Thin foil analysis in the SEM final revisedfy.chalmers.se/~f10mh/Halvarsson/EM_intro_course... · One of the main reasons for studying thin foil specimens, apart from BF and DF imaging,

Thin foil analysis in the SEM

M Halvarsson1, T Jonsson and S Canovic

Microscopy and Microanalysis, Department of Applied Physics, Chalmers University of Technology, SE-412 96 Göteborg, Sweden

E-mail: [email protected]

Abstract. This paper explores the possibilities for imaging and chemical analysis of thin foil specimens in the SEM. Bright field and dark field imaging provide high resolution imaging with crystallographic information within the grains. In multiphase materials with varying electron transmission the dark field images generally provide a more even contrast in all phases. It is possible to obtain high-quality quantitative EDX data with high spatial resolution.

1. Introduction Scanning electron microscopy (SEM) is widely used for analyzing bulk samples. Often secondary electron (SE) and backscattered electron (BSE) imaging are used in combination with chemical analysis using energy dispersive X-ray analysis (EDX), see e.g. [1]. In order to improve the spatial resolution in the BSE images and the EDX analysis a lower acceleration voltage than the standard 20 kV may be used. One problem is that the EDX spectra do not contain information higher than the energy of the incoming electrons, which means that higher energy peaks are not present. This limits the EDX analysis to the low-energy part of the spectra, where the peaks generally are less well separated. For heavier elements this means that L- or M-peaks have to be used. For quantification these are theoretically more poorly described than K-peaks.

Another way to increase the spatial resolution in the SEM is to make use of thin foil specimens instead of bulk specimens. This will limit the interaction volume in the same way as in the Transmission Electron Microscope (TEM), without reducing the acceleration voltage. For the thin foil analysis there are special Scanning Transmission Electron Microscopy (STEM) detectors available that are placed below the sample in the SEM [2]. Electrons traveling straight through the sample may be detected by a bright field (BF) detector and scattered electrons may be detected by dark field (DF) detectors. This means that there is more information (than SE and BSE imaging) available from the sample by using the bright and dark field imaging modes.

The aim of this paper is to explore the possibilities for imaging and chemical analysis of thin foil specimens in the SEM. A comparison with thin foil analysis in the TEM will also be made.

2. Experimental Two different samples have been used in this work. One is an oxidized molybdenum disilicide composite with a thin alumina scale formed after oxidation in lab air at 1450°C for 5 minutes. Thin 1 To whom any correspondence should be addressed.

Electron Microscopy and Analysis Group Conference 2007 (EMAG 2007) IOP PublishingJournal of Physics: Conference Series 126 (2008) 012075 doi:10.1088/1742-6596/126/1/012075

c© 2008 IOP Publishing Ltd 1

foil specimens (~100 nm thick) were produced using the lift-out technique in a combined Focused Ion Beam/SEM instrument of the type FEI Strata 235 DB.

The scanning electron microscope used was a Leo Ultra 55 equipped with a field emission gun (FEG). The acceleration voltage was 30 kV. A STEM detector with 5 diodes was used. One central diode is for BF imaging and the other four are for DF imaging. An Oxford Inca EDX system is attached to the microscope. It has software for evaluation and quantification of both bulk and thin foil specimens. A Philips CM200 FEG TEM was also used for STEM imaging at 200 kV.

3. Results and Discussion A lift-out sample placed on a Cu-grid is shown in figure 1. The sample has been clamped in a special thin foil holder and some of the STEM detectors are seen below the sample. The BF detector is located directly below the sample, while there are four DF detectors placed around the BF detector. In the software any combination of detectors can be used for image formation. In this paper, in addition to the BF mode only a DF mode with two diagonally positioned detectors turned on were used; the other two DF detectors were then turned off. Let us consider Braggs law, 2dhkl sinθ = λ , and the geometry of electron scattering to the detectors, tan 2θ

!

= x /L , where dhkl is a typical interplanar distance for the material studied, λ is the electron wavelength at 30 kV, 2θ is the total scattering angle, 2x is the distance between two diagonal DF detectors and L is the distance between the specimen and the detector plane (“camera length”). It is possible to calculate an appropriate specimen-detector distance (L) using typical numbers for the other parameters (d=0.2 nm; λ=0.00698 nm; x=50 µm). One finds that L≈1.4 mm. Thus, the STEM detectors should be placed more than 1.4 mm below the thin foil sample in order to let Bragg-scattered electrons hit the DF detectors and not the BF detector, and thereby creating high contrast in the DF and BF images.

