AN EVALUATION OF PORTABLE X-RAY FLUORESCENCE FOR ARTIFACT SOURCING IN THE FIELD: CAN A HANDHELD DEVICE DIFFERENTIATE ANATOLIAN OBSIDIAN SOURCES?

Ideally chemical analyses used for sourcing studies could be conducted in the field, quickly, and nondestructively. Unfortunately, most techniques require immobile instruments. One technique with great potential for on-site analysis is portable energy-dispersive X-ray fluorescence (ED-XRF). Thermo Fisher Scientific's NITON analyzer is a handheld ED-XRF device about the size and shape of a cordless drill. It is typically used for environmental monitoring and scrap-metal testing. I evaluated an "off-the-shelf” model of NITON analyzer to see if it could differentiate Anatolian obsidian sources. The analyzer was used in a factory-set “bulk/soil” mode without further calibration, and the 21 elements measured were also preset. NIST standards were also analyzed to determine if its accuracy and precision for trace elements are sufficient for sourcing. Like all ED-XRF systems, this analyzer is susceptible to X-ray peak overlaps, causing misidentified elements and spurious element detection. A sample-size effect was also observed for some elements, as has been previously documented for XRF. Element concentrations were often under- or over-reported, typically by a factor of 2 to 3, although more at times. This strongly underscores the need for users to create custom calibrations for each material (e.g., obsidian based on rhyolitic glass standards) to improve accuracy. Testing established, though, that errors were mostly systematic. About two-thirds of the obsidian sources were clearly distinguished. Other source fingerprints overlapped, and some overlaps had important implications for archaeological interpretation. Also, if the data are only internally consistent, one cannot use them directly with existing source databases. Any archaeological applications of the analyzer must play to its strengths, notably portability. This instrument might be most useful in archaeology as a “first sort” tool: one can analyze hundreds of artifacts on-site, explore the data for chemical clusters, select representative samples from each cluster for export and later lab-based analyses, and supply the initial data in artifact export requests. Portable ED-XRF is one of only a handful of techniques that shows potential for initial on-site chemical sorting of artifacts, and I am planning further tests using my own custom calibrations.