Use of X-ray Fluorescence Analysis for the Determination of Rare Earth Elements

3.1 Energy-dispersive X-ray fluorescence analysis

With EDXRF, the sample is excited by the X-ray tube directly or through a filter (Figure 3). A semiconductor detector (e.g., silicon–lithium semiconductor, pin diode or silicon drift chamber) analyzes the X-ray fluorescence radiation that comes directly from the sample.

Figure 3:

A schematic setup of an energy-dispersive X-ray fluorescence spectrometer with direct excitation.

Here, the detector together with the associated electronics counts and sorts, according to energy, all of the photons that reach it. A pulse height spectrum that indicates the number of photons or impulses for a given energy is established. The detector typically has only a few μ s for processing, so that in case of a modern silicon drift detector (SDD), processing is accordingly limited to approximately 1,000,000 pulses/s. Using a filter, a portion of the exciting radiation can be screened out to avoid overloading the detector.

XRF with polarizing excitation represents an alternative design for EDXRF (Figure 4). Here, the radiation coming from the tube is deflected by 90° and is then used to irradiate the sample. The detector must be perpendicular to the plane determined by the tube, target and sample. The most important effect is that by deflecting the X-ray radiation by 90°, the radiation is polarized and the spectral background in the spectrum is reduced.

Figure 4:

4c vector – “Configuration of an energy-dispersive X-ray fluorescence spectrometer with polarized excitation”.

The analysis of REE by EDXRF suffers from big overlaps between the spectroscopic lines so that detector resolution becomes the limiting factor.

3.2 Wavelength-dispersive X-ray analysis

WDXRF uses, like some EDXRF, direct excitation (Figure 5). Here, it is also possible to work with filters to block or weaken components of the excitation radiation. The major difference is the method of detection of the X-ray fluorescence radiation for WDXRF. Using a goniometer, only one wavelength from the spectrum is fed to the detector, i.e., it measures only one line from one element. In order to conduct multiple element analyses, it is necessary to create a serial measuring program that drives to and analyzes all of the lines of interest, one after another. There are, however, so-called simultaneous spectrometers: In this case, there is a set channel for each element consisting of a fixed crystal with corresponding detector arranged around the sample. When combined with a goniometer, this forms a very fast, high-performance XRF instrument that is especially useful for process control.

Figure 5:

Schematic setup of a wavelength-dispersive X-ray fluorescence spectrometer.

Most of the XRF spectrometers nowadays are equipped with an X-ray tube with a target “Rh”. Nevertheless, for the analysis of REE, an X-ray tube with targets “Au” and “W” would be a better choice than Rh to excite the K- and L-lines of the REE [10].

The analysis of REE by WDXRF still remains a challenge even if the resolution of the goniometer is much better than for EDXRF. There are still a lot of line overlaps which cannot be totally resolved. The final performance of the XRF analysis of REE is therefore strongly depending on the availability of overlap-free analyte lines which could be used for the determination.

3.3 Comparison of EDXRF–WDXRF

In general, it is possible to say that the two techniques are complimentary; one supplements the other. EDXRF has a time advantage, as all elements are measured simultaneously, whereas the (serial) WDXRF measures the elements one after another.

WDXRF has a resolution and sensitivity advantage which is especially useful in the range of atomic numbers up to 30 and 55–80 which also includes most of REEs.

3.4 Other XRF techniques

4.1 Pressed pellets techniques

Pressing a sample leads to a defined density in the pellet. This guarantees reproducible sample preparation.

The Following procedures are used to press samples:

1. Direct pressing

2. Pressing in aluminum trays

3. Pressing in rings

4. Pressing with binders

5. Pressing onto a backing material

Manually mixing with mortar and pestle is arduous and not very effective. This is why, I recommend using a simple mixer, whereby the high reproducibility is also ensured.

4.1.1 In practice: Making a pellet

The complete procedure for making a pellet, from preparation of the sample to pressing, is shown in Figure 6 [2].

