Quantification using Integration Regions Combined with Synthetic Peaks

Routine measurements using XPS often involve acquiring a wide scan spectrum to determine the general composition of a sample. The survey spectrum is then used to select narrow energy regions where detailed structures are present. These narrow scan spectra are typically needed when overlapping peaks are responsible for the data envelope such as is often seen in the case of C 1s spectra. Synthetic line-shapes must be used to extract the chemical-state information in the data, but intensities determined from these models are not always comparable with intensities determined from integration regions. The most notable case is when asymmetric line-shapes are used to model the data. Moreover, the transmission characteristics of an instrument require that each operating mode must be characterized by a transmission function and although it is possible to correct the data for such differences, problems may arise due to changes in the system from aging or tweaks to lens functions by the operator. Both require recalibration of the instrument, without which peak areas from a wide scan acquired using one pass energy cannot be used with intensities measured via other modes.

If a spectrometer is characterized by a set of relative sensitivity factors (RSF’s) applicable only to the survey mode of an instrument, but the resolution of the survey mode is insufficient to provide the desired chemical state information, then one way to provided a more detailed quantification is to reference the intensities from the to high-resolution spectra to this survey mode. Peak areas determined from the high-resolution data provide the relative proportions for the chemical-state intensities and therefore the corresponding concentration from the survey data can be subdivided using these proportions, thus allowing detailed quantifications without the need for transmission correction. Such a procedure effectively takes out the transmission function from the quantification step and lessens the need for the time consuming calibration of each operating mode. Indeed, for many instruments transmission correct is not available and so this procedure represents the only way to combine data from different operating modes, which would otherwise require the maintaining of multiple sets of RSF values.

Figure 1 shows a typical situation from an XPS measurement where a survey spectrum provides the overall elemental quantification but a narrow C 1s scan offers a wealth of information lost by the limited resolution used for the survey spectrum. In the case of relatively pure samples, such as the one in Figure 1, it might be possible to run a set of high-resolution spectra for each of the peaks identified from the survey data, but many samples include so many elements that full quantification via high-resolution spectra would be both time consuming and costly. Degradation of a sample from exposure to X-rays and financial considerations make the use of high-resolution spectra unattractive as a routine analysis regime.

The data in Figure 1 is quantified via a table that shows the elemental concentrations plus a further breakdown of the C 1s elemental composition into chemical state concentrations. The proportions determined from the peak model for the high-resolution envelope are used to show how the elemental concentration is subdivided into chemical-state intensities. This is achieved by assigning a “tag” to each of the quantification items used in the analysis. That is to say, in addition to the name field in each integration region or synthetic component there is a Tag entry (in the form of a string), which is used to link intensities from synthetic peaks to a specific integration region.

Figure 1: C 1s high-resolution data used to proportion the quantification for C 1s in the survey spectrum.

If the information is available, CasaXPS will enter into the Tag field the element and transition associated with the acquisition, and only those items with tags are included in the quantification table. To remove an item from the quantification step delete the string from the Tag field and press return. The system will enter the key word NoTag when return is pressed to indicate that the item is not to be used in the quantification. Note that the NoTag entry is necessary to exclude from the elemental concentration table the integration region defined for the high-resolution scan seen in Figure 1. The background to the peak fit for the C 1s spectrum requires a region to be defined, but the intensity from this region must not be included in the results from the survey spectrum. In this example the tag field in the region used to define the background to the peak fit must be set to NoTag, while each of the synthetic components are tagged with the same name as the corresponding region in the survey quantification, namely, “C 1s”.

Not all forms of quantification tables use tags. For example, the Component annotation option used to display the component table in the upper spectrum does not discriminate between the components based upon tags, nor do any of the annotation options involving quantification tables unless the “Use Tag Field” checkbox is ticked. The tag mechanism is, however, always used when the Combined button on the Report Spec property page from the Quantification Parameters dialog window is selected. No other quantification reporting options use tags.

Although the primary reason for using tagged reports is to reference results from peak fits to survey spectra, it is possible to use the same mechanism to remove some of the ambiguities from comparisons between intensities derived from peak fits that employ different line-shapes. Any line-shape involves a functional form that may extend beyond the acquisition regions to which it is applied, especially when asymmetric peaks are involved. The tag mechanism allows the peak fitted results to be referenced to the underlying integration region and so intensities for the chemical states are calculated based upon the data rather than the implementation of the synthetic components.

Figure 2: Tag mechanism applied to differing line-shapes. The Al 2p has been modeled using Doniach-Sunjic profiles, while the O 1s spectrum is fitted with Gaussian-Lorentzian line-shapes.

Figure 2 shows a set of quantification tables. These tables derive from components and integration regions taken from the two high-resolution scans below the survey spectrum. Note that the Al 2p doublet is fitted using a Doniach-Sunjic asymmetric line-shape, while the O 1s spectrum is modeled using symmetric Voigt approximations. The table headed “Components Only” shows the atomic concentrations where the intensities are calculated from the line-shapes. These concentrations do not agree with the elemental results calculated from the integration regions alone, where the Al 2p concentration is computed to be 85.38% while components only reports 88.24% for the same quantity. The differences are due to the cutoff criterion used to limit the infinite area under a Doniach-Sunjic profile. Using the tag mechanism the peak fit can be used to subdivide the concentrations from the integration regions using the individual components from the peak models. The concentrations for the synthetic components are now consistent with the results of the integration regions but are in the proportions determined by the models.