Search

Electron spectrometers for low-energy electrons can be built with a better resolution and offer different spectroscopic modes for secondary, Auger, and backscattered electrons. Energy-filtering of lower-energy SE and an analysis of the spin polarization become an efficient method for investigation surface magnetism.

A vibrating Kelvin probe in form of a platinum wire loop is used to measure the surface potential Us on electron-irradiated free-floating metal and insulator specimens as a function of electron energy E. This allows an accurate measurement of the critical electron energy E2 for no charging. At energies below E2, the positive charging increases with decreasing energy to Us=2–5 eV at E=0.5 keV and switching off the collector bias of the Everhart-Thornley detector. A two-to threefold increase of Us is observed when the bias is switched on. For E > E2, the strong increase of a negative surface potential can be measured. Insulating films free-supported on a conductive substrate show a steep decrease to small positive and negative Us when the film thickness becomes lower than the electron range at a critical energy E3 > E2. At insulating specimen the temporal decrease of charging can be measured when the electron beam is switched off.

Solutions of the stationary and time-dependent equations of diffusion are presented for contamination when scanning a rectangular area in the scanning electron microscope (SEM). A method is described to record the thickness of the contamination layer by the signal of secondary or backscattered electrons and to measure the influence of electron current density, beam energy, and specimen cooling on the contamination rate.

In-depth imaging of subsurface structures in scanning electron microscopy (SEM) is usually obtained by detecting backscattered electrons (BSE). For a layer-by-layer imaging in BSE microtomography, it is preferable to use an energy filtering of BSE. A simple approach is used to estimate the contrast by using backscattering coefficients of bulk materials and the maximum escape depths of the BSE. The contrast obtained by BSE energy filtering is about twice that of the standard BSE method by varying the acceleration voltage. The contrast decreases with increasing information depth. The information depth is about four times smaller than the electron range. The transmission of the spectrometer influences the minimum current of the order of 10−8 A that is needed to get a contrast of 1%, for example.

A locally enhanced element concentration influences the result of an X-ray microanalysis at a neighbouring position. This influence was investigated for the first time systematically in organic (biological) material using sections of epoxy resin (thickness 0.5–2.5 μm) containing a layer of pure gold. Wavelength and energy dispersive spectrometers were applied to analyse the X-rays generated by 15–35 keV electrons. Characteristic X-rays could be detected up to distances of several μm from the gold layer. For example, for a 2.4 μm thick section and 35 keV electrons the measured apparent gold concentration was above 0.1% (weight% per dry mass) at a distance of 10 μm. Thus, the lateral resolution may be not better than a multiple of the section thickness. The apparent gold concentration at a given distance is proportional to the specimen thickness and increases with increasing electron energy. Monte Carlo simulations confirm the experimental results. The influence of a local enrichment depends on the particular specimen properties (e.g. thickness, density, mean atomic number), the electron energy, and the geometry of the detector with respect to the specimen.

A system of two opposite Everhart–Thornley detectors A and B has formerly been applied in conventional SEM for electron energies between 5 and 20 keV to separate material, topographic and other types of contrast by sum and difference signals. This technique can also be used successfully for low-voltage scanning electron microscopy. The decreasing information depth with decreasing electron energy shows differences in the surface composition and contamination which cannot be observed beyond 5 keV. Also below 5 keV material and topographic contrast can be separated and increased by the A + B and A − B signals, respectively.

It is well known that the differential Mott cross section for large-angle elastic scattering shows maxima and minima at angles depending on material and electron energy. For electron energies of 10–30 keV, the averaging by frequent elastic scattering processes results in approximate Lambert angular distributions of backscattered electrons (BSE). However, the present Monte Carlo calculations for electron energies E = 1–5 keV and different angles of incidence show strong deviations from a Lambert distribution which increases with decreasing energy. The signals of the BSE detector with five annular segments for different take-off directions show good agreement with the calculations for normal electron incidence.

Summary Complex scattering amplitudes are used to calculate the phase contrast of colloidal gold particles. Comparison of measurements of the phase contrast intensity at the centre of the gold particle as a function of defocus for unfiltered and zero-loss filtered images demonstrates the increase in phase contrast achieved by zero-loss filtering even for a thick carbon substrate film. The granulation of amorphous germanium films is measured by the spatial rms (root mean square) values of image intensity in a defocus series.

Summary The contrast thicknesses (xk) of thin carbon and platinum films have been measured in the transmission mode of a low-voltage scanning electron microscope for apertures of 40 and 100 mrad and electron energies (E) between 1 and 30 keV. The measured values overlap with those previously measured for E (≥ 17keV) in a transmission electron microscope. Differences in the decrease of xk with decreasing E between carbon and platinum agree with Wentzel-Kramer-Brillouin calculations of the elastic cross-sections. Knowing the value of xk allows the exponential decrease ∝ exp(—x/xk) in transmission with increasing mass-thickness (x = ρt) of the specimen and the increasing gain of contrast for stained biological sections with decreasing electron energy to be calculated for brightfield and darkfield modes.

