STEM Spectrum Imaging Tutorial Gatan, Inc. 5933 Coronado Lane, Pleasanton, CA 94588 Tel: (925) 463-0200 Fax: (925) 463-0204 April 2001
Contents 1 Introduction 1.1 What is Spectrum Imaging? 2 Hardware 3 Software 4 STEM Spectrum Imaging 4.1 Overview 4.2 Preparation 4.3 Acquisition 4.4 Visualization 4.5 Analysis 5 Quick Reference 1
1 Introduction This tutorial provides a step-by-step guide for the acquisition and processing of spectrum images using Gatan systems. The methods for acquiring such images are dependent on the relevant hardware specification. Here, we will focus on the case of EELS spectrum imaging in Scanning Transmission Electron Microscope (STEM) mode. The processing methods are, however, applicable to all spectrum images, irrespective of the mode of acquisition. We begin with a brief description of the physical principles that govern the technique. 1.1 What is Spectrum Imaging? The process of spectrum imaging refers to the acquisition of three-dimensional data that contains both spectral and spatial information about the specimen. The resulting data set is illustrated in figure 1-1, and is given the term spectrum image. Acquiring a spectrum image can be viewed as a progressive filling of the cube of data represented in the figure. x E y Figure 1-1. Schematic of 3d spectrum image data cube. This discussion of Spectrum Imaging is confined to Electron Energy Loss Spectroscopy (EELS) in the Transmission Electron Microscope (TEM) environment. The energyfiltered (or EFTEM) mode of Spectrum Imaging is covered in a separate tutorial. Here, we concentrate on the STEM mode, where the microscope is configured in such a way as to produce a nanoscale probe at a particular point (x,y) on the specimen. An energy-loss spectrum is acquired at each of a succession of probe positions as it is scanned across a survey region, and the data cube is thus filled line-by-line. The following sections 2
describe in detail how to acquire spectrum images in the STEM case, beginning with a survey of the hardware and software. x y E a) b) Figure 1-2. Acquisition modes for EELS a) EFTEM and b) STEM Spectrum Imaging.The arrows show the filling of the data set. 2 Hardware In order to generate STEM spectrum images, the microscope must be set up so as to allow for STEM EELS integration. In view of this the following hardware components are required: A STEM unit on the TEM or a STEM microscope. A dark field detector at the entrance to the spectrometer. Digital control of the beam. A spectrometer system. Software to integrate the whole system. This tutorial will focus on the specific case of the Gatan Digiscan system for scanning the electron probe to a acquire dark field images. For spectral information, one of the Gatan ENFINA or GIF systems is used. Details of both are given in separate manuals. 3
3 Software The general software requirements are FilterControl, to control the spectrometer, and DigitalMicrograph for viewing and processing the images. In addition, the Digiscan package contains software that mediates the scanning of the probe. The acquisition and processing of spectrum images is done through the Spectrum Imaging package, coupled with AutoPEELS ( when using the ENFINA system) or AutoFilter( when using the GIF ). Although not necessary, EELS Analysis is also useful for the quantification. 4 Acquiring STEM Spectrum Images 4.1 Overview Acquiring a set of data with both spectrum and image information requires integration of the STEM and spectrometer modes. The sequence of steps for achieving this are as follows: Electron beam focussed at sample Sample Diffraction pattern Dark field detector Electrons pass into spectrometer 4
Figure 4-1. Optics for conventional STEM work. Align the STEM to give a high quality dark field image at the spectrometer entrance. Acquire image and set as survey image. Align the spectrometer. Set up the acquisition conditions. Specify the region for taking the spectrum image. Mark areas for other corrections i.e drift. Set the spectrum image parameters. Acquire the spectrum image. Process data. Each of the above steps are described in further detail below, beginning with the procedure for microscope alignment. 4.2 Preparation Prior to acquiring a spectrum image, it is important to be able to acquire an image using the STEM unit on the TEM. Many older STEM units have poor control over the optics and require the TEM to be used in free lens mode to give optimal STEM images on either a bright field or dark field detector. To get a useful signal into the spectrometer and also give a bright, clear image on the dark field detector a number of conditions need to be set. To set up the microscope 1. Make certain that the image on the TEM screen is an in focus diffraction pattern. It should look like at CBED pattern as the beam is converged in STEM. 2. Set the camera length to be between 5 and 20 cm. Ensure that the bright field disc is smaller than the spectrometer entrance aperture and also the dark field detector s aperture. 3. Remove the objective aperture if not already done so. 4. If the STEM has Digiscan coils make sure that the diffraction pattern is stationary. 5. Ensure no light gets onto the detector. To take useful EELS data at the same time as a STEM image requires the same conditions that one uses for spectroscopy but with the beam rastering across the sample. The latter is controlled by the Digiscan and the details of acquiring an image in this case are given below. To collect a Digiscan image 5
