Spcfrroc&fica Acrrt. Vol. 408. No. 8, pp. 1139-l 141. 1985. Pn’ntcd in Great Britain.

0584-8547/85 s3.w+ .oo Peganon Press Ltd.

TECHNICAL NOTE

The determination of molybdenum and tungsten in solutions by thin-film X-ray- fluorescence spectrometry

(Received. 20 June 1984, in reuisedfonn 13 February 1985)

1. INTK~~UCTION

THE DETERMINATION of MO and W in process-plant solutions is not easy with conventional analytical techniques such as atomic absorption spectrometry (AAS), ind~ively~upl~ plasma atomic- emission spectrometry (ICP AES), and X-ray-tluorescence spectrometry (XRFS). &cause of this, a thin-film XRF technique was developed for the purpose.

The samples consisted of various concentrations of MO and W in alkaline and ammoniacal solutions containing a high concentration of sodium sulphide or sodium hypochlorite. Initial investigations had indicated that considerable analytical variations occurred owing to matrix effects over the concentra- tion range for W and MO from 50 mgf- ’ to 8 gl-‘. The solutions were not stable, and tended to precipitate if allowed to stand too long or when subjected to X-ray irradiation.

A technique in which filter paper is used as the support was first proposed in 1954 [l] and has been used mainly in pollution studies and also in the determination of palladium in titanium [2]. A hydrophobic wax ring impregnated into the filter paper was found [3,4] to reduce the spreading of the sample, thus improving its sensitivity. Although it has been reported [SJ that several sources of error are inherent in the technique, it was felt that, with careful control, these could be. minimized or eliminated,

2. EXPERIMENTAL

A Siemens SRS200 X-ray fluorescence spectrometer equipped with a gold target tube and operated at a potential of 60 kV and a current of 40 mA was used for all the measurements, together with a LiF (110) cut crystal and scintillation detector. The KK lines of MO, Ru, and Zn and the La, line of W were used for these analyses.

Background intensity m~surements were recorded at 27.50” 28 and 64.50” 28.

2.1. Dicme~er of the sample With processing-plant solutions containing widely varying concentrations of anions and cations, it is extremely

difficult for samples and standards to be matched. When these solutions are spotted and allowed to dry on filter paper, the elements present concentrate at different distances from the point ofapplication because variations in the solutions cause differences in the chromatographic effects. This can seriously affect the accuracy and precision of the analysis even if the sample is spun during the measurement, since the measured sensitivity will vary across the sample [6]. The sample should therefore be confined to the area that gives a near-linear response (Fig. 1). This can be achieved by the impregnation of the filter paper with a hydrophobic wax ring [3] or, more simply, by the use of filter paper of an appropriate diameter. The optimum diameter of the samples for the Siemens SRS200 instrument was determined by the recording of the change in intensity as a l-mm2 piece of tin foil was moved across the face of the sample holder. A plot of the normalized local intensity versus distance from the centre is shown in Fig. 1. When the sample was restricted to a diameter of 15 mm instead of the normal 34 mm, improvements of up to a factor of two in the limits of detection and sensitivity were obtained. The results also became more accurate, partictdarly in the

analysis of samples that had a tendency to precipitate (Table I).

2.2. Sample loading and matrix elg,,ts The effect of an increase in the amount of sample on the paper was determined by the spotting of 100,200 and

400~1 of a 500-mgl- ’ solution of MO and W onto filter paper and by a comparison of the resultant intensities. No significant change in the sensitivity of the MO KK~ spectral line was observed, but the sensitivity for the W Lu, line

changed from 10.0 to 8.8 c/s/fig. This is a significant change in that the analytical precision is better than 3 “/, This

[I] H. G. PFEIFFER and P. D. ZEMANY, Nurure, Zmd. 4426, 397 (1954). [2] K. IWASAKt, Anal. Chim. Acta 110,67 (1979). [3] G. ACKERMAN, R. K. KOCH, H. EHRHARDT and G. SANNER, Tulanta 19,293 (1972). [4] J. Skim and R. VAN GRIEKEN, Awl. Chim. Acta 88, 97 (1977). [5] G. K. H. TAM and G. LACROIX Anal. Lett. lS(A17), 1373 (1982). [6] E. P. BERTIN, Principles and Practice &X-Ray Spectrometric Analysis. Plenum Press, New York (1970).

1140 Technical note

1 I I I 5 10 I5 20

Distance from centre, mm

Fig. 1. Effect of sample diameter on intensity.

Table 1. Analytical results for samples

Concentration Molybdenum Tungsten

MO W 34* 158 34. 15; SampleNo. (mgl-t) (mgl-‘) (mgl-‘) (mgl-‘) (mgl-‘) (mgl-‘)

1A 4ooo 4ooo 3994 4083 4333 3956 1B 3961 4034 4104 4063 1c 5899 4369 4025 4269 2A 4000 500 4043 4126 529 501 2B 3906 4031 508 480 2c 6136 4246 509 514 3A 500 4000 529 534 4113 4140 3B 522 530 4109 4051 3c 745 560 4053 4022 4A 50 50 41 44 57 54 4B 38 41 51 43 4c 49 45 49 45

*Diameter of filter-paper disc (mm). A: Samples contain only MO and W in aqueous solutions. B: Samples contain MO and W in approximately 50 ml I-’ solution. C: Samples contain MO and W in approximately 90 g l- 1 Na$ solution.

matrix etlect can be compensated for either by the extrapolation of the results for more than one filter paper [7] to those for a sample with a thickness ofzero, or by the use of an internal standard. The latter approach was used in this study and was found to increase the types of sample matrix that can be analysed. For MO, the precision of analysis was improved by up to a factor of two. This improvement was due to the effect of the internal standard in correcting for small errors in the positioning of the sample disc.

