An Evaluation of Portable X-Ray Fluorescence for … XRF Handheld X-ray fluorescence (HH-XRF) is one...
Transcript of An Evaluation of Portable X-Ray Fluorescence for … XRF Handheld X-ray fluorescence (HH-XRF) is one...
An Evaluation of Portable X-Ray Fluorescencefor Artifact Sourcing in the Field:
Can a Handheld Device DifferentiateAnatolian Obsidian Sources?
Ellery FrahmDepartments of Anthropology and Geology & Geophysics
University of Minnesota - Twin Cities
Geological Society of America2007 Annual Meeting, Denver
Session: Sourcing Techniques in Archaeology
The Problem• Analytical instruments not portable
• Heavy, electricity and water, dust-free, etc.
• Must obtain export permissions for analysis
• Most are “destructive” in some way
• Powdered, polished, radioactive, pitted, etc.
• Ideal: accurate, nondestructive, fast, field
• Very few field-capable instruments
Ultra-Portable XRF
Handheld X-ray fluorescence (HH-XRF) isone of a few techniques showing promisePictured: Thermo Fisher Scientific NITON analyzers
XLt seriesXL3t series
Disclaimers• HH-XRF sold by different companies
• Innov-X, AppliTek, Oxford Instruments, etc.
• Likely differ in features, performance, etc.
• Only NITON analyzer evaluated
• Has about 80% of HH-XRF market share
• Lead-laden toys delayed further tests
• Second round of testing planned, delayed
AA
BB
CC
Typical UsesA: Pb Testing
Surveying houses, paint
EU’s RoHS compliance
B: Soil SurveyingMapping toxic metals
EPA-regulated surveys
C: Metals & AlloysMaterials identification
Scrap yards, recycling
Results of Its Pedigree• Rugged, easy to use: “point and shoot”
• Used on receiving dock, in scrap yard, etc.
• Fast, low cost/sample, no sample prep
• 1000s of analyses in one eight-hour shift
• Nondestructive analyses* *Produces heat in samples
• Often used in “pass/fail” mode
• Does a product meet RoHS standards?
ArchaeologyNITON advertises
Conservation/RestorationReconnaissance surveysResource management NAGPRA compliance
[Relative] Dating AuthenticationProvenance
HH-XRF Weaknesses• Low-resolution spectrometer: 300 eV
• SEM-EDS: 150 eV, EPMA-WDS: 5 eV
• Peak overlaps, element misidentifications
• Low-energy X-rays absorbed by air
• No Si, Al, Mg w/o extra equipment; no < Mg
• Radioisotope models heavily restricted
• ≤ 25 elements; limited by model, source
From NITON Literature
Ni & Fe in rock: Ni low by about 3%, Fe high by 14%**Error fits what is expected due to matrix effects; under-corrected data?
Is this good enough for sourcing purposes?
Ni Fe
Testing Details• Evaluated “off-the-shelf” model
• XLt series analyzer, X-ray tube source
• 5-min analyses, portable analyzer stand
• 21 elements were factory-set
• Factory-set calibration: “bulk/soil” mode
• Claims to correct for geometry, matrix, etc.
Samples Analyzed• NIST standards: three glass SRMs
• Included standards of different diameters
• 15 obsidian samples analyzed w/ INAA
• Comparison of INAA and NITON analyses
• 600 obsidian samples, 90 collection areas
• Are sources/regions clearly differentiated?
• 11 of 21 pre-set elements consistently detected
Standards: NITON vs NIST• SRM 1412 - Multi-component glass:
• Min %RSD = 22% for K; Max = 80% for Sr
• SRM 93a - Borosilicate glass:
• Min %RSD = 43% for Zr; Max = 310% for K
• SRM 610 - Trace-element glass:
• Min %RSD = 4% for Ca; Max = 230% for K
• Data often off by factor of 2-3; more at times
Sample Size Effect
Same thickness: 3 mm
Only diameter changed:30 mm (“big”) vs 5 mm
Obsidian: NITON vs INAA• 15 sub-samples from same collection set
• Error varies wildly by element and sample:
• K: Min %RSD = 3%, Max %RSD = 170%
• Zn: Min %RSD = 1%, Max %RSD = 120%
• As: Min %RSD = 9%, Max %RSD = 86%
• Zr: Min %RSD = 14%, Max %RSD = 120%
• INAA databases likely incompatible with NITON
Clusters “smeared” dueto confounding effect?Size/shape? Density?
Calibration error?
Problems Identified• Deviations from NIST & INAA analyses
• RSD sometimes < 10%; other times > 300%
• Incorrect element identifications
• 100% Au misidentified as 80% Nb & 20% W
• Density/matrix estimates imperfect
• 99.999% Cu reported as almost 600% Cu
• Definite size effect; shape effect possible
Use SuggestionsMost useful as a “first sort” tool:
• Analyze 100s or 1000s of artifacts on-site
• Explore data for initial chemical clusters
• Identify representative samples
• Request fewest samples possible for export
• Supply HH-XRF analyses in paperwork
• Conduct subsequent lab-based analyses
NITON Recommends…• From their literature: “The quality of data produced by
field XRF varies with site conditions, soil composition,and sample preparation. Quality assurance protocolsfor the field method usually require that a number offield samples be split and sent to a laboratory forconfirmatory analysis.”
• “Although portable XRF analysis was once thought tobe less reliable than traditional lab-based techniques,the US EPA now acknowledges that inspectors shoulduse the data to ‘adjust their testing strategy for theproperty’ in real-time to investigate unusual readings.”
Future Work• Further testing with custom calibrations
• Better accuracy, precision for sourcing?
• More compatible with existing databases?
• Analyze set of known rhyolitic glasses
• Evaluate factory calibrations, seek errors
• How can errors be minimized by users?
• Establish adequate sample size, shape?
Conclusions• Not a replacement for lab-based analysis
• Likely incompatible with extant databases
• Separated 2/3rds of sources; others overlap
• Overlaps have serious archaeological implications
• Uses must play to its strength: portability
• Most useful as “first sort” field tool
• HH-XRF shows potential for the future…
• …but be wary of claims in the sales literature