An automated system of sample analysis for a total carbon analyser

Standard analysis of the dissolved organic carbon (DOC) content of seawater involves the wet oxidation technique described by Menzel and Yaccaro [ 1]. In this method the sample is sealed in ampoules containing potassium persulphate. As the samples are heated in a sealed pressure vessel, the organic carbon is oxidized to CO2 which is then measured with a nondispersive infra-red analyser. Wet oxidation is best suited to seawater analysis because it is unaffected by high salt concentrations and has a high sensitivity. Due to the large number of samples generated in our research of estuarine and oceanic waters, a fast, economical, automated sample vial handling device was needed which could be used in conjunction with an Oceanography International 0524B Total Carbon System. Our present method of DOC analysis requires tedious and time-consuming manipulation of sealed ampoules and constant operator attention [2]. The automated sample handling system allows the operator to perform other tasks while samples are analysed automatically and with greater efficiency. Other automated analysis systems for organic carbon have been developed, but they involve photo-oxidation and would require complicated changes in the preparation and analysis of our samples [3 and 4-1. Therefore, we decided to automate our present manual system. This new system uses simple, consecutive logic and a series of limit switches and time delay relays to rotate the ampoule into position, break the top of the glass ampoule and purge each sample of CO. with nitrogen. We decided that it must be designed with as little alteration of the original instrument as possible and must be able to convert between the manual and automatic mode quickly and easily. Sets of standard curves were analysed on both the manual and automated systems. The curves were tested for homogeneity as a group using analysis of covariance as described by Zar [5-1 and Sokal and Rholf[6]. The analysis revealed that there was no significant difference between the regression lines ofthe standard curves analysed by the different systems although there is an inherent difference caused by variations in the preparation ofthe samples. Sets of identical samples were also analysed using both methods, and an increase in precision was noted in the automated analysis over the manual method.