The contrast of particles of colloidal gold in the backscattered electron mode of high-resolution low-voltage scanning electron microscopy is of interest for the use as markers in surface labeling techniques of electron microscopical histochemistry. Monte Carlo calculations show a contrast maximum when the particle diameter is comparable with the electron range in gold, and the electron energy showing the maximum decreases with decreasing diameter. The calculations are confirmed experimentally by measuring the contrast of 40 nm particles of colloidal gold and evaporated gold films on a silicon substrate for comparison.

The imaging of ferromagnetic domains by the transverse Kerr effect using a scanning optical microscope (SOM) with two or four directions of light incidence (± x, ± y) allows the digital record of the By and Bx components of the magnetic induction B. These components can be used to calculate a two-dimensional colour-coded map of B, which can be superposed by vector arrows.

Measurements of the electron range R, and the backscattering coefficient η and the secondary electron yield δ at normal and tilted incidence for different elements show characteristic differences for electron energies in the range of 0.5 to 5 keV, compared with energies larger than 5 keV. The backscattering coefficient does not increase monotonically with increasing atomic number; for example, the secondary electron yield shows a lesser increase with increasing tilt angle. This can be confirmed in back-scattered electron (BSE) and secondary electron (SE) micrographs of test specimens. The results are in rather good agreement with Monte Carlo simulations using elastic Mott cross-sections and a continuous-slowing-down model with a Rao Sahib-Wittry approach for the stopping power at low electron energies. Therefore, this method can be used to calculate quantities of BSE and SE emission, which need a larger experimental effort. Calculations of the angular distribution of BSEs show an increasing intensity with increasing atomic number at high takeoff angles than expected from a cosine law that describes the angular characteristics at high electron energies. When simulating the energy distribution of BSEs, the continuous-slowing-down model should be substituted by using an electron energy-loss spectrum (EELS) that considers plasmon losses and inner-shell ionizations individually (single-scattering-function model). The EELS can be approached via the theory for aluminium or from EELS spectra recorded in a transmission electron microscope for other elements. Measurements of electron range Rα En of 1 to 10 keV electrons are obtained from transmission experiments with thin films of known mass thickness. In agreement with other authors the exponent n is lower than at higher electron energies.

Energy-filtering transmission electron microscopy (EFTEM) can be realized either with a dedicated STEM or with an imaging energy filter in a TEM. The conventional STEM and TEM modes can be combined in this way with the modes of electron spectroscopic imaging (ESI) and electron spectroscopic diffraction (ESD), and different modes can be used to record an electron energy-loss spectrum (EELS). An EFTEM can therefore make full use of the elastic and inelastic electron-specimen interactions. This review summarizes the possibilities of EFTEM for the investigation of biological specimens.

SUMMARY The electron spectroscopic imaging (ESI), diffraction (ESD) and different types of electron energy-loss spectroscopy (EELS) modes in an energy-filtering transmission electron microscope can all be used for the investigation and analytical use of plasmon losses. Shifts of plasmon losses caused by differences in composition can be detected with an accuracy of 0.1 eV by parallel-recorded EELS (PEELS). The dispersion of plasmon losses and the cut-off angle θc can be observed by angle-dispersive EELS and by recording spectra at different scattering angles θ. ESD patterns with a selected energy window width of 1 eV enable the dispersion and its anisotropy to be imaged by characteristic intensity distributions between the primary beam and the first Bragg diffracted beams. The ESI mode can be used for the selective imaging of precipitates and for the investigation of the excitation volume of plasmons in small particles.

When the zero-loss transmission falls below 10(-3) for biological sections of mass-thickness greater than 70 micrograms/cm2, the energy window in the electron spectroscopic imaging (ESI) mode of an energy-filtering electron microscope (EFEM) can be shifted to the most-probable energy loss of the electron energy-loss spectrum. This enables mass-thicknesses up to 150 micrograms/cm2 or thicknesses of 1.5 microns to be examined. Electron energy-loss spectra of thick carbon films calculated by a Fourier method agree with experimental spectra. Measurements of the electron energy-loss spectroscopy and ESI image intensities with an additional platinum film confirm a scattering model for the calculation of the image intensity. This model considers the angular broadening at the most-probable energy loss by introducing an effective illumination aperture of the order of the full-width at half-maximum of the angular distribution.

