THE USE OF A HIGH ACCURACY ISOTOPE DILUTION MASS SPECTROMETRY (IDMS) METHOD FOR THE ANALYSIS OF SULFUR IN FUEL TO SUPPORT ANALYTICAL TECHNOLOGIES USED IN INDUSTRY

摘要

A significant source of atmospheric pollution is the emission of sulfur from the combustion of fossil fuels. In addition, sulfur in fuel can reduce the efficiency of catalysts containing the platinum group elements such as those in car exhaust catalysts, leading to increased emissions of other pollutants such as NO_(2). Despite concerns about environmental impact, fossil fuels remain a significant source of energy. In 2001, the UK alone produced 106 million tonnes of crude oil from offshore sites [1]. European Community directives concerning the sulfur content of liquid fuels reflect the increasing concern about the effect of sulfur on the environment. In 1993, the legislation for sulfur content was 500 and 2000 (mu)g g~(-1) for petrol and diesel fuels respectively [2]. The current limits (set in 1998) are 150 and 350 (mu)g g~(-1) respectively and the limit for both fuels is set to be reduced to just 50 (mu)g g~(-1) by the year 2005 [3]. Similar legislation exists in other developed countries. The petroleum industry while wishing to meet legislative requirements, is also keen to avoid the additional costs of removing sulfur below the legal limit. This is a major economic reason why the industry needs high accuracy sulfur measurements. Routine techniques used in the industry for the analysis of sulfur in fuel are being put under increased pressure with the establishment of these dramatically reduced limits. This paper describes how developments in the high accuracy analysis of sulfur in fossil fuels by IDMS have lead to the support of other more routine techniques. This has been achieved by providing reference values for a round robin exercise and by the certification of reference materials. These reference materials are in the concentration range of 1 - 450 (mu)g g~(-1), which encompass all of the planned legislative limits within the EC and also any likely further reduction in these limits. Together, the reference values and reference materials provide a quality assurance of routine techniques and of their day-to-day operational performance. A European working group (WG27) was set up to evaluate which analytical techniques were best suited to meet the new European Commission levels in 1998. The working group organised a round robin exercise (1998-1999) [4] in which 85 laboratories across Europe analysed 15 petrol and diesel samples in a concentration range 50 - 500 (mu)g g~(-1). These laboratories used one of five recognised techniques for this analysis: wavelength dispersive X-ray fluorescence, energy dispersive X-ray fluorescence, Wickbold combustion, microcoulometry, and combustion/ultraviolet fluorescence. The round robin exercise was used to compare each of the methods in terms of repeatability, reproducibility and accuracy of results relative to each other. In October 2000, the working group decided to organise a second round robin exercise in order to define the precision of test methods in the concentration range 1 - 60 (mu)g g~(-1) [5]. This round robin was carried out during 2001 and consisted of 8 petrol and 8 diesel samples sent to 102 laboratories across Europe. The laboratories used one of the 5 analytical techniques as for the previous round robin study, with the exception of Wickbold combustion, which was excluded due to the poor precision of the results in round 1. Two of the round robin samples (one each of petrol and diesel) were considered "quality control" samples because they had been assigned a reference value by IDMS analysis. The IDMS results were of the highest accuracy with total expanded uncertainties (with a 95percent confidence interval) of only 2.6percent and 2.0percent for the petrol and diesel samples respectively. This was an important development for the working group since the IDMS analysis was used to provide reference values for these samples and allowed the group to define the absolute performance of the other techniques rather than just the accuracy relative to each other. Results from