1. Lift up the TEM screen. 2. Collect an image by pressing Start on the View tab of the DigiScan floating window. The view parameters are typically set to give a lower resolution but faster updating than the corresponding acquire mode. Typical parameters in both cases are given below. The long dwell time( in all cases, an acquisition time greater or equal to 16 µs is recommended ) ensures accurate positioning of the microscope beam in subsequent analysis. Figure 4-2. View and Acquire parameters for Digiscan. 3. Adjust magnification, focus and sample shift as required to produce a focussed image of the region of interest. Figure 4-3. A typical dark field Digiscan image. 4. In some cases it may be necessary to alter the Digiscan setup. Click Setup on Digiscan floating window. Change the flyback time to the value shown. 6
Figure 4-4. Digiscan set up. 5. Press Stop to terminate scanning of probe. The image on the STEM system can be adjusted with the brightness and contrast controls for the detector and then aligned to be sharp, focused and stigmatic in the normal way on the microscope. It is now necessary to specify the acquired image as a survey image for spectrum image acquisition. To assign a survey image 1. Bring the acquired Digiscan image to the front by clicking on it. 2. Go to the SI menu. Select the menu item ASSIGN SURVEY IMAGE. When using the STEM system, it is important to know the location of the beam prior to taking a spectrum. Rastering the beam or random positioning can lead to results that are misleading. The spectrum imaging software accommodates this and facilitates integration between the image and the spectrometer by creating an extra tool, referred to as the exploration tool. This positions the beam and commences the spectrum viewer. To engage it, click on the tool shown below, which forms part of the Standard Tools floating window. Figure 4-5. The exploration tool. At this stage, it is useful to carry out some steps to acquire a preliminary EEL spectrum. This will provide some insight as to the nature of the data set that one might expect from the final spectrum image. For example, it is useful to check that the current configuration 7
for the start energy or exposure time does not cause the CCD to saturate or that the sample is thin enough to produce a good EELS signal. It also provides a basis for defining the setup parameters that should be specified prior to acquisition. We begin by describing the steps toward producing a sharply focussed and aligned zero loss peak (ZLP). The specification of certain EELS acquisition parameters such as the dispersion and exposure time will differ depending on which of AutoFilter or AutoPEELS is being used. In view of this, the procedure in each case is described separately. To Align and Focus the ZLP with AutoPEELS 1. Adjust the AutoPEELS View and Acquire parameters if necessary to correspond approximately to those given below. The parameters are chosen so as to produce a non saturating zero loss peak centred close to 0 ev. Note that, as the absolute probe intensity is dependent on microscope details, such as the brightness, spot size, choice of condensor aperture, the actual values for the Integration time might vary. Figure 4-6. AutoPEELS View and Acquire parameters. 2. Select the exploration tool from the standard tools floating window. 3. Locate a hole in the sample from the survey image, if one is present. If not, locate a very thin region. 4. Use the exploration tool to position the probe at an arbitrary point on the hole. This is done by clicking at the chosen point in the survey image. The probe position is marked on the image and spectra are continuously gathered and displayed corresponding to this position. A sharp and intense zero loss peak should be visible. Note that the display may be expanded/contracted by clicking and dragging the pointer tool at a point close to the relevant axis, while pressing the CTRL key. 5. To align the zero loss peak, click on the Align ZLP button in the AutoPEELS Tuning floating window. Following alignment, the ZLP should be centred on, or very close to, 0 ev. 8
