Assessment of the physical properties, and the hydrogen, carbon, and oxygen content in plastics using energy-dispersive X-ray fluorescence spectrometry

https://doi.org/10.1016/j.sab.2020.105771Get rights and content

Highlights

  • Approach for quantification of hydrogen, carbon and oxygen in plastics with XRF.

  • Use of XRF to get information on physical properties of plastic samples.

  • Chemometric processing of XRF spectra for analysis of plastics.

Abstract

Energy-dispersive X-ray fluorescence spectrometry (EDX) is a widely used modern elemental analysis method. Most of the commercially produced EDX spectrometers cannot determine elements with atomic numbers below 11 (sodium). The EDX spectra contain scattering components and their intensity depends on the elemental composition and physical properties of the samples. The use of scattering as an analytical signal provides the opportunity to determine the integral characteristics of various samples. Since the light elements (with atomic numbers <11) also contribute to the scattering, it is possible to indirectly quantify these elements. In this study we demonstrate the use of X-ray scattering signals and chemometric tools to assess the physical properties and quantify the carbon, oxygen, and hydrogen content of various plastics.

Introduction

X-ray fluorescence spectrometry is a common tool that is used for elemental analysis. It is widely applied as a routine method in material and environmental science, biology, chemistry, industrial process monitoring, and numerous other fields. Energy-dispersive X-ray fluorescence spectrometry (EDX) is the most popular type of this method, allowing for fast and non-destructive multi-element analysis. The limitation of the majority of EDX devices is that the light elements (with atomic numbers <11) cannot be quantified. This hinders the applications of EDX for the analysis of organic samples. The typical EDX spectra contain two types of signal: characteristic X-ray fluorescence lines (associated with the content of particular elements) and scattered radiation from the X-ray source. The intensity of the scattering depends on numerous issues: elemental composition of the sample, density of the sample, geometry of the experiment, and scatter factors [1]. Thus, the scattered radiation can be used as a source of information on the integral properties of the samples. Since the scattering regions of the XRF spectra have no specific selective signals from particular elements (or parameters, e.g., density), the extraction of the useful chemical information from such spectra can be performed using chemometric tools. This idea was already explored in the literature. Thus, in [2] the scattering region of XRF spectra was used together with principal component analysis (PCA) to classify different vegetable oils. It was shown that the oils can be separated into clusters according to the type (soybean, sunflower, corn, and canola) and also according to the variety (extra virgin olive oil versus ordinary one). A similar approach was employed in [3], where the authors differentiated samples of dog hair according to gender and melanin. Another part of this study was devoted to the differentiation of varieties of coconut, PCA and hierarchical cluster analysis (HCA) were applied to explore the data. An interesting study that used a synchrotron radiation source to obtain XRF spectra was reported in [4]. The authors managed to distinguish different aliphatic alcohols (methanol, ethanol, 1-propanol, and 2-propanol) and they showed that the separation between the samples was much worse than when an ordinary Rh anode X-ray tube was applied as a radiation source. This study also demonstrated the potential of the scattered radiation signals for distinguishing between different carbonaceous materials (coke, graphite, activated carbon, and carbon nanotubes).

Scattered XRF spectra can not only be used to distinguish between samples with different compositions, it can also be used to produce quantitative prediction models to assess different numerical parameters in the sample. Goraieb et al. [5] employed partial least squares regression (PLS) constructed using XRF scattering data to predict the degree of sweetness in aqueous solutions of various sugars. Whilst the reported root mean square error in cross-validation was quite low, a possible limitation of these results is the small sample set size (15 samples) and the rather large number of latent variables (LV) in the models (16 LV) and this may indicate an overfitting of the PLS model. Another example of XRF quantitative analysis based on scattered spectral region can be found in [6]. The authors aim was to determine micro- and macronutrients in soil samples. The most interesting result of this study was the possibility of nitrate quantification, since nitrogen and oxygen do not have their own characteristic lines in the spectra. Regression modelling was carried out using PLS and artificial neural networks (ANN), the latter approach produced more precise results regarding the quantification of nitrate. Scattered XRF signals have also been used for the quantification of light elements, such as carbon and hydrogen [7], in paints and varnishes. These elements also do not have fluorescence lines in the EDXRF spectra. The results obtained in the PLS modelling of various scattering regions demonstrated the high potential of this methodology. Another study which addressed the topic of the quantification of light elements (hydrogen, carbon, and oxygen) looked at pure organic compounds, such as various alcohols, sugars, and acids [8]. The authors of this work did not employ multivariate spectral processing, but they used the intensities of scattered Cu and Pd lines to assess the content of light elements.

Some other interesting examples using scattered radiation signals on XRF can be found in the recent review by Leani et al. [9]. Further developments in this field are very interesting in order to broaden the number of XRF-detectable elements and to improve the performance capabilities of XRF spectrometry.