Figure 6:

Illustration of all steps to produce a pressed pellet with binder.

After grinding the sample, 4 g of the sample is mixed with 1 g binder (e.g., wax [2]). Before pressing, the pressing tool can be covered with a pellet film, which prevents subsequent sticking, simplifies cleaning and prevents contamination of the pellet from the hardened steel (Cr, Fe). Aluminum cups help stabilize the pellet and enable easy labeling but have no additional usefulness. Table 2 presents figures based on experience for the pressure required depending on the diameter of the pressed pellet. Here, it is important to remember that the pressing procedure must be reproducible, i.e., the same density must be obtained in each pressed pellet. Direct pressing from 0 to, e.g., 30 tons can prevent the air contained in the sample from completely escaping. This leads to cracks in the pressed pellet. Use of a lower pressure or a pressing procedure in several steps is recommended.

Table 2:

Recommended pressuredepending on the diameter ofthe pressed pellet.

Diameter	Pressure (t)
32 mm	15–20
40 mm	20–30
35-mm ring inside	max. 30

When there is not enough sample material, but a stable pressed pellet is still required, then the sample can be pressed onto a backing. In literature [13], this is presented for yttrium oxide: 400 mg of sample (dissolved in acid and precipitated as oxalate) is mixed with an equal amount of boric acid and then pressed on a boric acid layer as backing. However, since December 1, 2010, boric acid is classified as GHS08 health hazard which means that it should be replaced by a non-hazardous material. In Ref. [2], a cellulose-type material called BOREOX is presented as a good non-hazardous replacement.

It has to be mentioned that sample thickness should always be higher than the penetration depth of the XRF line used for calibration. Table 3 lists recommended minimum sample thicknesses or masses depending on K- or L-lines for REE analysis.

Table 3:

Minimum required sample thickness [2] or mass [10] of a pressedpellet for REE analysis depending on the analyte line and pellet diameter.

REE XRF lines	Minimum sample thickness (mm)	Minimum mass/diameter
L-lines	0.5	2 g/D30 mm
K-lines	8	20 g/D30 mm
		40 g/D40 mm

4.2 Fusion technology

In XRF, most solid samples are prepared as pressed pellets, whereby the achievable precision suffers from so-called particle size effects. They describe the phenomenon that the intensity of X-ray fluorescence is depending on particle size and shape. If the achieved precision with pressed pellets is not sufficient, the sample must be prepared using a fusion process.

The fusion provides an ideally homogeneous sample with a defined density and without particle size effects. In addition, homogeneity and a perfect surface lead to a much smaller calibration error.

Figure 7 shows the fundamentals of the fusion process.

Figure 7:

Principle of the fusion process.

The following conditions must be fulfilled for fusion of a sample:

1. Finely ground oxidic sample material

2. Borate as the fusion material and glass-forming agent

3. Platinum–gold crucible

4. Platinum–gold casting dishes

5. Temperatures higher than the borate melting point

For ICP-OES or ICP-MS measurements, the melted sample could be cast directly into the acid instead of casting a bead.

Further sample preparation methods using fusion known from literature:

Rock samples are mixed 1:5 with lithiumtetraborate and fused at 1,000°C in a muffel furnace. Then dissolved in HF and HNO₃ followed by a separation of REE from the matrix by anion exchange and subsequent preconcentration. Finally, the REE are coprecipitated with rhodizonate and tannin. The precipitate is collected on membrane filter which is measured with XRF [14].

To fuse REE rock samples, three different dilutions between sample and flux are suggested by Nakayama and Nakamura [15]:

A ratio of 1:1 for REE, 1:2 for Th and U and 1:10 for major and minor elements. It is a challenge to perform a low-dilution fusion with a ratio 1:1 to avoid inclusions and bubbles. Therefore, the fusion had to be repeated a second time and LiCl was added to avoid devitrification. The LLDs achieved with this method are in the range of 0.6 (Th, U) and 0.7–6.5 μg/g for REE.

