1.2 XRF Data

The XRF data can be split into two groups — “raw” and “processed” data. The raw data is contained in a separate folder called XRF data, and consists of a single file, beginning with L000000.spe and incrementing sequentially. This file can be read using a text editor and is tab-delimited. The first part is a header, containing metadata information. The second part is a table of all the channels of the detector and the corresponding count for each channel. Increasing channel numbers represent increasing energy, but some thought needs to be given to calibrating channels into an energy — this step is usually performed using specialist software. In addition, a file called sumspectra.spe is often included in the root directory; this is simply the sum of all the *.spe files in the XRF data directory, and is sometimes useful in processing the data.

Processed data comes from the Q-Spec software (Cox Analytical Systems, Sweden) provided with the machine. Its function is to process the spectral data files (*.spe) into peak-areas for each element of interest by fitting a model to the data. The model needs some user input and intervention to optimise it, and the quality of the model can be assessed using a number of diagnostic parameters, the most important being the root-mean-squared-error (RMSE). The Q-Spec software can also perform some quantitative calibration of the data, although this is a less typical use-case. Often the operator will include a file that contains all of the settings used by Q-Spec to translate the raw data files into the peak area output file — this file will have the extension *.dfl and will often simply be called settings.dfl.

The processed elemental data comes in the form of a text file comprising of a tab-delimited table, with a single row for each measurement step, and a column for parameters including individual element peak areas in counts (n) or intensities (n/mA). The data files commonly have names like result.txt or Results.txt, but may have been subsequently renamed. These are the ED-XRF data most commonly worked with by analysts.