https://doi.org/10.3390/min13020166
“Large-scale, high-density geochemical explorations entail enormous workloads and high costs for sample analysis, but, for early mineral exploration, absolute concentrations are not essential. Geochemists require ranges, dynamics of variation, and correlations for early explorations rather than absolute accuracy. Thus, higher work efficiency and lower costs for sample analysis are desirable for geochemical exploration. This study comprehensively analyzed the reliability and applicability of portable X-ray fluorescence (pXRF) spectrometry in geochemical exploration. The results show that pXRF can be applied effectively to rock and rock powder samples, and sample preparation and a longer detection time have been shown to increase the precision of the pXRF results. When pXRF is used on rock samples, if less than 30% of the samples are assessed as containing an element, the element is usually undetectable using pXRF when these rock samples are prepared as rock powders, indicating that the data about the detected element are unreliable; thus, it is suggested that some representative samples should be selected for testing before starting to use a pXRF in a geochemical exploration project. In addition, although the extended detection time increased the reliability of the analysis results, an increase in detection time of more than 80 s did not significantly affect the accuracy of the results. For this reason, the recommended detection time for the pXRF analysis of rock powder samples is 80 s for this study. pXRF has the advantages of being low-cost, highly efficient, and stable, and its results are reliable enough to exhibit the spatial distribution of indicator elements (arsenic, nickel, lead, sulfur, titanium, and zinc) in polymetallic mineralization exploration. Therefore, pXRF is recommendable for practical use in geochemical exploration.”
”
2.2. pXRF Analysis
An energy-dispersive pXRF (Thermo Fisher Scientific Niton XL3t 950 GOLDD+; Waltham, MA, USA) was used to analyze the 316 rock and rock powder samples. To analyze the rock samples, one piece of rock was randomly selected from each sample, the detection time was set to 120 s, and the MINING Cu–Zn mode was used. To analyze the rock powder samples, all rock samples were ground into a powder with a particle diameter of less than 75 μm. Then, the rock powder samples were packed into snap-and-seal polyethylene sample bags (note that a pile of new empty polyethylene bags was tested to make sure they contained no contaminants for the rock powder samples), and each sample was placed on a mobile test stand. Four maximum detection times were applied (40, 80, 120, and 200 s) under the conditions of the MINING Cu–Zn mode. Each of the four built-in exciter filters in the pXRF spectrometer required a quarter of the total detection time for one sample.
”
”
3.1. Usability Analysis of pXRF
To investigate the usability of pXRF technology in geochemical exploration, a pXRF was applied to the rocks and corresponding rock powder samples were collected from the study area. The detection time was set to 120 s for the rock samples and 40, 80, 120, or 200 s for the rock powder samples.
A total of 35 elements were identified with pXRF when analyzing the 97 randomly selected rock samples ().
Figure 2. The percentage of rock samples in which an element could be detected by pXRF.
In total, 26 of the 35 identified elements were measured by operating the pXRF spectrometer in copper–zinc mode to analyze the rock powder samples derived from the above-mentioned 97 rock samples (). The results revealed that the detection time could notably affect the results, especially for most of the trace elements. All of the identified elements can be classified into three groups (G1: Mg, Ni, W, and Bi; G2: Al, Ba, Ca, Cl, Si, S, K, V, Ti, Mn, Fe, Rb, Sr, Zn, and Zr; and G3: As, Cr, Cu, Mo, Nb, P, and Pb) according to their different detection times. Very little information was obtained from the G1 group since these elements were only detected in a small percentage of samples. The G2 group was mainly composed of rock-forming elements that could be detected in a detection time as short as 40 s. The G3 group’s results indicated that the detection capability increased as the detection time increased. The G3 group’s results also indicated that G3 contains perhaps three of the most important geochemical pathfinders for hydrothermal ore deposits: Cu, As, and Mo.
Figure 3. The percentage of rock powder samples in which an element can be detected by pXRF based on different detection times.
3.2. Influence of Detection Time on Analytical Error
To elucidate the relationship between the detection time and analytical error for each element, four different detection times (40, 80, 120, and 200 s) were applied when the pXRF spectrometer was used to analyze the 97 rock powder samples.
The analytical error results (i.e., two times the standard deviation) were compared based on the detection time for each element detected by the pXRF spectrometer, and the comparative analysis revealed that the analytical error decreased as the detection time increased. The analytical error results for three major elements (aluminum, silicon, and iron) and three trace elements (copper, lead, and zinc) are presented as examples in . shows that the reduction in the analytical error was most significant when the detection time was increased from 40 s to 80 s, and there is an exponential relationship between the detection time and the analytical error.
Figure 4. Analytical error of rock powder samples by pXRF based on different detection times.
Table 2. The relationship between detection time and mean analytical error.
Increasing the detection time from 40 s to 80 s can result in a relatively significant decrease in error, and the error fluctuation also tended to decrease with the increasing detection time (). Therefore, it can be stated that the analytical error and its fluctuation is relatively low and acceptable.
3.3. The Reliability and Stability of pXRF
To verify the results of the pXRF-based analysis method, the rock and rock powder sample results were compared to those obtained via laboratory analysis. The results for eight common indicator elements in mineral exploration (i.e., arsenic, copper, molybdenum, nickel, lead, sulfur, titanium, and zinc) are shown in .
Figure 5. Similarity between the analysis results of pXRF and laboratory. The 40 s, 80 s, 120 s, and 200 s are the results of rock powder samples reported by pXRF.
The result in shows that the reliability of the concentrations via pXRF analysis was not consistent across different elements, with the results for molybdenum being the least reliable. This is because the molybdenum concentration determined via pXRF was imprecise regardless of whether a rock or rock powder sample was used. However, the pXRF-derived results for arsenic, copper, nickel, lead, sulfur, titanium, and zinc were relatively reliable, especially for the rock powder samples. The results for most of the elements (e.g., arsenic and lead) were observed to be more accurate when a longer detection time was applied. However, the extent of the increase in accuracy was modest for most of the elements.
To investigate the robustness of the pXRF results, 32 rock powder samples were randomly selected and individually subjected to three successive rounds of pXRF-based analysis. The correlation coefficient results for these repeated measurements are shown in . It can be seen that all the correlation coefficients, except Ni and Cr, were close to one.
Figure 6. Correlation between different repeated measurements of rock powder samples by pXRF.
“