The authors of Ref. [16] describe a method for raw materials with major and minor concentrations of REE: the REEs are separated from the main matrix by dissolving them with HF/HNO₃ and Na₂CO₃/Na₂O₂ and final precipitation as oxalate. Then, a fusion of the residue with lithiumtetraborate was done. Cr and Mo were used as internal standard. Results are presented for light REEs, Y and Th over a wide concentration range (La, Ce, Nd, Pr 0.02–25 %; Sm, Gd, Eu, Y and Th 0.005–1 %).

In Ref. [10], monazite samples were fused with pure lithiumtetraborate or a mixture of lithiumtetraborate and lithiummetaborate (90:10) in a ratio of 0.5 g sample and 9.5 g flux and a few drops of LiBr (25 %) as releasing agent.

4.3 Additional sample preparation techniques

Roelandts [17] describes the analysis of REEs in apatit. The sample is dissolved in HNO₃, then the REEs are separated by a mixed solvent anion exchange and kept by the resin which is then analyzed by XRF on top of a disk.

Another preconcentration method in combination with a thin film measurement is presented in Ref. [18]: 1-g sample was fused with 4-g Na₂O₂ and finally dissolved in acid. The REEs were separated from the main components by ion exchange. The XRF measurement was done at an ion-exchange paper.

Also in Ref. [19], the XRF measurements were performed directly on a filter. After digestion of the sample with acid, the REEs are preconcentrated by using an arsenazo III complex which was collected on a filter by chemofiltration.

5.1 Reference materials

To perform and to verify an XRF calibration, reference materials are required. Table 4 lists the reference materials mentioned in the literature cited in this article. Most of them are already sold out. However, there are replacements and many other reference materials which could easily be explored by free accessible databases in the internet. Table 5 gives an overview.

5.2 Measuring parameters

Starting the configuration of an XRF spectrometer for the determination of REE, first the spectral line for calibration must be selected. REEs have three K-lines and about seven L-lines which could be used. However, the sensitivity of the analyte lines differs a lot. From the K-line series, the Kα1 line and for the L-lines, the Lα1 and Lβ1 show the highest sensitivity.

K series lines could be used for La, Ce, Pr, Nd, Sm if pressed pellets are used as sample preparation. Based on the large penetration depth, particle size effects will be reduced.

Figure 8 shows the K series lines for La to Sm measured with WDXRF equipped with a target X-ray tube “Rh”, 60 kV excitation, lithium fluoride (LIF) 220 crystal, 0.15° fine collimator and SC as detector. The graph demonstrates how the sensitivity decreases from La to Sm.

Figure 8:

Overlapped K-line spectra of REE measured at 40,000 μg/g single element samples from FLUXANA, WDXRF, Rh tube 60 kV, LIF 220, 0.15°, SC.

For comparison, Figure 9 shows the K series lines of the elements La to Sm measured with an EDXRF equipped with a target X-ray tube “W”, 60 kV excitation combined with a Al₂O₃ polarizer target and a Silicon SDD. The resolution is better than in WDXRF and there are less line overlaps in the spectrum. Based on the fact that the concentrations for the REE shown in Figure 8 are 40,000 μg/g each and the concentrations for the REE shown in Figure 9 are only 100 μg/g each, it can be seen that also the sensitivity is much higher.

Figure 9:

Overlapped K-line spectra of REE measured at 100 μg/g single element samples from FLUXANA, EDXRF, W tube, 60 kV, Al₂O₃ polarizer, SDD.

L-series lines must be used with fusion disks because of infinite thickness limitations. Figures 10–12 show the complete L-line spectra of La, Gd and Lu as an example measured with WDXRF, Rh target X-ray tube, 40 kV excitation, LIF 220 crystal, 0.15° fine collimator and flow counter detector (FPC). The L-lines show always the same line pattern (out of the Lβ4 line).