The most effective detection systems for secondary electrons (SE), which has a low noise level and a large bandwidth, is the Everhart-Thornley detector. Electrons are collected by a positively biased grid in front of a scintillator biased at +10 kV. The light emission is recorded by a photomultiplier tube. Scintillation detectors can also be used for backscattered electrons (BSE) when the solid angle of collection is increased. Other alternatives for BSE are semiconductor detectors, microchannel plates, or the conversion of BSE to SE. The recording of SE in the in-lens mode or in LVSEM needs special detector designs for rotationally symmetric collection fields. For the separation of different types of contrast and for more quantitative signals, improvements can be obtained by employing multiple detector systems.

Backscattered electrons (BSE) and secondary electrons (SE) are the most important signals for image recording. A knowledge of the dependence of the backscattering coefficient η and the secondary electron yield δ on surface tilt, material and electron energy and their angular and energy distributions is essential for the interpretation of image contrast. The spatial exit distributions and information depths of these electrons govern the resolution if the latter is not limited by the electron-probe size. The shot noise introduced by the incident electron current will be increased during SE and BSE emission.

The most important topographic contrast mode with secondary electrons is a consequence of the dependence of the SE yield on the local tilt of the specimen surface. A fraction of the SE signal is excited by the primary electron probe and carries high-resolution information since the exit depth of the SE is small. Another fraction of poorer resolution is excited by the BSE. If, instead of using the conventional SE detector, the SE are sorted according to their exit momenta, a more quantitative interpretation of the topography may be possible.

Magnetic stray fields in front of the specimen can be detected by the deflection of secondary electrons by the Lorentz force (magnetic contrast type—1) or by the deflection of primary electrons incident parallel to the surface. The Lorentz force due the internal magnetization affect the trajectories of backscattered electrons, resulting in type—2 magnetic contrast. The surface magnetization can be analysed by detecting the spin-polarization of SE.

Electron beams generate electron—hole pairs or minority carriers in semiconductors within a small volume. They are therefore excellent tools for measuring semiconductor-device parameters such as the diffusion length, the surface recombination velocity, the relaxation time, and the position and width of depletion layers by recording the charge-collection current or electron-beaminduced current in a depletion layer. Schottky barriers as well as diffused and ion-implanted p-n junctions can be studied. By modulating the CRT with the charge-collection current, images of depletion layers and of crystal defects, which influence the recombination of minority carriers, can be displayed.

The deceleration of fast electrons in the Coulomb field of a nucleus can result in the emission of a background x-ray quantum. The de-excitation of an inner-shell ionization results either in the emission of a characteristic x-ray quantum or in the emission of an Auger electron, both of which can be used for elemental analysis.

Electron diffraction methods are employed for the identification of substances by measuring the lattice-plane spacings and for the determination of crystal orientations in polycrystalline films (texture) or single-crystal foils. Extra spots and streaks, caused by antiphase structures or plate-like precipitates, for example, may also be observed when imaging a selected area.

The inner-shell ionization of atoms results in an emission of characteristic x-ray quanta or Auger electrons. A wavelength- or an energy-dispersive x-ray spectrometer can be coupled to a transmission electron microscope to record x-ray quanta emitted from the specimen. The quantitative methods developed for the x-ray microanalysis of bulk materials can be transferred to the investigation of thin specimens.

The theoretical treatment of electron diffraction at crystals needs the concepts of lattice planes and the reciprocal lattice, as in x-ray diffraction. Kinematical theory leads to the Bragg condition and to a description of the influence of the structure of a unit cell and of the external size of a crystal on the diffracted amplitude in terms of structure and lattice amplitudes, respectively. The observed diffraction pattern is equivalent to the points of intersection of the Ewald sphere of radius 1/λ with the reciprocal-lattice nodes.

The elastic scattering of electrons by the Coulomb potential of a nucleus is the most important of the interactions that contribute to the image contrast. Cross-sections and mean-free-path lengths are used to describe the scattering process quantitatively. A knowledge of the screening of the Coulomb potential of the nuclei by the atomic electrons is important when calculating the cross-sections at small scattering angles.

A crystal can be imaged with the primary beam (bright field) or with a Bragg reflection (dark-field). The local intensity depends on the thickness, resulting in thickness (or edge) contours, and on the tilt of the lattice planes, resulting in bend contours, which can be described by the dynamical theory of electron diffraction. In certain cases, the intensity of a Bragg reflection depends so sensitively on specimen thickness that atomic surface steps can be observed. The most important application of diffraction (Bragg) contrast is the imaging of lattice defects such as dislocations, stacking faults, phase boundaries, precipitates and defect clusters. The contrast depends on the Bragg reflection excited and its excitation error, the type of the fault and its depth inside the foil. The Burgers vector of a dislocation or the displacement vector of a boundary can thus be determined quantitatively. The resolution of the order of 10 nm when a strongly excited Bragg reflection is used can be reduced to the order of one nanometre by the weak-beam technique, which allows us to measure the width of dissociated dislocations, for example. Different types of contrast for precipitates are associated with coherent and incoherent precipitates, which can hence be distinguished. Electron spectroscopic imaging can remove the inelastically scattered electrons in the background of a diffraction pattern and increase the contrast and resolution of defect images.

