Manipulating Spectra for Comparative Quantification

A common analysis scenario is comparing spectra from a batch of similar samples, where the principal difference between the recorded spectra is a set of minor peaks typically located near some dominant structure in the spectrum. If the set of samples are the result of a modification to a standard material, then manipulating the spectra using a systematic procedure may produce comparative results which otherwise would be difficult to achieve.

Consider the set of spectra shown in Figure 1. These data [1] correspond to a base material, referred to as the blank sample, and three spectra acquired following surface modification, designated samples 20107a1, 20107b1 and 20107c1. The data are acquired using different acquisition settings and is also subject to varying charging conditions. The feature of interest is the small Cr 2p doublet; variations in the intensity of the Cr 2p characterises the experiment. Two problems present themselves, namely, the Cr 2p lines are distorted by the relatively large O 1s plasmon loss structure and the variation in the acquisition procedure means that the data cannot be directly compared to one another. A solution is to shift and normalise the spectra with respect to the Si 2p lines and then subtract the normalised blank sample data from the modified sample spectra.

Figure 1

The VAMAS file shown in the Experiment Frame in Figure 2 includes the spectra displayed in Figure 1. Spectra acquired from each of four samples are collected into a single VAMAS file where each sample is identified by a different index number (entered as the experimental variable). The resulting Experiment Frame displays the logical file as a set of spectral regions organised as rows of VAMAS blocks. CasaXPS provides a means of energy calibrating and intensity normalising the rows of such a file with respect to a particular peak within a spectral region on a row-by-row basis.

The steps required to both energy calibrate and intensity scale a file of spectra are as follows:

1) Choose a spectral region on which to calibrate the data. In the example shown in Figure 2 the Si 2p line has been chosen to provide the reference for each row of data.

2) Create a region and/or component on the reference spectra. Note that the first integration region or the first synthetic component will determine the position and/or the intensity used to calibrate the data. The term “first” means the region or component that appears first in the scrolled lists displayed on the Quantification Dialog Window.

3) In the right-hand-side browser, select the set of VAMAS blocks for which the calibration region or component is defined. In Figure 2, all the Si 2p VAMAS blocks are selected since these contain an integration region from which the position and scale factors will be determined.

4) On the Spectrum Processing dialog window, select the Calibration property page. Enter the energy for the selected spectral region consistent with the chosen region or component.

5) If the spectral intensities are to be normalised, the Use Reference Intensity checkbox should be ticked and an appropriate scale factor entered into the Reference text-field. The value entered into the text-field is simply to allow the scaled data bins to have values greater than one (if desired).

6) Press the appropriate button Apply by Row (1^st Region) or Apply by Row (1^st Comp) to perform the calibration for each row for which a corresponding VAMAS block is selected. If the Use Reference Intensity checkbox is ticked, a new Experiment Frame is created in the image of the original file, however the data is adjusted with respect to acquisition time instead of maintaining the dwell-time and number of scans as VAMAS fields. The data is adjusted under these circumstances to ensure that the spectra are normalised with respect to counts per second and removes the need to perform further pre-processing steps as part of the Calculator operations. This does mean, however, that once performed, the data file should be saved to a different file name otherwise information previously held within the original VAMAS file will be lost.

Figure 2

Once the data has been normalised and calibrated, the Calculator Property Page can be used to subtract the blank sample spectra from the modified spectra.

Figure 3

Step to subtract the blank sample Cr 2p spectrum from the modified spectra are as follows:

1) Display the blank sample Cr 2p spectrum in the Active Tile.

2) On the Calculator Property Page, press the Set Operand pushbutton (Figure 3). The Active Spectrum in the Active Tile is loaded into the Calculator when the Set Operand pushbutton is pressed.

3) Press the Offset Adjust pushbutton. The spectrum loaded into the Calculator is shifted in intensity so that the maximum value for the data is zero. This ensures that the operand used in the calculation is below the spectra to which the operation is performed, thus resulting in positive values in the resultant (following the subtraction of the operand loaded into the Calculator).

4) Select the spectra in the right-hand-side browser for which the subtraction is required.

5) Select the Subtract radio button.

6) Press the Apply to Browser Selection pushbutton.

7) View the selected spectra in the left-hand side.

Figure 4 shows the Cr 2p spectra after subtraction of the equivalent data recorded from the blank sample. The peak to the right of the Cr 2p doublet is a Ti 2s line, which is consistent with the Ti 2p features shown in Figure 1 but is difficult to see in the data before subtracting the blank sample (Figure 3). The key point that has made this analysis possible is the consistent nature of the Si 2p lines, which offer a reference for both energy and intensity calibration.

Figure 4

[1] Dr. P.C. Thüne, Dutch Polymer Institute, Eindhoven University of Technology, email: p.c.thuene@tue.nl (private communication) 2002