Figure 4-7. Alignment using AutoPEELS. 6. To optimise the energy resolution, one can focus the ZLP by clicking on the Focus button. Figure 4-8. Focussing using AutoPEELS. Now that the ZLP is properly positioned and focussed, one can acquire a spectrum from the sample. To Align and Focus the ZLP with AutoFilter 1. Switch to spectroscopy mode by clicking on the EELS button within the AutoFilter floating window. Figure 4-9. AutoPEELS View and Acquire parameters. 2. While pressing the ALT key, click on the EELS button once again to generate the AutoFilter Technique Options dialog. Specify the appropiate beam energy,dispersion and aperture size. Typical values for these parameters are given in figure 4-10 below. 9
Figure 4-10. AutoFilter Technique dialog. 3. Click on Turbo button in the AutoFilter floating window. Figure 4-11. AutoFilter View buttons. 4. Then, while pressing the ALT key, click on it again to generate the Spectroscopy Setup dialog. Typical values for parameters are given in figure 4-12 below. Figure 4-12. Spectroscopy setup dialog 10
5. Specify the start energy range of interest (in this case, the zero-loss). Do this by clicking on Zero Loss button within the AutoFilter floating window. Figure 4-13. AutoFilter Energy buttons. 6. Select the exploration tool from the standard tools floating window. 7. Locate a hole in the sample from the survey image, if one is present. If not, locate a very thin region. 8. Use the exploration tool to position the probe at an arbitrary point on the hole. This is done by clicking at the chosen point in the survey image. The probe position is marked on the image and spectra are continuously gathered and displayed corresponding to this position. A sharp and intense zero loss peak should be visible. Note that the display may be expanded/contracted by clicking and dragging the pointer tool at a point close to the relevant axis, while pressing the CTRL key. 9. To align the zero loss peak, click on the Align ZLP button in the AutoFilter floating window. Figure 4-14. AutoFilter Commands buttons. 11
To acquire a first spectrum 1. Identify a region of interest in the survey image for SI acquisition. 2. Use the exploration tool to position the probe on to this part of the specimen. Notice that the spectra are continuously refreshed as the probe is shifted from one part of the specimen to another. For the given view parameters, one should observe a decrease in intensity of the zero loss peak and also the appearance of plasmon peak(s), whose number and integrated intensity vary with thickness. If the signal diminishes substantially, one could increase the integration time. The specification of certain spectrum imaging acquisition parameters such as the start energy and dispersion will depend on compositional details specific to the sample. As an example of how values may be assigned to these quantities, consider the case of a specimen containing the elements C, N, O, Cu, Al and Si. Here, the potential edges of interest are the Si L and K edges, at 99 ev and 1839 ev respectively, the C K edge at 284 ev, the N K edge at 401 ev, the O K edge at 532 ev, the Cu L edge at 931 ev and Al K and L edges at 73 and 1560 ev. Given the difficulty in background fitting and the uncertainty in edge computation at the low energy edges, we might choose to ignore the Si and Al L edges, and instead attempt to quantify at the associated K edges. For the purposes of ensuring reliable quantification, one would require an extra energy window of about 200 ev ( say 100 ev above the highest energy edge onset and 100 ev below the lowest energy edge ). In this case, given the lowest lying edge is the C K edge at 284 ev, the Start energy may feasibly be set to 150 ev. The resulting energy window of about 1800 ev spread over the entire CCD of 1340 channels might lead to a dispersion choice of 1.5 ev per channel. The acquisition time is chosen to be sufficiently large to produce a quantifiable signal at the high energy K edges while still ensuring that the CCD remains unsaturated in the lower energy regime. In addition, one should note the balance between spectral quality and speed of acquisition. Whilst this is true for EELS technique in general, it is particularly so for the case of spectrum imaging. This is exemplified by the fact that even a small image of 50x50 pixels requires 2500 spectra to be acquired. Figure 4-15. AutoPEELS parameters for acquisition of spectrum image. 12
The CCD camera takes differing amounts of time to read a frame depending on the binning set. To change the binning, ALT-Click on the View or Acquire buttons. Note that the SI software looks at the acquire binning. Although the maximum brightness can be seen an x = 1, y=1-5 binning, the greatest speed is achieved at x=1, y= 100. Increasing the x binning also increases the speed but reduces the number of spectrum channels and is therefore not normally used. Another method of increasing the acquisition speed is to reduce the dispersion so as to increase the signal into each pixel. In the case of AutoFilter, the equivalent of Start energy may be specified by clicking the Custom button. Then, while pressing the ALT key, click the button again. Then adjust the field associated with the E and Custom parameters to the required values. Figure 4-16. Specification of start energy loss using AutoFilter. To Define the spectrum image region. 1. Bring the survey image to the front by clicking on it. 2. From the ROI tools floating window, select the rectangular ROI tool. Figure 4-17. The ROI tools floating window. 3. Locate the region of interest on the survey image. 4. Click and drag to specify this region on the survey image. 13