Nb and Ru can be used as internal standards in the determination of MO provided that there are no absorption edges (from elements in the sample matrix) between the spectral lines of the internal standard and those of MO. As Nb is not easily kept in solution, Ru was chosen as the internal standard. Ru is suitable provided the sample contains no Nb ot Zr.

Zn was used as the internal standard for the determination of W. Hf or Ta would normally be preferred but, like Nb, these elements are not easily kept in solution. Fluctuations in the tube. potential or current can reduce the effectiveness of Zn as an internal standard, since the Au La, tube-spectrum excites the Zn Ka spectrum but not the W Lu, spectrum. However, with modern instruments, such fluctuations should not occur.

2.3. Sample preparation and analysis

An amount (0.25 ml) of sample was mixed with an equal amount of an internal-standard solution containing Ru (900 mgl- ‘) and Zn (2000 mg l- I). Self-zeroing pipettes of the “Lang Levy” pattern were used for the measurement

[7] E. J. FELTEN, I. FANKUCHEN and J. STEIGMAN, Anal. Gem. 31(11) 1771 (1959).

Technical note 1141

of these aliquot portions. A No. 41 Whatman filter paper of 15 mm diameter was placed on a sheet of Mylar film, and 0.2 ml of the mixture was pipetted onto it and allowed to dry. A second sheet of Mylar film was placed over the first, and the sample and the Mylar film were clamped in place in the solution-cell holder. Synthetic standards containing the analyte elements in the concentration range 10-6000 mg I- ’ were made up in an aqueous solution containing 50 g 1-l tartaric acid and 50 mg I-’ ammonium hydroxide. The samples and standards were measured with the instrumental parameters discussed earlier.

The net peak intensities for the analyte and internal-standard elements were determined by multiplication of the intensity of the background at the measured positions by a predetermined constant factor (background factor). This factor is the ratio of the intensity of the peak position to that of the background in a sample containing no analyte or internal standard. The ratio (R) of the net peak intensity of the analyte to the peak intensity of the internal standard was then determined This ratio is a function of the analyte concentration and is usually almost linear. A slight improvement in the accuracy of values for concentrations of the analyte below 100 mgl- ’ was obtained by use of a second-degree equation for the calibration function rather than a simple linear equation. Because no spectral interferences were encountered, no corrections were necessary.

The procedure was tested on a series of twelve synthetically prepared samples. These contained MO and W made up in an aqueous medium in 50 ml I- ’ sodium hypochlorite, or in 50 g I- ’ sodium sulphide. These solutions were spotted onto papers of 34 and 15 mm diameter respectively. The results are shown in Table 1. The precision was calculated after the analysis of ten portions of samples containing MO and W at concentrations of 50and 2000 mg I - ‘, respectively. The precision of the MO analysis varied from 0.048 to 0.007 (relative standard deviation, s,) while the precision of the W results varied from 0.056 to 0.012 (s,), depending upon concentration. The limits of detection [S] were calculated as 0.63 pg for MO and 0.37 pg for W on the paper. In addition to samples with the matrix variations shown in Table 1, the technique was found to be suitable for the analysis of samples containing ammonium chloride in concentrations up to 200 g I- ’ without any change in the calibration standards.

3. DISCUSSION

Thin-film X-ray-fluorescence spectrometry has proved to be very useful in the routine analysis of plant solutions generated in the mineral-processing industry. The technique eliminates the danger of leakages in the sample cells, which result in the damage of costly X-ray tubes, and is therefore suitable for unattended measurements to be made throughout a 24-h period. Larger variations in the type of sample can be accommodated with this technique than with methods involving the direct measurement on solutions. The differential chromatographic effect must be reduced by the optimization of the sample size for the type of instrument being used because this is probably the foremost potential source of error. The maximum sample mass that can be measured must also be known for each type of sample.

SUMMARY

Thin-film X-ray-fluorescence spectrometry was applied to the analysis of solutions containing MO and W in concentrations ranging from 50 mg 1-l to 6 gl- ‘. The diameter of the sample and its corresponding effect on minimizing errors due to a differential chromatographic effect are discussed. The effect of sample loading was also investigated. The precision of analysis varies from 0.7 to 5.6 %, and the limit of detection for MO is 0.63 pg of element on the filter paper.

Acknowledgement-This paper is published by permission of the Council for Mineral Technology (MINTEK).

Council for Mineral Technology Analytical Science Division Private Bag X301 5 Randburg 2125 Republic of South Africa

B. T. EDDY and A. M. E. BALAES

[S] R. JENKIN$ An Introduction to X-Ray Spectrometry. Heyden, London (1974).