Figure 10:

L-series lines of 100 μg/g La in cellulose measured with WDXRF, Rh-tube, 40 kV, LIF 220, 0.15°, FPC.

Figure 11:

L-series lines of 100 μg/g Gd in cellulose measured with WDXRF, Rh-tube, 40 kV, LIF 220, 0.15°, FPC.

Figure 12:

L-series lines of 100 μg/g Lu in cellulose measured with WDXRF, Rh-tube, 40 kV, LIF 220, 0.15°, FPC.

For comparison, Figure 13 shows the overlap of single element L-line spectra of the REE measured with an EDXRF with W target X-ray tube, 40 kV excitation with a Mo secondary target and an SDD. However, the resolution of the detector is not sufficient to resolve the different lines.

Figure 13:

Overlapped L-line spectra of REE measured at 100 μg/g single element samples from FLUXANA, EDXRF, W tube, 40 kV, Mo secondary target, SDD.

The most difficult work in creating a measurement program for REEs is the selection of the analyte lines and the corresponding background positions. Because of the high numbers of line overlaps, a general recommendation is very difficult. Figures 14–17 show a full scan of all analyte lines in the region of the REE lines. This measurement was performed on the glass drift monitor sample FLX-S13 from FLUXANA which contains most of the elements which show lines in this spectrum. The detailed study of all line marks gives a good impression of where to select the analyte lines and the corresponding background positions for the required application.

Figure 14:

Scan 1/4 of FLUXANA drift monitor sample FLX-S13 with WDXRF, Rh tube 40 kV, LIF 220, 0.15°, FPC.

Figure 15:

Scan 2/4 of FLUXANA drift monitor sample FLX-S13 with WDXRF, Rh tube 40 kV, LIF 220, 0.15°, FPC.

Figure 16:

Scan 3/4 of FLUXANA drift monitor sample FLX-S13 with WDXRF, Rh tube 40 kV, LIF 220, 0.15°, FPC.

Figure 17:

Scan 4/4 of FLUXANA drift monitor sample FLX-S13 with WDXRF, Rh tube 40 kV, LIF 220, 0.15°, FPC.

Table 6 lists all X-ray fluorescence lines of REE which show relevant intensities and Figure 18 lists all line overlaps which have to be taken into account for REE.

Figure 18:

X-ray fluorescence line overlaps relevant for REE in the range of 3–10 keV [2].

5.3 Analyte lines

Table 7 gives an indication which analyte lines should be selected for the analysis of REE. The selection considers a minimum of line overlaps which have to be taken into account by spectrum scans or the measurement of single element standards. Also, background positions have to be chosen based on individual measurements on the samples which have to be investigated.

A complete measurement program for WDXRF equipped with target x-ray tube “Rh” to measure rock samples is presented in Ref. [15].

5.4 Lower limit of detection (LLD)

Figures 19 and 20 show LLDs of REE in different matrices and measured with WDXRF equipped with a target x-ray tube “Rh” and EDXRF equipped with a target x-ray tube, W both working with 60 kV. While EDXRF shows here a higher sensitivity for the K-lines, WDXRF shows a higher sensitivity for most of the L-lines.

Figure 19:

LLD of REE measured with Kα 1-line (WDXRF: 20 s per line in lithiumtetraborate matrix, EDXRF: 600 s in cellulose matrix).

Figure 20:

LLD of REE measured with Lα 1-line (WDXRF: 20 s per line, EDXRF: 600 s, both in cellulose matrix).

6.1 Other calibration strategies mentioned in literature

Macháček and Weiss [16] describe for the calibration of REE an empirical correction in combination with internal standard elements Cr and Mo.

Michelsen [22] gave an overview about the use of an internal standard like Sr for Y, the use of standard addition for REE in rock samples and dried solutions on filters.