Figure 1. (a) The thin foil sample (arrowed) is placed on a Cu grid above the BF diode. The four DF diodes are placed around the BF diode. (b) Close-up of the specimen and the underlying diodes. The sample is shown in SE-mode in figure 2(a), in BF-mode in figure 2(b) and in DF-mode in

figure 2(c). It is obvious that these three images carry different information. The different phases can be seen in the SE and DF images, while the BF image mostly carries information within the oxide scale. It is obvious that the transmission through the molybdenum disilicide composite is much lower than for the alumina scale. This is due to the relatively low acceleration voltage used (30 kV). If the sample is imaged in a TEM at 200 kV the transmission is more equal in the oxide scale and the subjacent material, as can be seen in figure 2(d), where the different grains in the molybdenum disilicide composite are clearly visible. As can be seen in figure 3, the dark Mo rich particles within the alumina scale are easily visible, even if their sizes range down to a few nanometers. The resolution in the STEM images in the SEM is generally stated to be just below 1 nm. If the sample contains phases that vary much in transmission in the BF mode, the DF mode often produces better images, as seen in figure 2(c), with a more even contrast.

40 µm

DF

DF

DF

BF

a b

500 µm


2

One of the main reasons for studying thin foil specimens, apart from BF and DF imaging, is the possibility of obtaining EDX data with high spatial resolution in the 0-20 keV range. EDX maps produced at 30 kV are shown for oxygen and aluminium in figure 4. As the oxide scale is about 2 µm in thickness and details are easily recognized within the scale it is clear that the spatial resolution is quite good. Generally it is of the same order as in the TEM. The specimen thickness in our case is estimated to be 100 nm and therefore the EDX spatial resolution is around 10 nm. Extra care must be taken when mounting the thin foil samples for EDX analysis. There are STEM specimen holders on the market that will generate spurious X-rays in the spectra. For instance certain holders made from aluminium will generate extra Al X-rays that are detected by the X-ray detector. It is of importance that the holder has as little material as possible close to the specimen and that transmitted/diffracted electrons do not generate extra X-rays that can be detected. The holder should also provide a free path to the detector for the X-rays originating from the specimen. In our case a homemade holder based on a TEM graphite holder for Jeol 2000FX was used.

Figure 2. Images of the molybdenum disilicide composite thin foil sample. (a) SE – SEM image with labels (b) BF – SEM image (c) DF – SEM image and (d) BF – TEM/STEM image.

Figure 3. BF image of dark Mo rich particles located in the alumina scale (SEM).

a b

c d

3 µm

100 nm

Mo5(Al,Si)3

Mo(Al,Si)2

Pt

Al2O3


3

Figure 4. EDX maps acquired from a thin foil specimen in the SEM; oxygen and aluminium.

Table 1. EDX results (at.%)

Spectrum O Al Si Mo Mo(Al,Si)2 1 2.8 29.7 34.3 33.1 Mo(Al,Si)2 2 2.0 29.5 34.8 33.7 Mo(Al,Si)2 3 2.0 30.4 34.2 33.4 Alumina 1 60.1 39.6 0.2 0.1 Alumina 2 62.2 37.6 0.0 0.1 Alumina 3 61.9 38.2 0.0 0.0

Of key importance is the quality of the EDX quantification results. A relevant question is if the thin

foil approximation in the EDX quantification algorithms is valid when the electron energy is only 30 keV in the SEM, instead of 100 – 300 keV as in typical TEMs. Therefore a number of EDX analyses were carried out on phases with known composition. In table 1 results from standardless analysis from single phase regions of alumina and Mo(Al,Si)2 are shown. The errors are estimated to be less than 1 at.%. The nominal composition of alumina is 40 at.% Al and 60 at.% O. For Mo(Al,Si)2 the composition should be 33.3 at.% Mo with the sum of Al and Si being 66.7 at.%. By comparing these values with the measurements in table 1 it can be concluded that the agreement is excellent. Thus, it is possible to correctly quantify the composition of thin foil specimens in the SEM.

4. Concluding discussion Using STEM analysis of thin foils in SEMs extends the information that is possible to obtain in these instruments. The BF mode not only provides spatial resolution in the nanometer range, the images generally carry more crystallographic information within the different grains than SE and BSE imaging do. However, in multiphase materials with varying electron transmission the BF images can be difficult to acquire with good contrast in all phases. Then the DF mode is often more useful.

The spatial resolution in EDX maps and spot analysis is generally of the same order as in the TEM, i.e. 10 nm. However care has to be taken when using available STEM holders on the market, as some will produce a large number of spurious X-rays registered by the detector. The EDX thin foil quantification works very well even if the acceleration voltage in the SEM is relatively low (30 kV) compared to standard TEMs (>100kV). Due to the generally easier use of SEMs than TEMs, the results presented in this paper suggest that SEM/STEM challenges TEM/STEM in many cases.

Acknowledgements This work was carried out within the Swedish High Temperature Corrosion centre (HTC). A grant from Knut and Alice Wallenberg Foundation for acquiring the FEG-SEM is gratefully acknowledged.

References [1] J. Goldstein et al., Plenum Press, New York, 2003. [2] J. P. Vermeulen, A Novel STEM Detector System, 2005 Imaging & Microscopy 1 22–23

O Al

3 µm


4

Documents

Thin foil analysis in the SEM final revisedfy.chalmers.se/~f10mh/Halvarsson/EM_intro_course... · One of the main reasons for studying thin foil specimens, apart from BF and DF imaging,