Graphite furnace technology, where are we?
A few years ago it was proposed that characteristic masses for each furnace analyte represented a stable property of the analyte and the furnace system. L'vov has put this proposal on a firm theoretical footing by calculating the characteristic mass values from spectroscopic data, a mathematical model of the furnace and the assumption that the analyte is totally atomised in the modern furnace. His calculated values agree fairly well with the experimental characteristic masses for 23 of the 30 most common furnace analytes. In this paper the stability of the characteristic mass values is considered further. It is also shown that these data provide an important quality assurance opportunity for furnace analysts. An instrumental system that does not adequately reproduce the reported characteristic masses is not likely to be working optimally. The frequent current practice of using the method of standard additions in preference to an analytical working curve to quantify furnace analyses can mask errors, particularly background correction errors, and provide erroneous results. Nevertheless, we are still unable to obtain stable values for the characteristic mass with the precision available with modern instrumentation. The problems and deficiencies which cause this to be true are discussed.