Abstract
An automatic instrument for organic oxygen determination has been constructed by combining a portable nondispersive infrared analyzer with an automatic operation mechanism of CHN analyzer. A sample was instantaneously pyrolyzed in a quartz decomposition tube, and the decomposition gas was passed with carrier gas (helium) through a hot zone of platinized carbon granules. The resultant carbon monoxide was slowly withdrawn during 5.5 min into a stainless steel pump having a reproducible capacity of 150 ml, and the gas was then homogenized in the pump for 1 min. The pump pushed out the gas into the infrared analyzer, which measured the CO concentration as the output signal on an attached recorder. All the flow paths were then scavenged by pure helium for 1 min before starting the next analysis. One cycle of the automatic operation required 13 min.
The concentration in the pump is independent of the pressure and temperature; this fact renders the interpretation of the output signal quite simple. On the other hand, the change of the temperature difference between the pump and the infrared detector caused fluctuation of the sensitivity of the apparatus because the gas expanded or contracted on entering into the detector; for example, with a sample containing 800 μg of oxygen, 1°C change of the temperature difference caused an error of 4 μg in oxygen content. This error could be reduced by isolating the detector unit from the infrared analyzer and installing it together with the pump in a single oven; but still it was found that the temperature variation of the detector produced an error of 20 μg/°C, probably due to the change of radiation efficiency of the light source and to the vibration of the membrane in the detector. Calculation demonstrates that the temperature of the oven containing the pump and the detector unit should be controlled within a limit of ± 0.1°C or less.
A series of analysis of a numbers of standard samples with different chemical constitutions was carried out, and the estimate of standard deviation of oxygen content was 0.18%, which was thought to be acceptable for microanalysis.
