The UH Hilo EDXRF Method

On this page:

Calibration
- Rb
- Sr
- Y
- Zr
- Nb
- Na2O
- MgO
- Al2O3
- SiO2
- S
- P2O5
- K2O
- CaO
- TiO2
- V
- Cr
- MnO
- Fe
- Ni
- Cu
- Zn
- Ba
- Mo
- La
- Ce

We use a Thermo Scientific ARL Quant'X EDXRF spectrometer with an extended sample chamber and a silicon drift (SDD) detector. The large sample chamber allows us to non-destructively analyze many artifacts that would not otherwise fit in the spectrometer. The specific conditions that we employ have been slightly modified since we first began operating the lab in 2004, and when we make modifications or recalibrate the spectrometer after replacing internal parts, we give each method a new name based upon when it was developed. Our current method is "2-21" (meaning that it was developed and calibrated in February of 2021). The 2-21 method generates quantitative or partially quantitative data for 26 elements: Na, Mg, Al, Si, P, S, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Rb, Sr, Y, Zr, Nb, Ba, Mo, La, Ce, and Pb. The analytical conditions of the 2-21 method are posted below:

Analytical conditions screenshot Analytical conditions

Each condition listed above generates a range of x-ray energy by directing a beam of x-rays from a rhodium (Rh) tube (a non-radioactive source) through filters, or through no filter at all, towards the sample. Atoms in the sample absorb the original x-ray energy by bumping their electrons up to higher orbits. When these electrons return to their normal orbits, they give off their own characteristic x-rays based on the specific orbits of electrons in the elements with which they are associated. Each analytical condition is sensitive to different elements, based roughly on atomic weight. For example, the Low Za condition (6kV with no filter at all) is best for exciting the lightest elements (Na, Mg, Al, Si, S, and P). To run one archaeological sample through the four conditions takes about 30 minutes. The SDD detector in the spectrometer sorts the x-rays generated by the sample into different energy ranges. At the end of the analysis, the spectra show how many x-ray photons were detected at different energy levels for each sample and matched with defined energy regions (regions of interest, or ROI) that are characteristic of specific elements. The spectrum posted below shows the different peak intensities for Rb, Sr, Y, Zr, and Nb for a sample of Mauna Kea adze quarry basalt:

Mid-Z spectrum of a sample of Maunakea basalt

Calibration

To determine the concentrations of each element in a sample (usually measured in parts per million for trace elements, and percentages for major oxides), the peak intensities shown above are compared with peaks generated from samples of known composition. To calibrate the 2-21 method, we have used 22 geological standards. These standards are mostly rock and sediment samples that have been reduced to powders and extensively analyzed in different labs. Eleven of the standards are produced by the USGS (AGV-2; BCR-1; BCR-2; BHVO-1; BHVO-2; BIR-1; DNC-1; GSP-1; QLO-1; STM-1, and W-2). Six others come from Japan (JA-1; JA-2; JA-3; JB-2; JB-3; JSl-2). One additional basalt standard comes from China (NCS-DC73303). The remaining 4 standards are in-house whole-rock standards. We make pressed pellets from the powdered standards using 3g of powder and 3 drops of polyvinyl alcohol binder which are placed in a 31 mm SPEX evacuable pellet die and then pressed in a 25-ton hydraulic press. In the graphs below, you can see how the calculated values for standards compare with the published values. Note that we have a very high degree of agreement between calculated values and published values for the 'Mid-Z' trace elements Rb, Sr, Y, Zr, and Nb, and poorer agreement for the lighter elements. In a perfect world, all the red dots would be on a straight line, but they are not because of a number of factors ('matrix effects' is one of the biggest problems). The graphs thus demonstrate a little about the level of analytical error that can be expected with different elements. Elements analyzed at 50 Kv (particularly Ba) are susceptible to error when comparing pressed powder with whole rock, and to adjust for this, the in-house whole rock standards are relied upon to calibrate for Ba.