Collecting Tomographic Tilt Series in STEM

By Chris Booth and Robin Harmon, Gatan Software, Pleasanton, CA

Combining Scanning Transmission Electron Microscopy (STEM) with Tomography provides a unique opportunity and leverage the strengths of both techniques to address questions at the leading edge of science. STEM Tomography enables researchers to use directly interpretable Z-contrast imaging while trying to understand complex 3D specimens^{1, 2}. With all the advantages that the two techniques provide, there are still some significant challenges that need to be overcome to make this technique routine for data acquisition. Fortunately there are now many tools to help overcome the problems associated with the collection of STEM Tomography Data. These include holders that help to minimize the missing wedge problem, STEM detectors that utilize collect bright field, dark field and high angular dark field signals and software that robustly automates much of the data collection process.

Holders Designed To Minimize the Missing Wedge

One of the major limitations of electron tomography is that it is extremely difficult to tilt a specimen through 180 degrees in an electron microscope³. Specimen preparation and goniometer design both contribute to this problem, making it difficult to image a specimen through the full range of angles needed for a complete reconstruction. STEM Imaging compounds this problem in that the analytical objective pole pieces typically used for high resolution STEM Imaging are often even smaller than those traditionally used in TEM Tomography. There are two approaches that can be used to minimize missing data when in electron tomography.

The first approach to maximize the achievable tilt angle is to make the holder tip as small as possible. In some cases reusable tips that can be securely fixed to the specimen can be fitted into the end of a high tilt holder minimizing the part of the holder that could touch the pole piece at high tilt angles.

The end of Gatan’s 912 high tilt tomography holder (left) and an array of reusable tips (right) that can be clipped into the end of the holder. Gatan’s 912 high tilt tomography holder is designed to maximize the tilt range that can be achieved in narrow gap pole pieces typical in STEM Applications.

An alternate way to try to minimize the missing wedge is to use dual axis tomography³. Dual axis tomography requires specially designed holders that can rotate 90 degrees in plane. In this scheme, two orthogonal tilt series are collected of the same object.

Gatan’s 927 Dual Axis Tomography holder allows an 90 degree in plane rotation so that two orthogonal tilt series may be collected. Dual axis tomography merges the two orthogonal datasets to minimize the missing wedge effect.

For more information on Gatan’s complete line of tomography holders please visit http://www.gatan.com/holders.

Detectors Designed To Take Make Image Interpretation Easier

One of the primary advantages of using STEM is the ability to collect multiple signals simultaneously (bright field (BF) images, low angle annular dark field (ADF) images, and high angle annular dark field (HAADF) images are all common).

ADF and BF signals can be useful very useful because the signal intensity is relatively high compared with other types of imaging. However, one drawback of BF and ADF STEM is that for crystalline specimens electron diffraction can cause dramatic contrast reversal that is dependent on the orientation of the specimen rather than the specimen material itself. In STEM Tomography this problem is especially significant because the orientation of the specimen changes throughout the tilt series.

The HAADF signal is related primarily to the atomic number of the sample material (termed Z contrast). HAADF images are less affected by diffraction contrast and for crystalline specimens can provide an image that is easier to interpret.

A schematic representation of the solid angle collected by the Gatan 805 ADF detector (left) and an example of a semi-conductor device imaged with the 805 ADF Detector.

A schematic representation of the solid angle collected by Gatan’s 806 HAADF detector (left) and an example of the same semi-conductor device imaged in the previous figure. Note the absence of diffraction contrast in the image in comparison with the previous image.

In some applications it is useful to be able to acquire the BF or ADF signal simultaneously with the HAADF signal. Some detectors have been designed to allow the collection of both types of signals simultaneously. Having a BF and HAADF signal from the same experiment can allow one to understand both the high Z elements and the lighter low Z elements from the same specimen.

For more information on Gatan's new HAADF STEM detector, please visit http://www.gatan.com/analysis/806.php


Robust STEM Tomography Automation

In principle, tomography is simply a matter of tilting the specimen stage and collecting images. Unfortunately, the picture gets a little bit more complicated when doing high resolution electron tomography. Any stage tilt will result in an additional movement of the stage in X, Y or Z that must be addressed by any software package that hopes to collect tomography data sets automatically. STEM Imaging brings its own challenges to tomography, as measures of focus traditionally used in TEM are not applicable in STEM mode, and it is impractical to switch back and forth between TEM and STEM to adjust focus. Keeping the probe focused on a tilted specimen is a major challenge in STEM Tomography and needs to be handled well by the software and hardware.