5. From the SI menu, select the menu item ASSIGN SPECTRUM IMAGE REGION. The defined ROI is now labelled as Spectrum Image To Define the spatial drift region. 1. Bring the survey image to the front by clicking on it. 2. From the ROI tools floating window, select the rectangular ROI icon. 3. Locate an area on the survey image that demonstrates some contrast. 4. While pressing the ALT key, click and drag on the mouse to specify a square region of interest on the survey image. The ALT key constrains each side of the ROI to be of length 2 N pixels, where N is an integer. 5. From the SI menu, select the menu item ASSIGN SPATIAL DRIFT REGION. The defined ROI is now labelled as Spatial Drift. Figure 4-18. Survey image with spectrum image marked with Spatial Drift and Spectrum Image regions To set up spectrum image parameters 1. On the SI menu, select the SPECTRUM IMAGE SETUP menu element. This is necessary for the cross-correlation routine that computes the relative image shift. 14
The following dialog is displayed. Figure 4-19. Sampling parameters for spectrum image acquisition. 2. Specify the sampling parameters by clicking on Sampling tab. The spatial resolution is improved by increasing the x and y sampling, but is compensated by the resulting increases in integration time and memory requirement. 3. Click on the EELS Spectrometer tab. Verify/adjust the parameters accordingly: Figure 4-20. Spectrometer options for SI setup. 4. Click on the Corrections tab to specify type(s) and method(s) of correction. This tutorial deals only with the spatial drift correction is as it is considered the most significant contribution to distortions in the final spectrum imaging data set. 15
Figure 4-21. Corrections in SI setup. In general, some experimentation with the spectrum image parameters may be necessary. Additionally, given the relatively large acquisition time, one should bear in mind features for optimising the speed of acquisition (such as the binning set ). Note also, that for the given set up parameters, the dialog displays the expected acquisition time and memory requirements. 4.3 Acquisition We are now ready to acquire a spectrum image. On the SI menu, select the ACQUIRE SPECTRUM IMAGE menu element. The spectrum image window is displayed and the pixel intensities are filled in a row by row fashion as spectra are gathered at each probe position within the spectrum image region. Figure 4-22. A typical acquired spectrum image. 16
4.4 Visualization To view slices of the spectrum image 1. Select the Slice floating window. Figure 4-23. The slice floating window. 2. Drag topmost positioning tool from left to right. This corresponds to scanning across a series of energy planes in the spectrum image. Observe the changing intensity in the spectrum image. Notice that as the tool is scanned across the edge threshold energy of a given element, regions of the image containing that element appear bright. The resulting display reflects the spatial distribution of the element within the analysed region. Figure 4-24. Energy slice of spectrum image at Al K edge. 17
To extract spectrum from the spectrum image 1. Select the extraction tool from the standard tools floating window. The tool is displayed in figure 4-25 below. Figure 4-25. The Spectrum Extraction tool. 2. Locate a point of interest on the spectrum image and position the spectrum imaging tool at this point by clicking. The acquired spectrum at this position is retrieved and displayed. It may be necessary to expand the energy and/or intensity scales to view features in the spectrum. Figure 4-26. Extracted spectrum with the Al K edge clearly visible. 3. Drag the SI acquisition tool through different parts of the spectrum image. The changes in a edge intensity of a given element is reflective of compositional differences within the analyzed region. 4.5 Analysis The compositional information present in the spectrum image is best illustrated by constructing a background subtracted map. 18
