In vacuum emission spectrometric analysis of pig iron by D. C. low voltage spark excitation in argon, an unusual or improper discharge, whose symptoms are a prolonged integration time and whitishlooking burn marks on the sample surface, is encountered on apparently good sound samples and gives seriously erratic results.
The experiment showed that under the discharge in question the samples sparked with an integration time between 5 and 30% longer than normal and that the negative errors of 0.04% silicon, 0.025% manganese, and 0.02% phosphorus were common. The exception was sulfur but the results were less reproducible.
An optical and electron microscopic study of the sample surface and burn marks revealed that the improper excitation was caused owing to a preferential sparking to flakes of graphite.
A detailed study of the analytical conditions was carried out and it was found that a long pre-spark of 60 seconds altered the spark characteristics and improved the analytical accuracy and precision to a considerable extent.
Some important problems associated with the spectrometry, i. e., the distribution of excited atoms in the analytical gap and the selection of internal standard lines, were also investigated and discussed.
