected in engine samples when hydrogen was used as the fuel.
The combustion of aromatic and alicyclic hydrocarbons. V. The products of combustion of benzene and its monoalkyl derivatives