Gatan’s STEM Tomography Software. The palette on the right allows the user to interact with the running tilt series collection, to pause, verify and continue the acquisition process. The top two images shown are a simultaneous 806 HAADF series (left) and a 805 BF series (right) being collected. The middle three images are from the x,y feature tracking measurement used in this data set. Note that the effective magnification of the x,y feature tracking does not have to be identical to that used for high quality data collection (top). The bottom image shows a snapshot of the STEM focus tracking algorithm that maximizes image signal to noise ratio. The line plot on the right shows the cumulative feature tracking that has been performed by the software

Feature tracking is the fundamental issue that automated tomography software needs to solve to collect useful data. Only those images which contain the region of interest from a tomography data set are of any use in 3D reconstruction. There are three typical ways for automated software to track the region of interest through a tilt series: by using a stage calibration to predict feature movement, by real time measurement of a feature’s movement throughout the tilt series, or by using a combination of prediction and measurement to track the region of interest as the stage is tilted. A summary of the strengths and weaknesses of each method is summarized in Table 1. One feature that is important for real time measurement of feature movement is to be able to select a lower magnification for tracking being used for tilt series collection. This is especially important when working at a very high magnification or if there region of interest has the potential to shift beyond the field of view when the tilt angle is incremented. Automated tomography software should provide a way for the user to follow the region of interest by taking measurements at a lower effective magnification.

Table 1: A comparison of the different methods for feature tracking provided by Gatan’s Tomography Software. Note that the setup times for Stage Calibration and Combination tracking are initially longer because they require a stage calibration to be performed. Once a stage calibration exists, these methods only take a few minutes to setup and run.

Automated STEM Imaging and STEM Tomography in particular require methods for automatically focusing the STEM probe on the specimen that are fundamentally different from those used in TEM. Algorithms for STEM focusing should work on a variety of specimens and at a variety of magnifications. Typically STEM images are focused by maximizing the image variance while changing the illumination conditions. STEM focusing methods that simply maximize the image variance are likely to fail for a number of reasons. High image variance can be achieved by imaging far from focus and in practice we have found that STEM focusing routines that search for high image variance are not robust when using a wide variety of specimens. Alternatively STEM images can be focused by trying to maximize the image signal to noise ratio. We have found that these methods are effective across a broad range of magnifications and specimens (nanowires, precipitates, calibration samples).

In tomography, once the STEM probe is focused in the middle of the image, the probe focus still needs to be adjusted during the acquisition to keep the image in focus on tilted specimens. This is particularly critical for modern Cs corrected STEM due to the narrow depth of field resulting from the large convergence angles available.

A schematic representation of the difference between a tilted STEM image acquired with Dynamic Focus and without as the electron beam is scanned across the image. The electron beam is represented in green and the tilt specimen is represented by the diagonal line in the middle. On the left the probe is only focused on the tilted specimen near the middle of the image while on the right the entire image is focused as sharply as possible through the entire image.

An example of a STEM Image acquired with Gatan’s STEM Tomography software during a tilt series without using Dynamic Focus (left), and with Dynamic Focus enabled (right). Looking at the magnified region from the top of these images the STEM probe is not able to be focused at the top of the tilted specimen when dynamic focus is not used.

Due to the valuable nature of the data and the time it takes to acquire a tomography tilt series electron microscopists often wish to pause the running process, verify that the imaging conditions are still optimal and continue the acquisition with having to start over from the beginning. Automated Tomography Software should make this possible and allow the microscopist to reacquire any previous data without having to restart the entire acquisition. In addition, software should report back to the user if it is unable to complete a given feature tracking step. The addition of these two features can give the microscopist confidence that the data being collected is of the highest quality possible.

For more information on Gatan’s Tomography Software including TEM tomography and EFTEM Tomography please visit http://www.gatan.com/software/tomography.php.

Click on either image to see the associated movie clip

An interplanetary dust particle was imaged from -61 to 68 degrees on a Gatan 805 BF detector and a Gatan 806 HAADF detector simultaneously. Study of these types of objects enables identification of the distribution and quantity of metal particles embedded in the glass matrix, as well as the morphology of the matrix itself. This dataset is provided courtesy of Dr. Ilke Arslan of Sandia National Laboratories Dr. John P. Bradley of Lawrence Livermore National Laboratory

For information on ways to reconstruct, visualize and interpret tomography datasets look for a follow up to this article in the next KnowHow edition.

Gatan would like to thank Dr. Ilke Arslan of Sandia National Laboratories for providing samples and helping to test the 806 HAADF detector and the STEM Tomography Software. Gatan would also like to thank Dr. John P. Bradley of the Institute for Geophysics and Planetary Physics at Lawrence Livermore National Laboratory and Dr. Roseann Csencsits of the Lawrence Berkeley National Laboratory for kindly providing specimens for study.

References:

1. Midgley, P.A., Weyland M., Yates T.J.V., Arslan I., Dunin-Borkowski R.E., Thomas J.M. Nanoscale scanning transmission electron tomography. Journal Of Microscopy. 2006, 223(3), pp 185-190.

2. Friedrich H., McCartney M.R., Buseck P.R., Comparison of intensity distributions in tomograms from BF TEM, ADF STEM, HAADF STEM, and calculated tilt series. Ultramicroscopy. 2005, 105(1), pp 18-27.

3. Arslan I., Tong J.R., Midgley P.A., Reducing the missing wedge: High-resolution dual axis tomography of inorganic materials. Ultramicroscopy. 2006, 106(11-12), pp 994-1000.

We hope that you will find this news letter both interesting and useful. If you do not, simply click here to be unsubscribed and removed from our list.

Gatan Inc. Corporate Headquarters, 5933 Coronado Lane, Pleasanton, CA 94588
Tel. (925) 463 0200 Fax. (925) 463 0204
Contact: info@gatan.com