The purpose of this study was to explore the potential of chemometric processing of the scattered region of XRF spectra for the assessment of various light elements and quality parameters in different plastics. Presently, these materials are widely employed for construction and packaging, and a technique for the fast, non-destructive analysis of plastics urgently needs developing, especially in the context of waste recycling. Although there are no characteristic fluorescence lines from hydrogen, carbon, and oxygen in EDX spectra, the content of these elements can still be assessed in an indirect manner. The physical fundamentals of such an evaluation are based on the fact that the scattered radiation intensity depends on the elemental matrix composition, including the light elements. The initial X-ray tube radiation spectrum contains several lines, depending on the X-ray tube anode. Each of these lines can scatter on the sample, and thus one will have a variety of non-selective signals that are influenced by the content of all the elements in the sample. Multivariate data processing in this case can relate these non-selective signals with the content of light elements (or some other sample parameters that are related to the intensity of scattered radiation).

The first steps in this research direction were already done in the study by Nečemer et al. [10], which addressed the problem of plastic museum artifacts classification using linear discriminant analysis and particular numerical parameters derived from the scattering spectra. Whilst the classification was successful, no attempts regarding the numerical prediction of particular parameters were made. The present study aims to fill this gap.

Section snippets

Samples

Forty samples of various plastics were obtained from Röchling Engineering Plastics Ltd. (St. Petersburg, Russia). There were 13 different types of amorphous and semi-crystalline thermoplastics; polyamide (PA, 17 samples); polyoxymethylene (POM, 10 samples); polycarbonate (PC, 2 samples); polyphenylensulfone (PPSU, 1 sample); polysulfone (PSU, 1 sample); polyethylene terephthalate (PET, 1 sample); polyethylene (PE, 1 sample); polypropylene (PP, 1 sample); polyether sulfone (PES, 1 sample);

Results and discussions

The acquired spectra of the plastics are given in the Fig. 1a,c. Fig. 1a shows the spectra obtained in the light channel along with a zoomed-in image of the scattering region (2.6–2.9 keV) where the signal from Rh L-series can be observed. Aside from the scattering the spectra of different plastics may contain the characteristic X-ray fluorescence signals from particular elements, such as chlorine and sulphur (from polymers), and potassium, calcium, titanium, iron, and cobalt etc. (from

Conclusion

The EDX spectra that are traditionally used for the determination of elements with an atomic number above 11 can be used for the successful analysis of organic materials composed of lighter elements, if the scattered parts of spectra are taken into account. Chemometric processing of these scattered regions of the spectra allows the discrimination of various types of commercial plastics according to their chemical composition. Common multivariate regression techniques, such as PLS, allow for the

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

V.S. and V.P. acknowledge financial support by Ministry of Education and Science of the Russian Federation, Russia, State Project № 075-00780-19-00 (Subject № 0074-2019-0007).

Cited by (15)

  • Experimental performance of a LED thermal management system with suspended microencapsulated phase change material

    2022, Applied Thermal Engineering
    Citation Excerpt :

    Elements Na, S, and P came from the background because of their low atomic fraction (less than 0.6% for each element). H, an element of the eicosane core of the MPCM, was not detected because the atomic number of H is 1, and the commercially available energy-dispersive X-ray fluorescence spectrometry cannot detect elements with atomic number below 11 [57]. As shown in Fig. 3, the morphology of MPCM after 1, 10, and 100 cycles remains similar, which is spherical with a size ranging from 5 to 30 μm.

  • Application of chemometrics using direct spectroscopic methods as a QC tool in pharmaceutical industry and their validation

    2022, Profiles of Drug Substances, Excipients and Related Methodology
    Citation Excerpt :

    A noticeable review of the pharmaceutical application THz has been published by Shen and Jin [100]. Chemometrics using XRF spectroscopy and MIRS can be applied for the quality determination of the packaging materials such as vials, plastics or cling films as described by previous works [94,95]. The EDX spectra that are traditionally used for the determination of elements with an atomic number above 11 can be also used for the analysis of organic materials composed of lighter elements, if the scattered parts of spectra are taken into account, so the contents of hydrogen, carbon, oxygen and other quality parameters in commercial plastics can be determined and evaluated using chemometrics method [95].

  • Does chemometrics work for matrix effects correction in X-ray fluorescence analysis?

    2021, Spectrochimica Acta - Part B Atomic Spectroscopy
    Citation Excerpt :

    These approaches do not rely on some physical model, but they seek for numerical relation between the input data (XRF spectra) and output data (target parameter value) through special mathematical procedures: accounting for matrix effects in multielemental calibrations and numerical evaluation of sample parameters that are not accessible in traditional XRF. The examples of the last type are quantification In quantitative analysis these approaches can be employed for two different types of tasks of sugar cane quality parameters (content of fiber and sucrose) using XRF spectra of stem, leaves and syrups [9] and ED-XRF based quantification of hydrogen, carbon and oxygen and integral parameters in commercial plastics [10]. A recent study by dos Santos et al. [11] explored a quantification of soil organic carbon using ED-XRF and partial least squares regression.

View all citing articles on Scopus
View full text