To create a background subtracted map 1. Place (if necessary) the SI tool on the spectrum image to choose a spectrum. 2. Identify an edge energy of interest, and move the tool to an appropriate point in the spectrum image so as to obtain a maximal signal in the corresponding spectrum. 3. On the Standard tools floating window, select the Pointer tool. 4. If the EELS Analysis package is installed, click on the extracted spectrum in the vicinity of a relevant edge and continue with steps 6-8. If EELS Analysis is absent, proceed directly to step 5. 5. The background and signal windows may be specified by clicking and dragging on the extracted spectrum, while pressing the CTRL key. In specifying these windows, ensure that there are no overlaps in the signal window and that the background does not include other edges. Proceed to step 10. Figure 4-27. Specification of signal and background in absence of EELS Analysis. 6. On the Quantification floating window, click the Identify selection button. 7. If the identified edge corresponds to an element suspected to be present in the sample, click Add to quantify list. If this is not the case, select from the quantification element list, a suspected element whose associated edge is closest in energy to the identified element, and then click on Add to quantify list. 8. Repeat steps 4-6 for all edges of interest. 19
Figure 4-28. Addition to EELS Quantification list. 9. Click on the Quantify tab of the Quantification floating window and select a given element from the list. Check the Show windows box to view the background and signal windows superimposed on the spectrum. While dragging the extraction tool over different regions of the spectrum image, ensure that no edges lie in the background window and that there are no overlaps in the signal window. This is essential if reliable maps are to be extracted. Repeat this for all edges of interest. 10. If using EELS Analysis: In the SI menu, for a selected element, for example Al, and corresponding edge, K, select CREATE BACKGROUND SUBTRACTED MAP OF Al K. If not using EELS Analysis, select CREATE BACKGROUND SUBTRACTED MAP from the SI menu. 20
Figure 4-29. Creating a background subtracted map of Al K. As a final step towards visualization of elemental maps in the case of a number of elements, one could generate an color map. This is possible using the Color Mix functionality in DM (version 3.6 ). To generate a color map 1. Bring two or three chosen elemental maps to the front by clicking on the relevant images. For the example shown in figure 5-6 below, the elemental maps correspond to Al, C and N. 2. Select the menu element COLOR MIX from the ANALYSIS menu. Figure 4-30. The Color Mix sub-menu. 21
3. To produce a color map, click the Generate button. Figure 4-31. A color map showing distribution of Al(blue), C(green) and N(red) in the analyzed region. For further details on the use of Color Mix, including the alignment of the individual channels of the final image, please refer to the corresponding tutorial. 22
5 Quick Reference To collect a Digiscan Image 1. Lift up TEM screen. Press Start on View tab of Digiscan floating window. Adjust magnification and focus. 2. Press Stop to terminate acquisition. 3. On SI menu, select ASSIGN SURVEY IMAGE menu element. To acquire a Spectrum Image 1. Use exploration tool to position beam on hole/very thin part of specimen on survey image. If using AutoPEELS, Click Align ZLP on AutoPEELS floating window. Then click Focus on same window. If using AutoFilter, click Align ZLP in AutoFilter floating window. 2. Use edge energies of majority elements (if known) to adjust parameters such Start energy, dispersion and exposure. If using AutoPEELS, this is done within the AutoPEELS View floating window. If using AutoFilter, these parameters may be accessed/set by pressing on ALT key and clicking on each of the Custom, EELS and Turbo buttons respectively. 3. Locate an area on specimen. Position exploration tool within this area to generate spectrum. Press space bar to terminate acquisition. 4. From ROI tools window, select rectangular ROI icon. Click and drag to define an ROI on the survey image. On SI menu select ASSIGN SPECTRUM IMAGE REGION. 5. From ROI tools window, select rectangular ROI icon. While pressing the ALT key, click and drag to define a square ROI (of side 2 N ) on the survey image. On SI menu select ASSIGN SPATIAL DRIFT REGION. 6. From SI menu, select SPECTRUM IMAGE SETUP and verify/adjust parameters accordingly. Then, from SI menu, select ACQUIRE SPECTRUM IMAGE. To create a background subtracted map with EELS Analysis 1. Use the extraction tool to select a spectrum from the spectrum image. 2. From the EELS menu, select the QUANTIFICATION menu item. Click on spectrum in vicinity of edge. On the Quantification dialog, click Identify selection. Verify that the identifed element appears reasonable (if not, select an alternative from the element list). Click Add to quantify list. Repeat for all edges of interest. Click on Quantify tab in Quantification dialog. Select an element from the quantification list using the pointer tool. 3. Bring extracted spectrum to the front. On SI menu, select CREATE BACKGROUND SUBTRACTED MAP. 23