As previously stated, imaging structures requires that the radiation have a wavelength on a similar scale to the features which are to be observed. Therefore, for nanometre scale features this makes visible light microscopy unsuitable, X-rays can be used for diffraction experiments but are not suitable for imaging. Transmission Electron Microscopy (TEM) is a commonly used method for imaging nanoscale structures, where imaging is achieved with monochromatic beams of electrons with a
Figure 3-14 Unprocessed tiff image from Pilatus detector showing (004) peak from a GeSn epilayer. The vertical axis is in the 2θ direction and the
de Broglie wavelength sufficiently small such that the features of interest are observable.
In this work cross-sectional TEM (X-TEM) imaging is used to determine Ge1-xSnx
epilayer and Ge buffer layer thicknesses; to obtain qualitative information of the crystal quality by determining if the epilayer material is monocrystalline, polycrystalline or amorphous; determine the type, location and prevalence of lattice defects including differentiating between edge and threading dislocations which can be identified by inspection. TEM is a commonly used characterization method in materials science as it provides a wealth of knowledge about a material.
3.7.1 Theory
The minimum feature size resolvable by electrons is limited by the de Broglie wavelength used for imaging, which is proportional to the energy of the electron as given by equation 3-5 the general de Broglie wavelength, equation 3-6 the de Broglie wavelength for electrons. In TEM the electrons are accelerated to relativistic velocities, and this must be taken into account, as done in equation 3-7.
𝜆 = ℎ 𝑝= ℎ 𝑚𝑣 3-5 𝜆 = ℎ √2𝑚0𝑒𝑉 3-6 𝜆 = ℎ √2𝑚0𝑒𝑉(1 + 𝑒𝑉 2𝑚0𝑐2) 3-7
Where λ is the electron de Broglie wavelength, V is the accelerating voltage, h is Plank’s constant, m0 is the electron rest mass, e is the electron charge and c is the
have a wavelength of 0.0025 nm, which is smaller than the features to be investigated in this work.
Unlike lab based X-ray wavelengths, which is limited to the emission lines of metal targets, in TEM the wavelength of electron beam used as a probe can be chosen simply by controlling the accelerating voltage, 200 kV is standard for imaging semiconductors. This value is a compromise, while a higher accelerating voltage decreases the electron wavelength, the increased electron kinetic energy can produce significant damage to samples. While lower electron beam energies, and hence larger wavelengths, do not facilitate as high resolution.
For many electron microscopes any additional resolution facilitated by using acceleration voltages above 200 kV cannot be utilised due to aberrations to the electron beam. Beam aberrations are caused by imperfect electron sources and imperfect electron lenses and reduce the attained image resolution. A schematic diagram of a transmission electron microscope is shown in Figure 3-15. The microscope requires many electron lenses and thus aberrations from each lens accumulate contributing to a corrupted, i.e. lower resolution, image. For particularly sensitive samples, such as many biological samples, lower accelerating voltages of 120 keV or 80 keV are commonly used in order to reduce sample damage from the electron beam, at the cost of attainable resolution. The theory and practice of TEM measurements is covered in detail in Williams and Carter, Ref. [120].
X-TEM observations are made through a cross-sectional volume of material, which must be thinned to be transparent to the electrons, as discussed below. All TEM in this work uses the JEOL JEM-2000 FX operating at 200 kV, unless otherwise stated.
Figure 3-15 A schematic diagram of the configuration of transmission electron microscope, with series of electron lenses, with their purpose and also marked is the path of the electron beam. The electron source produces
high velocity electrons, which are condensed, transmitted through the sample, focused and then either directly projected onto a CCD or
3.7.2 Sample Preparation
In this work TEM samples were prepared by the ‘lapping’ method, rather than using a focused ion beam. For this method wafers are cleaved to produce two approximate 1 × 1 cm squares. The surface is cleaned with acetone, then a thin layer of adhesive is spread across the surface, and the two pressed together. Both ‘backs’ of the sample are cleaned with acetone and a ~1 × 1 cm square of plain silicon wafer coated with adhesive is pressed to the back of the sample, as shown in left of Figure 3-16. The complete wafer bar is compressed as the adhesive sets. Once the adhesive sets, the structure is sawn in half to produce a cross-section, one half is attached to a metal block and then sanded down along the cross-section until transparent to red light. A copper ring is glued to the sample with the interface of interest at the centre of the ring. A precision ion polishing system is then used to further thin the central interface with argon ions, as show in Figure 3-16.
Figure 3-16 - Schematic diagrams of the stages in sample preparation for TEM. (left) Initial gluing of ‘sandwich’ surfaces together completed bar structure. Subsequently the bar is cut with a diamond saw, the sample is then
attached to metal block, ground and then polished. Finally a TEM copper support ring glued to the sample surface. (right) Precision ion polishing system beaming process, with argon ions used to thin the central interface
Samples which cannot be prepared by this lapping method, for example if they are particularly hard such as SiC, or too delicate such as electrical devices, can be prepared for TEM by cutting a cross-section and attaching it to a copper TEM grid using a focused ion beam-scanning electron microscope (FIB-SEM). While this method can be used for materials unsuited to the standard preparation method, it is significantly more time consuming and can induce significant ion implantation damage to the surface of the sample under investigation.
3.7.3 Imaging
In TEM imaging is achieved with the electron beam being projected after passing through the sample. The electrons are detected by a charge couple device (CCD) or phosphorus screen, traditionally photos were obtained using photographic film, but this has become less common. Image contrast comes from the detected electron
Figure 3-17 A TEM image of a Ge1-xSnx/Ge/Si structure in the (220) dark
field diffraction condition. The different layers have a different contrast. The interface between the Ge buffer and the Si substrate is visible and the lattice misfit dislocations. At interface the GeSn epilayer and Ge buffer no defects
intensity. Changes in image contrast occur in part due to the dependence on the material atomic number, allowing images to have rudimentary elemental contrast. The thickness of material also effects the contrast, with a hole in the samples giving a very bright signal and a thick sample appearing dark. A TEM image of a GeSn/Ge/Si structure is shown in Figure 3-17, each layer has a different contrast, the slightly undulating pattern of light and dark contrast is a thickness effect of the sample.
3.7.4 Imaging in a Diffraction Condition
When studying crystalline materials in TEM, a diffraction lens can be used to produce a diffraction pattern, containing ‘Kikuchi lines’ from many lattice planes [120]. In diffraction contrast imaging the sample is tilted to align on a particular lattice plane. An aperture is used to select a section of the beam from a particular diffraction condition, excluding signal from other contributions, including the straight through beam. This aperture increases the relative contrast of sample features under investigation. By aligning onto a particular lattice plane, particular sample attributes are emphasised. With Si(001) samples it is common to use to (004) and (220) to determine epilayer thickness with strong compositional contrast, and also to examine the prevalence of lattice dislocations, respectively.
3.7.5 Measurement Attributes
TEM is a commonly used characterization technique due to its versatility. It can be used to determine layer thickness, relative crystal quality, prevalence of alloy segregation, and any major surface features. An indication of alloy composition can be inferred from the contrast of layers, though with little accuracy other than the difference between layers. In this work TEM is the sole method able to identify
defect concentration. For TEM measurements, the sample is consumed during preparation and this process is time consuming.