Why STEM Not TEM?
The Transmission Electron Microscope (TEM) and Scanning Transmission Electron Microscope (STEM) were both initially developed in the 1930s. At first sight, they appear to be very similar in function but in reality, they can provide very different types of images and information. In particular, the STEM offers unique imaging modes and enhanced microanalysis capabilities.
The TEM The TEM has a similar optical configuration to an optical microscope. A flood beam of electrons illuminates a thin sample. The electron transmitted through the sample are projected onto a viewing screen or camera for observation (Figure 1). Samples must be thin (around 100 nm) and the beam energies must be high. Electrons may either pass through the sample without being scattered or may be diffracted off axis by interaction with the sample. Electrons may also be backscattered in the sample to re-emerge from the top surface. The primary beam also interacts with the sample to produce characteristic X-rays. With crystalline samples, most of the detail in the image is a result of Bragg diffraction. By choosing the position of the aperture, either the diffracted beam (dark field) or the unscattered electrons (bright field) can be used to form the image. The TEM is used extensively in the life sciences, where its similarity to the light microscope is self evident, and also in materials science. In materials science, the combination of diffraction and imaging provides a unique capability for understanding the properties of crystals and defects in crystalline materials, which can be interpreted in detail. Imaging is possible to the nanometer scale and the spatial resolution can now extend to approach the atomic level.
The STEM The STEM was initially developed at about the same time as the TEM, but its evolution was much slower until the work of A V Crewe in 1970 demonstrated its potential. The STEM operated in a very similar way to a scanning electron microscope (SEM). A fine, highly focused beam of electrons is scanned over a thin specimen (Figure 2). Electrons which pass through the sample can be collected to produce a variety of transmission images, but, as with
the TEM, backscattered electrons and X-rays are also produced. Secondary electrons (SE) are also produced, giving yet another imaging mode. One of the most common ways of carrying out STEM has been to add transmission detectors to an SEM, although this usually limits accelerating voltage to around 30 keV. It is also common to add scaaning coils to a TEM but the minimum probe diameter is large and the resolution of microanalysis is limited. To get the best results, dedicated STEMs such as the HD-2300 from Hitachi HighTechnologies should be used. Such dedicated STEM instruments are much more efficient than a TEM operating in STEM mode, they permit much higher resolution microanalysis than previously attainable in TEM and are also easier to use.
Brightfield imaging in the STEM Transmitted electrons collected on axis in the STEM give the bright field or phase signal. These electrons have either not been scattered at all or have been inelastically scattered through angles of milliradians or less. They exhibit largely crystallographic information. Under certain conditions, the brightfield mode in the STEM is identical to that in the TEM (Figure 3). The TEM and STEM ray paths are identical except for the direction of travel, providing ac = bs and bc = as. For TEM brightfield imaging, ac is approx 10-4 radians (i. e. the beam is collimated) and bc is approx 10-2 radians for best resolution. Obtaining the equivalent conditions in the STEM does give a small spot size, but the detector angle is small compared to the beam convergence so collection efficiency is very low. Thus under TEM conditions, similar images are obtained from the two instruments but the images from the STEM are rather noisier. There is clearly little benefit in operating the STEM in this non-optimum way. However, by using a much larger detector and large as, the STEM becomes much more efficient. Another significant benefit that results from this mode is the facility to look at much thicker samples than is possible in the TEM. This is because as electrons travel through the sample, they lose energy depending on the thickness of the sample. In the TEM there is an objective lens below the sample and electrons of different energies are focused at different focal positions - the well-known effect of chromatic aberration. This leads to blurring of the image and a loss of resolution and contrast. In the STEM, however, there is no lens below the sample so there is no defocusing effect. Bright field STEM can therefore satisfactorily image samples up to a few microns thick at 200 keV, compared to only about 0.5 micron for a
TEM at the same energy. With this extra sample thickness, sample preparation is simplified and there is a better chance of finding the feature of interest in the viewable volume.
Unique Dark Field Imaging Mode The STEM also offers significant benefits in dark field operation with a unique imaging mode, High Angle Annular Dark Field (HAADF) imaging. Here, (Figure 4) the inner angle of the annular darkfield detector is made so large (30 milliradians) that no Bragg diffracted electrons are collected. The images therefore come from elastically scattered electrons which have passed very close to the atomic nuclei in the sample. High (single atom column) resolution is possible with no unwanted diffraction contrast which can mask structural information. The HAADF signal is directly proportional to the density and thickness of the specimen and proportional to Z3/2 where Z is the atomic number. Thus it is possible to produce images which shows contrast due to the mass-thickness (ie the signal is proportional to the number of atoms) or Z contrast images (where the signal is proportional to the atomic number of the sample). HAADF is suitable for inorganic and organic samples and for crystalline and amorphous materials. Figure 5 shows HAADF images from two different samples. The dark field resolution of the STEM is mainly determined by the probe diameter. The HD-2300 has a measured FWHM width of 0.17 nm at 200 keV which brings the ability to achieve atomic level imaging on many samples. Switching between bright field (phase contrast) and dark field (Zcontrast) imaging is done by the simple click of an icon and no additional adjustment of the electron optics is required (Figure 6).
