Auger Imaging and the Advantages of Spectra at Pixels


Auger images, when measured at a peak maximum are often difficult to interpret. Fluctuations in the signal as a function of position on the sample causes variation in the pixel intensities independent of the chemical information desired from the surface of the material. The difficulty arises because Auger images are typically collected in direct mode at the peak maximum and unlike spectral measurements, are often taken in isolation; background variations are sufficient to hamper the interpretation of such images. In an attempt to reduce these effects, images are often also acquired at an energy representative of the background and sometimes at a second background energy, with the view to including a contribution from the background to the final image and thereby improving the estimate of the peak height at each pixel. A common calculation for an Auger analysis is the, so called, topographical correction computed from the peak maximum N1 and a background image N2, namely (N1-N2) / N2. A variation on a theme is to compute an image from (N1-N2) / (N1+N2).


The principal reason for analysing Auger images using peak and background intensities is the time constraint of doing more, however, the benefits of applying spectroscopic techniques to image analysis may, when possible, out weight the time penalty of performing the extended acquisitions of images and allow spectra at each pixel to be determined. The example examined below (data provided by Pennsylvania State University, USA), is a simple experiment in which a gold grid is placed over a clear silver surface, and a set of Auger images are acquired at unit eV step size across a silver Auger peak. As luck would have it, a weak gold Auger line is also included in the energy range over which the images were recorded and therefore the data set offers a means of illustrating a simple, yet effective procedure for creating elemental Auger images. Rather than using the direct spectra, following the conventional wisdom for Auger spectra (Figure 1), the data is differentiated and peak-to-peak measurements made for each pixel in the image. Simple quantification regions are defined on the spectra, namely, one for the silver Auger line and one for the weak gold Auger line, then two images of the peak-to-peak intensities are constructed, these two images in turn are quantified using the standard procedure of quantifying spectra. The results are two images (Figure 2), where the silver image is free from intensity variations other than the elemental surface composition. In many ways, the improved quality of the gold and silver images are reassuring, since for years it has been assumed Auger spectra taken from a single point on a sample can be compared by determining atomic concentrations from the relative intensities of differentiated Auger peaks. The images in Figure 2 were determined from the peak-to-peak intensities without application of relative sensitivity factors, however provided the appropriate data is available, proper atomic concentration image are possible and these images allow the surface to be view using the same regime normally adopted for spectra.






Figure 1: A sample of spectra taken from individual pixels constructed from the original set of images.


Figure 2: Top two images are generated from peak-to-peak intensities from spectra and regions similar to those shown in Figure 1. Bottom two images, while representing an improvement over the peak maximum image, still include artifacts of the measurement process.


Analysis of Auger Images in CasaXPS


The essential sequence of steps required to process the image data is as follows:

  1. Convert the image set to spectra at each pixel.
  2. Differentiate all the spectra.
  3. Define quantification regions for each transition and propagate these quantification regions to each VAMAS block containing a row of spectra.
  4. Convert the region intensities to images.

Within these steps there are potentially sub-steps and these sub-steps are detailed below based on the original data for the given example.


The data in Figure 1 and Figure 2 were acquired on a PHI Auger system. Auger images were acquired, in this particular case, using two data files in which the saved images correspond to a sequence of unit energy steps over an interval spanning 370 eV to 331 eV. Since the data are in two separate files, the first job is to assign the correct energy to the experimental variable for each image and then move the images into a new Experiment Frame.

