Assessment of the Effect of Soil Sample Preparation, Water Content and Excitation Time on Proximal X-ray Fluorescence Sensing

Shuo Li, Jiali Shen, Thomas F. A. Bishop, Raphael A. Viscarra RosselSensors 2022 · DOI

TL;DR

This study systematically investigates how soil sample preparation, moisture content, and X-ray fluorescence (XRF) excitation time affect spectral data quality. The findings highlight that while all factors influence XRF spectra, longer excitation times (60s) generally improve signal intensity and definition, suggesting an optimal setting for field applications.

What Problem Does This Paper Address?

Portable X-ray fluorescence (pXRF) is a powerful, rapid, and non-destructive tool for analyzing elemental composition in soils, finding increasing use in environmental monitoring and research. However, the accuracy and precision of pXRF measurements can be significantly impacted by several factors, particularly in field settings where samples are often analyzed without extensive preparation. These factors include the physical state of the sample (e.g., particle size, surface roughness), its moisture content, and the duration of the X-ray excitation. While previous studies have explored some of these variables individually, there's a need for a comprehensive understanding of their combined effects and interactions to optimize pXRF protocols for reliable soil analysis. This paper aims to fill that gap by systematically evaluating the influence of sample preparation, water content, and excitation time on XRF spectral data.

Methods

The researchers employed a three-way factorial experimental design to investigate the effects of three factors:

  • Sample Preparation: Natural aggregates, soil ground to <2 mm, and soil ground to <1 mm.
  • Gravimetric Water Content: Air-dry, 10%, and 20% (by weight).
  • Excitation Time: 15, 30, and 60 seconds.

Four different soil samples were used as blocks. A portable XRF analyzer system was used to collect spectra. After spectral smoothing, elemental peaks for Si, K, Ca, Ti, Fe (Ka), and Cu were identified. Analysis of Variance (ANOVA) was performed to determine significant differences in peak intensity and area across the tested factors and their interactions. Post hoc tests (LSD) were used to pinpoint specific differences.

Datasets

The study utilized four distinct soil samples collected from intact soil cores. These samples were selected to represent a range of clay contents. The physicochemical properties of these soil samples (pH, clay, silt, sand, TOC, EC) are summarized in Table 1. No new benchmark datasets were introduced; the focus was on experimental data generated from these soil samples under controlled conditions.

Key Insights and Ideas

  • Excitation Time is King (for signal): Longer excitation times (up to 60s) consistently lead to more defined spectral peaks and increased signal intensity. This suggests that while speed is important in the field, sacrificing a bit of time for a stronger signal is often beneficial.
  • Moisture Matters, But Not Always Dramatically: While water content can attenuate X-ray signals, its effect was less pronounced than expected, especially for coarser samples. Significant differences were often only observed at higher moisture levels (20%) or when samples were finely ground.
  • Grinding is a Trade-off: Fine grinding (<1 mm) generally enhanced signal intensity, but it's a time-consuming process. For many elements, the benefit of grinding might not outweigh the effort, especially for less sensitive elements or when samples are already relatively fine.
  • Interactions are Subtle: While individual factors had significant effects, the interactions between them were less pronounced. The most notable interaction was between sample preparation and water content for certain elements, suggesting a complex interplay.
  • Field Optimization is Possible: The findings provide practical guidance for optimizing pXRF measurements in the field, suggesting that a 60-second scan on dry or slightly moist samples might offer a good balance of speed and data quality.

Key Results

  • All three factors (sample preparation, water content, and excitation time) significantly affected the XRF spectra.
  • Longer excitation times (60 s) generally resulted in higher peak intensity and area, with more defined spectral features.
  • For most elements, there was no significant difference in spectral response between air-dry and 10% water content.
  • Moisture content had a less pronounced effect on coarse samples unless they were ground to <1 mm.
  • Fine grinding to <1 mm generally yielded the highest intensity and area values for most elements.
  • ANOVA revealed significant interactive effects between sample preparation and water content on the peak intensity of Si, K, Ca, and Fe.

Limitations and Future Work

  • Focus on Spectral Quality, Not Quantification: The study focused on the influence of factors on spectral peak intensity and area, not on the accuracy of elemental concentration estimates. Further work is needed to assess how these factors impact quantitative results.
  • Wet Samples (>20%): The study did not extensively explore samples with very high moisture content (>20%), which are noted as potentially complicating analysis. Future research should address this.
  • Physical Mechanisms: The authors acknowledge that fully understanding the physical mechanisms behind some observed interaction behaviors can be challenging.
  • Element-Specific Behavior: The effects of these factors varied for different elements, highlighting the need for element-specific considerations in protocol development.

The authors suggest that for field applications, a 60-second scan on dry or slightly moist samples is a promising option. However, more work is needed for wet samples, which may require specific correction methods.

Summary updated

BibTeX
@article{li2022assessment,
  author = {Shuo Li and Jiali Shen and Thomas F. A. Bishop and Raphael A. Viscarra Rossel},
  title = {{Assessment of the Effect of Soil Sample Preparation, Water Content and Excitation Time on Proximal X-ray Fluorescence Sensing}},
  journal = {Sensors},
  year = {2022},
  doi = {10.3390/s22124572},
}
Figure