EFFICIENCY AND EVALUATION - Total-Reflection X-Ray Fluorescence Analysis and Related Methods - page 396

measurement (PMM). Aliquots of 40 g were forwarded to interested labora-
tories for a charge of € 300, with the so-called reference values undisclosed. Two
hundred thirty-nine laboratories from 43 countries took part in this extensive
proficiency test. Half of them were accredited, and 13 different methods of
analysis were applied. The results had to be delivered before a deadline of six
months. The final report was given by IRMM after eight further months [82].
Klockenkämper etal . participated in the test with the established technique
of flame atomic absorption spectrometry (FAAS) and with the younger
methods of TXRF and ICP-MS [83]. For TXRF, 500 mg of the sediment
sample were poured into 4 ml of nitric acid (65%) and 0.5 ml of hydrochloric
acid (30%). This mixture was digested by high-pressure ashing at a maximum
temperature of 320 ° C and a pressure of 13 MPa. The insoluble residues of
siliceous components were treated with 2 ml of hydrofluoric acid (40%) and
dissolved. After evaporation almost to dryness, the residue was taken up with
10 ml of deionized water. In this way several different sediment samples and
also acid blanks were prepared. For TXRF, aliquots of 1 ml were diluted 1 : 10
and spiked with selenium as an internal standard of 5 mg/l (5 ppm). Droplets of
10 μ l of each solution were pipetted onto siliconized quartz-glass carriers, dried
under IR light, and the dry residues of some micrograms were analyzed by
TXRF with a counting time of 300 s. Altogether, six samples were analyzed via
sensitivity values, which had been determined by aqueous standard solutions
sometime beforehand. Six elements could be determined quantitatively (Cr,
Fe, Cu, Ni, Zn, and Pb) with concentrations between 0.4 mmol/kg (ppm) for
lead and 460 mmol/kg for iron.
The results are demonstrated in Figure 6.5 as a horizontal-bar plot [83].
Thecolorederrorbarsmoreorlessoverlapwiththereferencebars,
demonstrating a high accuracy of the three techniques. There is no evidence
for a systematic error. The relative deviations between measured and refer-
ence values range from 10% to about + 11% for all three methods.
The arithmetic mean is + 0.6% for TXRF, + 1.0% for FAAS, and + 3% for
ICP-MS. The root mean square of the deviations is 6.4% for TXRF, 3.5% for
Figure6.5. Horizontal - bar plot for six different elements of a sediment sample (Cr, Cu, Fe, Pb, Ni,
and Zn) determined by FAAS (light gray), ICP-MS (dark gray) and TXRF (gray). Mean values and
confidence intervals are shown by points and bars with a length of 2 σ . The reference values are
plotted in black. Figure from Ref. [83], reproduced with permission. Copyright2001, Royal
Society of Chemistry.