Monte Carlo simulations of electron diffusion become of increasing interest for scanning electron microscopy (SEM), X-ray microanalysis (XRMA) and Auger electron spectroscopy (AES) due to the increasing speed and storage capability of modern PCs. Depth distribution functions can be calculated in less than a minute also for complex specimen structures. To apply Monte Carlo simulations in the energy range of 0.1–50 keV it is necessary to use a data base of Mott elastic cross-sections calculated by the partial-wave method. For most applications it is sufficient to consider inelastic scattering by the Bethe continous-slowing-down approximation and inner-shell ionisations with energy losses larger than 100–200 eV by the Gryzinski cross-section. In future, energy-loss functions obtained by a Kramers-Kronig analysis of experimental electron energy-loss spectra (EELS) will become of interest for a better consideration of straggling effects during the slowing-down.

A FORTRAN 77 program package has been developed for the Monte Carlo simulation of electron trajectories and of secondary electrons (SE), backscattered electrons (BSE) and X-ray linescans for scanning electron microscopy and microanalysis. A data base contains the relativistic elastic Mott crosssections of all elements, the X-ray absorption coefficients, and the most relevant ionisation and characteristic X-ray energies.

Whereas an image in a conventional transmission electron microscope (CTEM) contains contributions from elastically and inelastically scattered electrons that pass the objective diaphragm, the electron spectroscopic imaging (ESI) mode of an EFTEM allows us to separate the contributions by inserting an energy-selecting slit in the energy-dispersive plane of a filter lens or a prism spectrometer; energy-loss windows E ± δ/2 with a width δ= 1 - 20 eV can be selected (see Sect. 1.4.2). The choice between zero-loss filtering and the use of an energy-loss window in the loss-spectrum of inelastically scattered electrons depends on the specimen and the information wanted. A schematical energy-loss spectrum in Fig. 7.1 shall demonstrate the ESI modes [7.1–5] which are discussed in detail in this Chapter.

In principle, electron diffraction studies in a transmission electron microscope (TEM) provide much the same crystallographic information as does X-ray or neutron diffraction. Compared to the latter two, two important problems arise in electron diffraction owing to the much stronger interaction of electrons with matter: 1) The kinematical approximation is valid only for very thin films of 1–5 nm and 2) Inelastic scattering processes contribute to a diffuse background. For thicker specimens this background dominates the entire diffraction pattern.

The earliest crystallites in dentine appear as chains of "dots" in ultra-thin sections viewed by transmission electron microscopy. These dots rapidly coalesce along the longitudinal directions of the collagen microfibrils to form needle-like structures that coalesce preferentially in lateral directions to form ribbon-like or plate-like crystallites. This morphological interpretation is supported by line-scans of the corresponding zero-loss filtered electron spectroscopic diffraction patterns, which demonstrate the crystalline structure of the dentine mineral (apatite). The intensity ratio of the Debye-Scherrer rings of the characteristic Bragg-reflections (002 to 300, together with 1 or 2 unresolved reflections) shows a maximum in the region of early chain-like and needle-like crystallites, decreasing with maturation of the dentine mineral to the ribbon-plate-like crystallites. Detailed investigations using line-scans of the zero-loss filtered electron spectroscopic diffraction patterns through the dentine zone show that the intensity ratio found near the mineralisation front is repeated 3-5 times at distances of about 10-20 microns. This may represent a circadian pattern of mineralisation corresponding to light microscopically visible incremental lines in dentine.

The rapid progress in LVSEM instrumentation is based on the development of Schottky emission and field-emission guns rather than the thermionic cathodes used in conventional SEM. The smallest electron probe size is limited mainly by the chromatic and spherical lens aberrations, and electron probe diameters of 1-2 nm at 20-30 keV and about 10-20 nm at 0.5-1 keV are attainable. The most efficient detector systems are scintillator-photomultiplier combinations, and semiconductor and microchannel-plate detectors. The improvement of electron optics using in-lens operation and through-lens detection of emitted electrons depends on new detector strategies.

Charging of insulators can result in a strong negative charging for high electron energies, resulting in charging artifacts. One of the advantages of LVSEM is weak positive charging for electron energies below a critical energy E2. The inelastic scattering processes result in heat generation and ionization processes that cause specimen damage, contamination, and exposure of resists in electronbeam lithography.

The backscattering coefficient η and the secondary-electron yield δ depend on electron energy E, atomic number Z, and surface tilt angle ϕ, and show characteristic energy and angular distributions that are important to the discussion of the image formation when using the signal of backscattered and secondary electrons.