Ordering the Images using the Experimental Variable


Initially, when converted through CasaXPS, the PHI .map files each contain half the images for the entire data set. The images are assigned the same experimental variable and the species/transition VAMAS block fields are all different, therefore the VAMAS blocks appear in the right-hand-side of the Experiment Frame as a single row (Figure 3). To convert a set of images to spectra at each pixel, it is necessary for each image to appear in the same column and the experimental variable for each image to be assign the value for the energy at which the image was acquired. Given data in the format seen in Figure 3, the first step is to assign the species/transition fields for each VAMAS block in the file:

  1. Select the entire set of VAMAS block in the Experiment Frame. Either click on the single experimental variable value to select the row of VAMAS blocks or click once on the button heading the column of experimental variable values. Both these actions will result in the full set of VAMAS blocks being selected. The former action relies on all the data appearing in a single row, whereas the latter selects the entire set of VAMAS blocks in the Experiment Frame.
  2. Press the button on the second toolbar and enter new Element/Transition strings on the resulting dialog window. When the OK button is pressed, all the selected VAMAS blocks are assigned the pair of strings entered on the dialog window and as a consequence, the right-hand-side of the Experiment Frame is reordered so that all the VAMAS blocks appear in a single column.

Once the VAMAS blocks appear as a single column in the Experiment Frame browser, the values for the experimental variable are adjusted to the energy at which each image was acquired. The two toolbar buttons provide a means of assigning the experimental variable value on bulk; however the left-most of this pair of buttons offers a means of specifying a range of values for the experimental variable, where each row of VAMAS blocks will be assigned intermediate values. Thus for the data in Figure 3, following rearrangement into a single column, are assigned the appropriate experimental variable values by specifying the range 370 351 eV. Similarly the second file can be adjusted into a column of VAMAS blocks by assigning the same element/transition strings to each VAMAS block as was used in the first file and then the toolbar button used to specify the range 350 331 eV. Having made these assignments for the experimental values to both files, the two sets of VAMAS blocks can be merged into a new Experiment Frame. After selecting all the VAMAS Block in one Experiment Frame, clicking on the second to make the Experiment Frame the active window, then press the toolbar button. The action of the toolbar button copies the selected VAMAS blocks from all the open Experiment Frames to the active Experiment Frame (Figure 4).



Figure 3


Although the VAMAS blocks now appearing in the same Experiment Frame (Figure 4) all are in one column and the map energy is assigned to the experimental variable, there is still one possible problem. As seen in Figure 4, the images are ordered from high kinetic energy to low and so the spectra, when generated, will be assumed to be XPS data rather than AES, as is the case. It is also possible that the merged files appear back to front, from the energy scale perspective, and so the true order of the images, with respect to energy, may yet to be realized. The functionality in the second button for assigning the experimental value is required to force the appropriate reordering. The dialog window invoked by the right of the two toolbar buttons allows the assignment of the experimental variable for a selected set of VAMAS blocks. Further, once the assignment for the selection is made, the VAMAS blocks are reordered with respect to the new set of experimental variables. To effect a reordering of the image, it is simply the case of selecting a single VAMAS block and pressing the OK button on the dialog window. Although no new value was actually assigned, the reordering will still take place and the images then appear in the Experiment Frame ordered with respect to the map energies.


Figure 4


Converting Images to Differentiated Spectra


Given the set of images, now ordered by the energy at which the data are acquired, the analysis proceeds by converting the data set into one viewed as a set of spectra at each pixel in an image. The conversion from images to spectra is achieved using the Image Processing property page on the Image Processing dialog window. To invoke the Image Processing dialog window, click on the displayed image to ensure the Options menu items are active, and then select the Image Processing menu option. Overlay the images in the Active Tile and press the Convert Images to Spectra button (Figure 5) on the Image Processing property page. It is important that all the images are acquired using the same acquisition time and that the experimental variable represents a sequence of evenly spaced energies. If the step in energy between the images deviates from a constant difference, then an error message will appear and no conversion to spectra takes place.


Figure 5


The converted spectra appear in a new Experiment Frame (Figure 6), where the experimental variable is labeled pixel and the values for the experimental variables represent row indices of the pixels in the original images. Each VAMAS block contains an entire row of spectra; the spectra are stored as corresponding variables in each VAMAS block. As a result, a new processing option on the Differentiation property page of the Spectrum Processing dialog window (Figure 6) specifies that the operation should be applied to all the corresponding variables in a VAMAS block, rather than simply to the corresponding variable currently displayed in the Active Tile. The data in all the VAMAS blocks must be differentiated, therefore once the data in the Active Tile is differentiated, the propagate mechanism should be used to process all the VAMAS blocks similarly. To propagate the differentiation: select the entire set of VAMAS blocks in the right-hand-side of the Experiment Frame and right-click the mouse button over the Active Tile containing the differentiated spectrum. On the Browser Operations dialog window, tick the Processing tick-box within the Propagate section and press the OK button. All the spectra within the targeted VAMAS blocks will be differentiated.


