PRS - back to the homepage   Pattern Recognition Systems - mail us {short description of image}
{short description of image}
INFO {short description of image}
    Product Facts
   |MS Resolver
News Download

10.04.04 mail uswebmaster

PRS as
MIX-M°hlenpris PB 24
5006 Bergen
+47 5532 5221
+47 5558 9496


Resolution of complex samples measured by multidetection chromatography

An analyst often encounters complex samples containing many chemical components, some of which are unknown. A traditional work strategy for such samples is to employ a chromatographic separation technique, thus enabling identification and quantification. To achieve proper instrumental settings so as to ensure adequate separation is not always an easy task, and smaller interferents and pollutants might get overlooked. This introduces errors, both in the qualitative and the quantitative stage. The introduction of multidetection instruments, e.g. by coupling spectroscopic detection with chromatographic separation represented a major step forward in this type of analysis. The data analytical tools commercially available, however, have until recently not kept up with the pace of the instrument manufacturers. Xtricator is a unique software package suited for quantitative and qualitative analysis of this type of data. This application note demonstrates the use of Xtricator on an air pollution sample form Hong Kong analysed on an HPLC-DAD 1050 instrument. The sample is a very complex one, as demonstrated by the mean chromatogram shown below.

At least 20 overlapping peaks can be identified from the chromatogram. To decide whether a peak is pure or not, is often very difficult - even when looking at the corresponding spectrum. Xtricator contains procedures designed to help you with this. For the remainder of this application note, we will focus on resolution of the region between 20 and 25 minutes. The mean chromatogram indicates that at least three components are present in this region. Before the resolution can take place, we need to remove a drifting baseline and a background spectrum. Xtricator contains procedures for this purpose, but they will not be demonstrated in this application note. One of the major tools of Xtricator is an evolving eigenvalue plot. It is created by moving a window through the rows of the data matrix, so that it covers only a small part (typically 2 - 10 spectra) of the matrix. This sub matrix is decomposed, and the resulting eigenvalues calculated. The window is the moved through the data matrix, and each sub matrix it covers, is decomposed. The resulting eigenvalues are plotted in an ETA plot. For this data set, an ETA plot with window size 8 is shown below.

The number of eigenvalues rising above the noise level indicates the number of components present in a region. The plot clearly demonstrates the evolving nature of the data, as the components appear and disappear at the detector. From the plot, five selective regions (only one component) and several two-component regions can be identified. You can order Xtricator to investigate the selective regions, as is done in the plot below.

By decomposing the selected region and plotting the resulting loading vectors, it becomes evident that this is a selective region. Only the first loading vector contains structural information. The size of the eigenvalues is used to confirm this hypothesis. After marking and investigating the five apparently selective regions, Xtricator resolves the data into the underlying chromatograms and spectra of the five components in the region. The results are shown below

In this case, every component had a selective region. However, Xtricator can handle more complex situations as well. Four of the five resolved components were later identified to be cyclopenta[c,d]pyrene, chrysene, benz[a]anthracene, and pyrene. This application note only demonstrates the use of some of Xtricator's powerful tools on a small region of the data. This complex sample was showed to contain 47 chemical components, which were all resolved by these methods.

A detailed report on this work cam be found in Shen, Liang, Yu, Li and Sun, Science in China (41), 21 - 29.

----------------------------Top of Page
© 1997-00 PRS AS. All rights reserved.