|
Quick Links Related Websites |
August 26, 2002 Modeling Sulfur K-XANES Spectra of Humic Acids With the Principal Component Analysis ApproachS. Beauchemin, D. Hesterberg, and M. Beauchemin Characterization of the chemical species of trace elements in soils, sediments, and other geochemical systems is essential for predicting their solubility and potential mobility in the environment. To determine metal speciation, scientists usually model x-ray absorption near edge structure (XANES) spectra by using standards of well-defined chemical species. But when scientists deal with complex matrices such as soil– in which many different chemical forms of the element may co-exist – other modeling techniques called principal component analysis (PCA) and target analysis, might be more convenient to use. The PCA was successfully applied to sulfur K-XANES spectra of six humic acid samples.
We tested a different modeling technique, called principal component analysis (PCA), to analyze soil mixtures. This technique considers the statistical variance within an experimental data set composed of a group of unknown samples. The data set is first redefined in terms of a reduced number of independent sources of variability, called components. These components have no chemical or physical meaning. A subsequent analysis, the target transformation, is used to test if suspected known species are part of the structural variation of the sample set. Based on the results from target analysis, abstract matrices obtained in the first step are transformed into chemically meaningful matrices. A major advantage of the PCA approach is that no a priori assumption is needed regarding the shape of the band. Also, suspected species can be evaluated individually without a priori knowledge of other species present in the sample.
PCA identified three main components. As illustrated in Figure 2, target analysis showed how closely the predicted targets fitted the suspected known species. Chitin sulfate was a better sulfate species than sodium sulfate to explain our experimental data. Sodium sulfate, cysteic acid and benzyl disulfide were marginal candidates, which means that the nature of the real species in the spectra of humic acids may be slightly different. Other standards were rejected. Therefore, we retained chitin sulfate, cysteic acid, and benzyl disulfide to derive the relative proportion of each selected standard species in each sample. The PCA approach provides a statistical basis for (i) gaining insight into the identification of chemical species in a multi-component sample, and (ii) defining the number of individual species to include in the spectral fitting. Therefore, it is a valuable complementary tool for band resolution or LCF as it helps to avoid an excessive number of bands or standards in the fit.
FUNDING PUBLICATION FOR MORE INFORMATION |
The National Synchrotron Light Source at Brookhaven National Laboratory in New York, is a national user research
facility funded by the U.S. Department of Energy's Office of Basic Energy Science. The NSLS operates two electron
storage rings: an X-Ray ring and a Vacuum UltraViolet (VUV) ring which provide intense light spanning the electromagnetic
spectrum from the infrared through x-rays. Each year over 2300 scientists from universities, industries and government
labs perform research at the NSLS.
Privacy and Security Notice | Contact Webmaster
X-ray absorption near edge structure (XANES) spectra from unknown
soil samples are usually modeled using techniques called spectral
deconvolution and/or least-squares linear combination fitting (LCF).
Both approaches rely on a priori information. Spectral deconvolution
is based on an assumption about the shape of the band, while
least-squares LCF involves fitting pure standard species to resolve
an unknown mixture. 
PCA coupled with target transformation was used to model sulfur
K-XANES spectra of six soil humic acid samples (Figure 1) using the
spectra of seven chemical compounds as sulfur standards: sodium
sulfate, chitin sulfate, cysteic acid, benzyl sulfoxide, benzyl
disulfide, elemental sulfur, and methionine. The objective was to
compare the results from PCA approach to those obtained from
least-squares LCF. 