Since the spectra are stored as multiple corresponding variables and only one corresponding variable from a given VAMAS block can be viewed at a time, it becomes necessary to step through the corresponding variables to inspect the results of the differentiation operation. To step through the corresponding variables in a VAMAS block, the Control + Page-Up and Control + Page-Down keyboard button are used. The corresponding variable index in each VAMAS block within an Experiment Frame is adjusted by the Control + Page-Up and Control + Page-Down keys. Similarly, Control + Home and Control + End move the corresponding variable index to the beginning and end indices for each VAMAS block in the Experiment Frame. To adjust the index for a corresponding variable on a VAMAS block by VAMAS block basis, using the Page-Up, Page-Down, Home and End keyboard keys whilst holding down the Shift key causes only the VAMAS blocks in the Active Tile to be adjusted.



Figure 6


Quantifying the Spectra at Pixels to Produce Images


Examples of the differentiated spectra are shown in Figure 1. The quantification of Auger spectra is typically performed using the peak-to-peak metric to measure line intensity. In this example, two Auger lines are evident in the energy range 331 370 eV and so two quantification regions can be defined on each spectrum as indicated by the grey bands over the data in Figure 1. These quantification regions are defined on the Quantification Parameters dialog window via the Regions property page. The regions for use with the spectra-to-images options are identical to those used for general spectral quantification, but with one slight difference. The default action of the Convert Regions to Images button on the Image Processing property page is to create images using the integrated intensity based on peak area. For differentiated AES data, the peak-to-peak intensity is required and to create images appropriate for the data in Figure 1, the Tag field in the region specification table must be assigned a keyword string, namely, peak to peak. The Tag field is also used to switch between other regions parameters such as fwhm, position and centroid; by using these various region outputs, the surface mapped by the images can be viewed with respect to peak broadening or peak shifts. Again the Tag field is entered with the appropriate keyword string e.g. fwhm, position or centroid. Figure 7 shows the Regions property page where two quantification regions have been defined on a spectrum in which the Tag field are set to the keyword for generating images using the peak to peak intensities.


Figure 7


Once quantification regions have been propagated to each VAMAS block in the spectrum file, the corresponding images are generated by overlaying all the spectra from the VAMAS blocks in the Active Tile and then pressing the Convert Regions to Images button (Figure 5).


Ideally, images for each element identifiable on the sample surface should be measures and a similar analysis to the Ag and Au data performed for each element. In this example only two elemental images were recorded, but nevertheless, proceeding to the quantification step based on the two available images has merit in that the resulting images (Figure 2 top) are normalized with respect to each other across the field of view. Figure 5 shows a section of the Image Processing property page, on which the button labeled Quantify Images can be used to perform the operation. The Quantify Images button assumes the images used in the calculation are generated from either quantification regions or synthetic components, both of which are in units of CPSeV (for area based intensities) or CPS (for peak-to-peak measurements), therefore the new images are generated simply using the formula Ij / (I0+I1+ +In). Intensity adjustments for relative sensitivity and/or transmission are accommodated by entering an appropriate value in the RSF field in the quantification regions or components. To perform the quantification step, overlay all the images in the Active Tile, then press the Quantify Images button. A new Experiment Frame is created containing the quantified images.


Note, a second button labeled Quantify Peak BG offers a second means of quantifying a set of images. The procedure in this case does not involve images preprocessed using spectral regions or components and therefore adjustments for time will be included in the calculation. This is neither appropriate nor desirable for data processed as described in this section, but is available for use with raw images, where a less sophisticated approach is adopted.