Encyclopedia of Glass Science, Technology, History, and Culture. Группа авторов
to improve spatial resolution. As described in Chapter 2.2, the first is to probe the material with photons of shorter wavelengths such as X‐rays: at 154 pm, the wavelength of Cu‐Kα X‐rays is for instance 1000 times shorter than that of visible light. A second way is to take advantage of the wave‐particle dualism of electrons or ions. Under accelerating voltages of 25 and 200 kV, the electron wavelengths are even shorter than that of typical X‐rays, with values of 8 and 2.5 pm, respectively. As a matter of fact, the former value is typical for studies of sample surfaces by scanning electron microscopy (SEM), whereas the latter applies to investigations by transmission electron microscopy (TEM) of bulk samples with a thickness of a few tens of nanometers at most, to limit electron absorption. In addition to these two techniques, to which we will pay particular attention, the third way to achieve high resolution will also be described. It relies on the use of scanning probes whereby a needle that is atomically sharp at its end is moved over the sample surface by a piezo drive that must be highly precise, since the key to resolution is, in this case, the accuracy of the needle position.
In practice, the formation of a given microstructure can be extremely complicated, including secondary, ternary, or even quaternary phase separation or the nucleation of multiple functional phases on finely dispersed seed crystals [1, 2]. Rather than trying to document all such processes, we prefer to illustrate the capabilities and complementarity of the methods described in this chapter with a single glass ceramics whose starting composition (in mol %) is 50.6 SiO2 · 20.7 MgO · 20.7 Al2O3 · 5.6 ZrO2 · 2.5 Y2O3, whence its MAS acronym from its main components MgO, Al2O3, and SiO2. Under an appropriate temperature and time treatment, this glass transforms to a glass ceramics whose microstructure shows a variety of features with which the pros and cons of the most common techniques are nicely illustrated [3–17]. And this glass ceramics also illustrates how a specific material property can be explained only in microstructural terms: in this instance an unusually high bending strength exemplifies differential expansion strengthening (Chapter 3.12) resulting from precipitation of ZrO2 (zirconia) and MgAl2O4 (spinel), two crystals with high thermal expansion coefficients, in a glassy matrix of lower expansivity.
2 Acronyms
AFMatomic force microscopyBSEbackscattered electron(s)CLcathodoluminescenceEDXSenergy‐dispersive X‐ray spectroscopyEELSelectron energy loss spectroscopyFIBfocused ion beamHAADFhigh‐angle annular dark field (imaging)MASmagnesium aluminosilicate (glass ceramics)SAEDselected area electron diffractionSEsecondary electron(s)SEMscanning electron microscopySTEMscanning transmission electron microscopyTEMtransmission electron microscopyXRMX‐ray microscopyWDXSwavelength‐dispersive X‐ray spectroscopy
2 Scanning Electron Microscopy
2.1 Image Formation
In SEM a finely focused primary electron beam possessing an energy in the 100 eV–30 keV range impinges onto the surface of a sample placed in vacuum (Figure 1). By interacting with the sample atoms, electrons generate many different signals. If the primary electron beam is scanned across the surface, an image can be formed not only from secondary electrons (SE) but also from backscattered primary electrons (BSE) and cathodoluminescence (CL). In addition, the chemical composition of the sample surface can be determined from the generated X‐rays and Auger electrons.
The primary beam diameter can be as small as a few nanometers, but the magnification is determined by the distance between the positions of the beam successively held to acquire one of the secondary signals and to generate an intensity signal in the digital image derived. Well below the sample surface exposed to the scanning electron beam, however, even the smallest primary beam undergoes some scattering upon interaction with the atoms in the bulk of the sample. Channeling effects may also occur if the sample is fully or partly crystalline, i.e. electrons may penetrate more deeply into some areas of the sample before scattering if a specific crystallographic orientation is well aligned with the direction of the electron beam. The spatial resolution of an SEM micrograph is, therefore, rather determined by the interaction volume (with typical extensions of several 100 nm in width and depth beneath the surface; cf. Figure 1) than by the initial spot size of the beam.
Depending on the type of (secondary) signal used for imaging and analyzing the sample, the depth of information and, accordingly, the lateral resolution as well will be very different. For instance, SE have energies lower than about 50 eV. Those that leave the sample surface can thus reach the detector system only if they stem from the “neck” of the excitation volume, i.e. down to approximately 50 nm beneath the surface. The SE generated at greater depths do not reach the surface to leave the sample because of self‐absorption effects resulting from their rather low energy. On the contrary, the highly energetic backscattered electrons (BSE) – as well as the X‐rays used for chemical analysis – may also stem from the “bulk” of the excitation volume, further below the sample surface down to some 100 nm. This interaction volume strongly depends on experimental factors that can be adjusted to some extent. For instance, reducing the acceleration voltage reduces the sample penetration depth and, thus, the size of the resolution‐determining excitation volume. It may also help to increase the resolution by reduction of sample charging, a problem that is always encountered when nonconducting samples (like most glasses) are analyzed.
Figure 1 Interaction volume and main secondary signals occurring when a primary electron impinges on a sample surface.
Such adjustments, however, might worsen other aspects of the experiment. When probing the sample composition from the energy of the characteristic X‐rays emitted, the primary electron beam should, for instance, have an energy that is high enough to generate these X‐rays. In other words, an acceleration voltage amounting at least to twice the X‐ray energy to be excited is needed, in spite of its ensuing implications on interaction volume, charging, and also sample degradation. In summary, one should always think first about the intended experiment – the crucial information to be gained – and adjust the experimental conditions accordingly.
An advantage of SEM is that sample preparation is simple. Glasses and glass ceramics are typically just ground and polished. More specific techniques rely on wet‐chemical etching to generate a topography related to the differing dissolution rates of the constituents of the microstructure (e.g. the vitreous “matrix” vs. crystalline precipitates) and on focused ion beam (FIB) etching to get access to regions buried under the surface.
When glasses are electric insulators, care must be taken to avoid accumulation of electrical charges on the surface, which could influence image formation by deflecting ingoing or outgoing electrons, particularly those that possess only low energies. Coating the sample with a few nanometers of carbon helps drain away such charges, but potentially covers fine details of the microstructure and adds a local source of contamination by carbon redistribution on the surface. Another way to prevent charging is imaging under poor vacuum conditions. This is done in so‐called environmental or vapor‐pressure SEMs, where a certain partial pressure of, for example, water vapor is maintained above the sample surface, whereas a pressure gradient ensures that the electron‐emitting source is kept in ultrahigh vacuum. In this experimental setup, primary electrons ionize water molecules that eventually neutralize charges on the sample surface. Last but not least, the total electron yield (SE and BSE leaving the sample) depends on the acceleration voltage. Provided that this voltage can be properly varied, conditions can be found such that as many electrons enter the sample as leave it, resulting in the absence of charging. Typically,