LIMITATIONS IN ANALYSIS BY X-RAY FLUORESCENCE
The technique of X-Ray Fluorescence has been widely used since the mid 1960s as a rapid means of determining the chemical content of materials by comparing the intensities of the spectral lines characteristic of each element with those of a standard sample of known composition. Because of its speed of application it has generally replaced the traditional method of ‘wet analysis’ in which compositions are determined by dissolving the sample in reagents and then measuring the concentrations of the elements present by titration, or measuring the volume of gas evolved (eg for a carbonate).
However, XRF suffers from the drawback that it is difficult (and sometimes impossible) to quantify elements of low atomic number, including carbon and oxygen. In the case of EDX (Energy Dispersive X-ray spectra), as used for example on a scanning electron microscope, the sample is caused to emit x-rays by excitation with an electron beam in a vacuum. In vacuum, the spectra of elements of lower atomic weight can be measured, but again only with difficulty using a ‘windowless’ detector to prevent absorption of the low energy spectra peaks emitted by elements of low atomic weight. Hence, the use of such techniques in analysing ores containing carbonates, as, for example, the Siderite ores of the Weald, require a correction to be made to account for the presence of the carbonate portion of the molecule undetected by the technique.
Because all analyses have to add up to 100wt% or the results would not be comparable with anyone else’s, the computer programs used in XRF determinations assume that each element is from a single compound and use the stoichiometry of that compound to determine the amounts of the ‘missing’ element.
With most commercial computer algorithms it is impossible to do this for two missing elements such as for the carbon and oxygen in carbonates and this seems to have caused some difficulties with the raw data presented in the table.
Since the unroasted ore is mostly Siderite, which is a ferrous compound, the correct way to quantify it is to assume that iron is present as FeO, (not Fe2O3 which is a ferric compound). When all the elements have been assessed as oxides of the appropriate valence state (depending on whether they should be treated as oxides or carbonates) the sum total ‘estimated’ by the program will not add up to 100wt% but the shortfall can be legitimately assumed to be CO2 bound up as compounds in the solid. (Any loss of gaseous products does not come into the analysis in XRF which only analyses what is present as a solid).
Hence, where a figure is given as a Fe2O3 equivalent, we can calculate what the FeO equivalent would be from the respective molecular weights. In the case of the hard, unroasted ore which was initially presented as 52.07wt% Fe2O3, the equivalent in terms of Fe is 36.44wt% Fe which would then lead to 46.9wt% FeO if this had been put into the calculation instead of Fe2O3. Hence the sum of the compositions in the Table becomes 64.99wt% and the other 35wt% can be attributed to CO2, most of which will be associated with the Siderite and some of which will be associated with the CaO as carbonate. Note that pure Siderite contains 37.9wt% CO2.
The same calculation can be done for the soft unroasted ore but X-ray Diffraction analysis, XRD, shows that the crystalline phase of the roasted ore is a mixture of siderite and maghemite (gamma Fe2O3 – which explains the magnetism). This cannot be properly quantified using a single valence state for the iron to derive the equivalent oxide as it contains a mixture of ferrous and ferric iron.
As a consequence, the result of assuming only Fe2O3 is incorrect and the relative amounts of FeO (from FeCO3) and Fe2O3 would need to be determined to arrive at the true analysis.
In practice, whether the oxide is taken to be in the ferrous form as FeO or the ferric form Fe2O3, makes no difference to the total amount of iron present in the ore. In the case of FeO, the iron represents 77.7% of the weight of the oxide ie 56 / (56+16), and in the case of Fe2O3 70.0% of the oxide (56×2) / (56×2 + 16×3). The important thing is that any ‘missing’ weight from the sum of the analysis is attributed to the CO2 present in the ore and not simply normalised to 100%