Improving Accuracy when Quantifying Non-Homogeneous Samples Using FlexiSpot

In quantitative Micro-XRF analysis, the information is integrated over the entire irradiated and detected sample area. A homogeneous sample is required for the quantification process to correctly measure its composition.

However, unlike manufactured samples such as glass or steel, non-manufactured samples like geological specimens are mostly heterogeneous in nature with an uneven distribution of elements.

Laborious preparation is required to provide homogeneous samples for traditional XRF analysis. Conversely, spatially resolved Micro-XRF requires minimal or even no sample preparation. However, finding positions may be difficult while carrying out analysis with micrometer-scale spatial resolution.

Here, the quantitative results represent the whole sample, or a specific partial sample area. Powders and milled samples typically show homogeneity at the millimeter scale, but large point-to-point deviations can be observed while measuring at a resolution of tens of micrometers.

M4 TORNADO’s feature FlexiSpot allows for large area analysis in micrometer resolution (Figure 1). It is possible to measure the samples with a spot size of <20 µm as well as with an additional spot size of a few 100 µm.

With this larger irradiated spot, samples with uneven surfaces or irregularly shaped and non-homogeneous sample such as powders can be quantified with high accuracy due to the integration of the information in the analysis over the enlarged detected area.

The M4 TORNADO

Figure 1. The M4 TORNADO

Functional Principle

The M4 TORNADO features a polycapillary lens, which generates a convergent X-ray beam to focus the X-ray tubes radiation. The plane where the beam waist lies is known as focal plane. Here, the beam width for Mo Kα radiation is less than 20 µm. An X-ray beam is typically broader below and above its focal plane. This fact forms the basis for the FlexiSpot feature.

The irradiated area turns out to be larger when the sample is moved to an “out-of-focus” position. When polycapillary lenses’ common beam divergence is taken into account, it is possible to achieve spot sizes of 100–200µm by shifting the sample roughly 2–3mm out of the focal plane (Figure 2).

Functional principle of generating a variable spot size

Figure 2. Functional principle of generating a variable spot size

Sample Preparation

A clay obtained from the Cretaceous-Paleogene-boundary in Petriccio, Italy, was used as the sample. The sample preparation involved milling and then pressing the clay into a pellet utilizing 20% binder. This standard sample preparation method provides reasonably homogeneous material for traditional XRF analysis. However, the sample is shown to be non-homogenous when imaged using an optical microscope with 100x magnification and at a 10 µm scale. The clay pellet sample has two different spot sizes as shown in Figure 3.

Clay sample in 100x magnification showing obvious inhomogeneity. The red circles indicate the set excitation spot diameters of 25 µm and 200 µm.

Figure 3. Clay sample in 100x magnification showing obvious inhomogeneity. The red circles indicate the set excitation spot diameters of 25 µm and 200 µm.

Measurement Conditions

A Bruker M4 TORNADO coupled to a polycapillary lens and an Rh X-ray tube was used to perform the measurements. It features fast data processing, high spatial resolution and a motorized high-speed X-Y-Z stage for positioning the samples. The standard measurement conditions used in this analysis included tube voltage of 50 kV, current of 200 µA, no primary beam filter, and chamber pressure of 20 mbar.

For an inhomogeneous sample, spatially resolved quantitative analysis needs to be carried out on multiple points to obtain a representative value for the whole area. Therefore, it is possible to consider the averaged quantitative values to be the “correct” result. The credibility of the result is provided by the standard deviation of individual results. This standard deviation may be very high for very small analyzed spots, so the results may not meet requirements or be relevant.

As shown in Figure 4a-b, six random positions were identified and quantified in the Multi-Point workspace of the M4 TORNADO software with spot sizes of 25 µm and 200 µm, respectively. The single point measurement time was set to 120 seconds to acquire good statistics for both minor and light elements.

The next step was mapping a 40 x 40 mm section of the sample under the same excitation conditions. The pixel size and the measurement time used were 50 µm and 4 ms per pixel, respectively. Six areas of 75 mm2 each were obtained from the HyperMap data cube acquired (Figure 4c) in order to have an equivalent overall measurement time of 120 seconds as with point measurements.

Mosaic image of the sample with defined measurement positions and areas.

Mosaic image of the sample with defined measurement positions and areas.

Mosaic image of the sample with defined measurement positions and areas.

Figure 4. Mosaic image of the sample with defined measurement positions and areas.

Results

The quantitative results for six measurement positions each quantified with a spot size of 25 µm and 200 µm, as well as six mapping areas are summarized in Table 1. The standard deviation for the quantified compositions are seen to decrease with increasing analyzed sample surface area. However, there is no change observed in the measured concentrations. This trend is visualized for the oxides of the key elements SiO2, CaO, Fe2O3, and Al2O3 in Figure 5.

