Sample preparation for X-ray fluorescence analysis VI.

---

Metal samples

Important sources include internal elemental segregation, cavities or defects in the measurement surface, inconsistent surface roughness, contamination, oxidation, and differences in metallurgical structure or history. Because XRF measures the specimen directly, these physical characteristics can affect measured intensities even when the bulk elemental compositions are nominally identical. Matching the preparation and metallurgical condition of unknown samples and calibration standards is therefore essential for accurate quantitative analysis.

Cooling rate affects metallurgical structure and the distribution or precipitation of elements within the alloy. Cast iron, for example, can give different calibration behavior depending on whether it is rapidly chilled or cooled more slowly in a sand mold. The effect can be especially important for lighter elements such as carbon, magnesium, silicon, phosphorus, and sulfur. Standardizing casting and cooling conditions helps ensure that calibration standards and unknown samples have comparable structures.

When segregation makes a polished section unrepresentative of the bulk composition, the sample can be drilled or milled to collect metal chips. These chips can then be remelted and molded to produce a more homogeneous analytical specimen. Another option is dissolution followed by liquid analysis. When remolding is used, the cooling conditions and resulting metallurgical structure should be made as consistent as possible with those of the calibration standards.

A rough surface generally produces lower X-ray intensities than a finely finished surface, and the effect is particularly important for light elements because their characteristic X-rays originate from relatively shallow depths. Experimental measurements on steel showed increasing Ni, Cr, and Si intensities as belt-sander grit changed from 80 to 120 to 240. For reliable quantitative analysis, however, consistency is more important than simply producing the smoothest possible surface: standards and unknowns should have comparable surface finishes.

Aluminum alloys containing more than about 4% silicon can contain silicon as discrete grains or island-like deposits. A lathe can pull these silicon-rich particles from the measurement surface, causing the measured composition to vary each time the surface is refinished. Preparing the surface somewhat more roughly with a file or grinder can reduce this preferential removal and improve calibration accuracy and measurement reproducibility. Rapid chilling during casting is also important because cooling rate affects silicon precipitation and distribution.

The composition and structure immediately beneath the cast surface can differ significantly from the representative chilled material. In cast iron, approximately the first 0.5 mm can form a surface region known as black scale and is unsuitable for analysis. This layer should be removed by grinding. At greater depths, another limitation appears because the material eventually becomes non-chilled; for the example discussed, the useful analytical region extends only about 2–3 mm into the cast specimen.

Yes. Silicon carbide abrasive belts can introduce silicon, while alumina abrasives can introduce aluminum. Abrasive binders can also be problematic; for example, a sulfone-type hardener can interfere with ultra-trace sulfur determination. The abrasive should therefore be selected according to the elements being measured. Grinder or file preparation can be preferable for ultra-trace sulfur analysis when contamination from conventional abrasive belts is a concern.

A freshly ground metal surface begins changing almost immediately through oxidation and adsorption of contaminants from the atmosphere. Airborne dust can contribute elements such as silicon, gaseous contaminants can affect sulfur measurements, and steel can rust under humid conditions. Storage in a desiccator can help maintain stable X-ray intensities for a period of time. If the surface visibly changes color, regrinding before measurement is recommended.