A Unique Equation to Estimate Flash Points of Selected Pure Liquids Application to the Correction of Probably Erroneous Flash Point Values
Authors:
- 1 December 2004
- journal article
- conference paper
- Published by AIP Publishing in Journal of Physical and Chemical Reference Data
- Vol. 33 (4) , 1083-1111
Abstract
A simple empirical equation is presented for the estimation of closed-cup flash points for pure organic liquids. Data needed for the estimation of a flash point (FP) are the normal boiling point the standard enthalpy of vaporization at 298.15 K of the compound, and the number of carbon atoms in the molecule. The bounds for this equation are: Compared to other methods (empirical equations, structural group contribution methods, and neural network quantitative structure–property relationships), this simple equation is shown to predict accurately the flash points for a variety of compounds, whatever their chemical groups (monofunctional compounds and polyfunctional compounds) and whatever their structure (linear, branched, cyclic). The same equation is shown to be valid for hydrocarbons, organic nitrogen compounds, organic oxygen compounds, organic sulfur compounds, organic halogen compounds, and organic silicone compounds. It seems that the flash points of organic deuterium compounds, organic tin compounds, organic nickel compounds, organic phosphorus compounds, organic boron compounds, and organic germanium compounds can also be predicted accurately by this equation. A mean absolute deviation of about a standard deviation of about and a maximum absolute deviation of are obtained when predictions are compared to experimental data for more than 600 compounds. For all these compounds, the absolute deviation is equal or lower than the reproductibility expected at a 95% confidence level for closed-cup flash point measurement. This estimation technique has its limitations concerning the polyhalogenated compounds for which the equation should be used with caution. The mean absolute deviation and maximum absolute deviation observed and the fact that the equation provides unbiaised predictions lead to the conclusion that several flash points have been reported erroneously, whatever the reason, in one or several reference compilations. In the following lists, the currently accepted flash points for bold compounds err, or probably err, on the hazardous side by at least and for the nonbolded compounds, the currently accepted flash points err, or probably err, on the nonhazardous side by at least bicyclohexyl, sec-butylamine, tert-butylamine, 2-cyclohexen-1-one, ethanethiol, 1,3-cyclohexadiene, 1,4-pentadiene, methyl formate, acetonitrile, cinnamaldehyde, 1-pentanol, diethylene glycol, diethyl fumarate, diethyl phthalate, trimethylamine, dimethylamine, 1,6-hexanediol, propylamine, methanethiol, ethylamine, bromoethane, 1-bromopropane, tert-butylbenzene, 1-chloro-2-methylpropane, diacetone alcohol, diethanolamine, 2-ethylbutanal, and formic acid . For some other compounds, no other data than the currently accepted flash points are available. Therefore, it cannot be assessed that these flash point data are erroneous but it can be stated that they are probably erroneous. At least, they need experimental re-examination. They are probably erroneous by at least 1,3-cyclopentadiene, di-tert-butyl sulfide, dimethyl ether, dipropyl ether, 4-heptanone, bis(2-chloroethyl)ether, 1-decanol, 1-phenyl-1-butanone, furan, ethylcyclopentane, 1-heptanethiol, 2,5-hexanediol, 3-hexanone, hexanoic acid methyl ester, 4-methyl-1,3-pentadiene, propanoyl chloride, tetramethylsilane, thiacyclopentane, 1-chloro-2-methyl-1-propene, trans-1,3-pentadiene, 2,3-dimethylheptane, triethylenetetramine, methylal, N-ethylisopropylamine, 3-methyl-2-pentene, and 2,3-dimethyl-1-butene.