It is possible to consider the mean and sigma values for the 75 mm2 areas as representative for the sample and prescribed measurement conditions. In the case of smaller analysis areas, the deviations of the measured values become larger, but the “correct” value is very well within the obtained uncertainties.

The increased spot size of 200 µm improves the deviations by a factor of 2 or better in comparison with the standard spot size of 25 µm (Table 1 and Figure 5). It has to be noted that this is relative to an identical measurement time.

Table 1. Measured values for different spot and area sizes

Oxide Spot size 25 μm Spot size 200 μm 75 mm2 area map
Mean value (stoich.wt.% ) Standard deviation (stoich. wt.% ) Relative Standard deviation ( % ) Mean value (stoich. wt.% ) Standard deviation (stoich. wt.% ) Relative Standard deviation ( % ) Mean value (stoich. wt.% ) Standard deviation (stoich. wt.% ) Relative Standard deviation ( % )
SiO2 34.36 2.07 6.0 35.07 0.81 2.3 35.05 0.18 0.5
CaO 30.94 3.55 11.5 30.94 1.77 5.7 30.69 0.20 0.7
Fe2O3 14.67 1.77 12.1 13.75 0.53 3.9 14.26 0.09 0.6
Al2O3 13.76 0.69 5.0 13.97 0.29 2.1 13.76 0.12 0.9
K2O 3.37 0.29 8.6 3.30 0.12 3.7 3.42 0.02 0.6
MgO 1.17 0.16 13.6 1.13 0.08 7.4 1.15 0.12 10.8
TiO2 1.11 0.10 9.4 1.16 0.11 9.4 1.14 0.01 0.9
SrO 0.16 0.02 12.0 0.15 0.01 3.4 0.16 0.00 1.7
MnO 0.14 0.01 9.4 0.13 0.01 5.0 0.13 0.00 2.4
P2O5 0.15 0.08 55.5 0.24 0.19 76.3 0.09 0.04 46.1
ZnO 0.047 0.016 34.9 0.041 0.004 10.5 0.041 0.001 2.7
Cr2O3 0.048 0.015 32.0 0.038 0.004 11.4 0.038 0.003 8.5
NiO 0.045 0.028 62.5 0.038 0.005 14.4 0.028 0.002 7.2
ZrO2 0.019 0.004 22.8 0.021 0.003 16.4 0.022 0.002 9.2
Rb2O3 0.020 0.003 15.6 0.017 0.002 11.1 0.019 0.001 6.8

Quantified values for the main matrix components with two sigma standard deviation (as obtained from six measurements)

Figure 5. Quantified values for the main matrix components with two sigma standard deviation (as obtained from six measurements)

As Figure 4 reveals that only a full area map yields reliable results, it has to be noted that the time taken for data acquisition in this area is five times more than the recording time for the multipoint measurements (1:04 hours in comparison with 12 minutes).

The deviations obtained from the point spectra measurements are fully adequate for most applications, except the results of low concentration light elements Mg and P that do not follow this trend. These variations are counting statistics of the inhomogeneity of the elemental distribution and the fluorescence lines.

Conclusion

The notable improvement of relative standard deviations achieved in large area analysis using M4 TORNADO’s FlexiSpot has been demonstrated by comparing the sample measurements with various spot sizes and areas. Point measurements with 200 µm spot size yield sufficient results in considerably shorter measurement time in comparison with full area maps.

This information has been sourced, reviewed and adapted from materials provided by Bruker AXS Inc.

For more information on this source, please visit Bruker AXS Inc.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Bruker AXS Inc.. (2019, August 28). Improving Accuracy when Quantifying Non-Homogeneous Samples Using FlexiSpot. AZoM. Retrieved on October 18, 2019 from https://www.azom.com/article.aspx?ArticleID=12950.

  • MLA

    Bruker AXS Inc.. "Improving Accuracy when Quantifying Non-Homogeneous Samples Using FlexiSpot". AZoM. 18 October 2019. <https://www.azom.com/article.aspx?ArticleID=12950>.

  • Chicago

    Bruker AXS Inc.. "Improving Accuracy when Quantifying Non-Homogeneous Samples Using FlexiSpot". AZoM. https://www.azom.com/article.aspx?ArticleID=12950. (accessed October 18, 2019).

  • Harvard

    Bruker AXS Inc.. 2019. Improving Accuracy when Quantifying Non-Homogeneous Samples Using FlexiSpot. AZoM, viewed 18 October 2019, https://www.azom.com/article.aspx?ArticleID=12950.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this article?

Leave your feedback
Submit