Secondary electron imaging in the STEM An additional bonus offered by the STEM is the ability to collect secondary electrons and backscattered images in the same way as a standard SEM. This makes it possible to correlate surface information (from secondary electrons) with bulk information from the STEM mode. It is also possible to use the secondary electron mode to image samples which are too thick even for STEM observation. The high accelerating voltages available in the STEM offer ultra-high resolution compared to a conventional SEM. Secondary electrons can be produced by the primary electron beam entering the sample or by backscattered electrons leaving the sample. These latter electrons carry bulk information from as far as 250 microns into the
sample at 200 keV for silicon. Figure 7 shows the effect on bright field and secondary electron images from a semiconductor sample of different thicknesses.
Live diffraction imaging The use of an optional live diffraction unit brings even more versatility to the STEM by allowing diffraction images and Z contrast images to be obtained simultaneously (Figure 8). This enables true nanoscale diffraction since the feature of interest can be precisely located and a diffraction pattern from this feature alone can be generated.
Microanalysis As well as its considerable and varied imaging capabilities, the STEM also offers significant microanalysis capabilities. Two microanalysis methods are available: energy dispersive X-ray analysis (EDX) and electron energy loss spectroscopy (EELS). In traditional EDX analysis in the SEM, the spatial resolution is limited to around one micron by the interaction volume of the beam within the sample. Although EDX analysis is possible in the TEM, the spatial resolution is then limited by the ability of the system to focus the beam. In the STEM, however, EDX analysis can be performed at the nanometer scale using thin samples. In addition to improved spatial resolution for EDX, the STEM also offers improved sensitivity compared to conventional TEM. This is because the solid angle of X-ray collection subtended at the EDX detector in the STEM is around 2.5 times larger than that in the TEM, resulting in a similar increase in sensitivity. For thin specimens, high spatial resolution can be achieved, but with low X-ray count rates from thin specimens, it may be necessary to collect the data over extended periods of time. This could be accompanied by specimen drift, which would limit the spatial resolution. The HD-2300 overcomes this problem with a dynamic drift compensation system, in which phase contrast images of the sample are compared to the original image and feedback adjustments made to the position of the electron beam to eliminate the effects of drift. This effect of drift compensation is illustrated in Figure 9. Dopant layers in a semiconductor device at levels lower than 0.001% of As can be mapped and a 2nm think oxide layer can be clearly observed. Another analysis method uses electron energy loss spectrometry of the transmitted electrons. This also offers elemental mapping at 2 nm spatial resolution. In addition, this technique achieves real time mapping over as little as 40 seconds (see Figure 10) and is particularly useful for light elements and transition metals,
since there is no overlap in the detection of the elemental peaks in the spectrum, unlike the EDX case. The ability to achieve two dimensional nanoscale analysis by these techniques, which was previously unobtainable, will have significant impact in materials investigation. Ease of use Even though TEMs have been commercially available for many years, the technique remains a specialist one, and dedicated operators are generally required to achieve the best results. The operation of the STEM, however, is remarkably similar to that of a scanning electron microscope, and the HD-2300 uses a similar user interface to the Hitachi range of SEMs. This means that SEM users can quickly adapt to the STEM technique and they can can perform high resolution imaging and nano-analysis quickly and routinely. The STEM is extraordinarily versatile with its unique imaging properties and is likely to become a much more widely used technique in the coming years.
Author: Paul Ansell, EM Sales Manager and Michael Dixon Product Group Manager, Hitachi Scientific Instruments, 7 Ivanhoe Road, Hogwood Lane Industrial Estate, Finchampstead, Wokingham, Berkshire, RG40 4QQ. Press Enquiries: Denis Bulgin, In Press Public Relations Ltd, PO Box 24, Royston, Herts, SG8 6TT. Tel: 01763 262621 Fax: 01763 262655 E-mail:
[email protected]. 208HIT-EM Approx. 1650 words 10 illustrations
Legends for figures Figure 1. Schematic for TEM Figure 2. Schematic for STEM Figure 3. Conditions for equivalence between TEM and STEM Figure 4.Conditions for HAADF, brightfield and SE imaging Figure 5. HAADF images of: (a) mouse duodenum (b) gold treated nanotubes Figure 6. (a) Phase contrast image (b) Z-contrast image Figure 7. Bright field and secondary electron images from a semiconductor sample of different thicknesses Figure 8. Simultaneous diffraction and Z contrast images from Au foil Figure 9 (a).High resolution EDX mapping of an N-MOS transistor without drift compensation (b).High resolution EDX mapping of an N-MOS transistor with drift compensation Figure 10. Electron energy loss image with thickness measurement of SiO and SiN layers
Figures (page 1 of 4)
Figure 1
Figure 2
Figures (page 2 of 4)
Figure 3
Figure 4
Figures (page 3 of 4)
Figure 5a
Figure 5b
Figure 6a
Figure 6b
Figures (page 4 of 4)
Figure 7
Figure 8
