International
Tables for Crystallography Volume C Mathematical, physical and chemical tables Edited by E. Prince © International Union of Crystallography 2006 |
International Tables for Crystallography (2006). Vol. C. ch. 4.3, pp. 259-429
https://doi.org/10.1107/97809553602060000593 Chapter 4.3. Electron diffraction
C. Colliex,a J. M. Cowley,b‡ S. L. Dudarev,c M. Fink,d J. Gjønnes,e R. Hilderbrandt,f A. Howie,g D. F. Lynch,h L. M. Peng,i G. Ren,j A. W. Ross,d V. H. Smith Jr,k J. C. H. Spence,l J. W. Steeds,m J. Wang,k M. J. Whelanc and B. B. Zvyaginn‡
a Laboratoire Aimé Cotton, CNRS, Campus d'Orsay, Bâtiment 505, F-91405 Orsay CEDEX, France,bDepartment of Physics and Astronomy, Arizona State University, Tempe, AZ 85287-1504, USA,cDepartment of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, England,dDepartment of Physics, The University of Texas at Austin, Austin, TX 78712, USA,eDepartment of Physics, University of Oslo, PO Box 1048, Blindern, N-0316 Oslo, Norway,fChemistry Division, Room 1055, The National Science Foundation, 4201 Wilson Blvd, Arlington, VA 22230, USA,gCavendish Laboratory, Madingley Road, Cambridge CB3 0HE, England,hCSIRO Division of Materials Science & Technology, Private Bag 33, Rosebank MDC, Clayton, Victoria 3169, Australia,iDepartment of Electronics, Peking University, Beijing 100817, People's Republic of China,jBeijing Laboratory of Electron Microscopy, Chinese Academy of Sciences, PO Box 2724, Beijing 100080, People's Republic of China,kDepartment of Chemistry, Queen's University, Kingston, Ontario, Canada K7L 3N6,lDepartment of Physics, Arizona State University, Tempe, AZ 85287, USA,mH. H. Wills Physics Laboratory, University of Bristol, Tyndall Avenue, Bristol BS8 1TL, England, and nInstitute of Ore Mineralogy, Akad. Nauk Russia, Staromonetny 35, 109017 Moscow, Russia The first section of this chapter concerns scattering factors for the diffraction of electrons by crystalline solids. An explanation of the theory of scattering by a perfect crystal is followed by a discussion of the kinematical, two-beam and phase-grating approximations. Relativistic and absorption effects are considered. Extensive tables of atomic scattering amplitudes for electrons for neutral and ionized atoms are presented. The second section of the chapter briefly discusses the parameterization of electron atomic scattering factors. Tables of useful parameters as a function of accelerating voltage and elastic atomic scattering factors for neutral atoms are given. Complex scattering factors for the diffraction of electrons by gases are discussed in the third section of the chapter. This section includes tables of scattering factors of interest for gas-phase electron diffraction from atoms and molecules in the keV energy region. In addition to the tables and a description of their uses, a discussion of the theoretical uncertainties related to the material in the tables is also provided. The tables give scattering factors for elastic and inelastic scattering from free atoms. The theory of molecular scattering based on these atomic quantities is also discussed. Electron energy-loss spectroscopy on solids is discussed in the fourth section of the chapter. Topics covered include: the use of electron beams; single and multiple scattering; the classification of the different excitations in a spectrum; instrumentation; and the excitation spectra of valence and core electrons. The fifth section of the chapter describes oriented texture patterns. Lamellar and fibre texture patterns are discussed and applications to metals and organic materials are mentioned. The computation of dynamic wave amplitudes in then described in the sixth section of the chapter. This section deals first with the multislice method. The numerical procedure is outlined and factors that influence the choice of thickness of the slice are discussed. Two checks that can be performed during a multislice calculation are noted. The Bloch-wave method is then described. The use of Bloch waves to describe electron diffraction and electron imaging in thin crystals is outlined together with the concept of the dispersion surface. These emerge as natural solutions of the Schrödinger equation with a periodic optical potential to generate the elastic scattering and also the loss of intensity from the coherent wave field due to thermal diffuse and inelastic scattering. The Bloch-wave approach is a useful complement to the multislice method and provides a clear picture of wave propagation in perfect and imperfect crystals. In the seventh section of the chapter, the measurement of structure factors and the determination of crystal thickness by electron diffraction are described. The use of convergent-beam electron diffraction to obtain integrated intensities is discussed and the relationship between intensity features and the dispersion surface is explained. The last section of the chapter concerns crystal-structure determination by high-resolution microscopy. This technique allows the arrangement of atomic columns in thin crystals to be observed directly. The resolution of the best instruments is now slightly below 0.1 nm. The images usually show a projection through a slice of crystal about 20 nm thick, however tomographic (three-dimensional) reconstruction is now possible at nanometre resolution. The images show the host of microphases, grain boundaries, twins, line and planar defects which broad-beam methods, such as X-ray diffraction, provide the average scattering from. These defects often control the properties of crystals, engineering materials and electronic devices. Individual nanostructures, such as carbon nanotubes and catalyst particles, may be imaged at atomic resolution. Fine twinning, polytypes, intergrowth of oxide phases etc. can be identified, and increasingly the detailed atomic structure of defects (such as oxide, superconductor and semiconductor interfaces) is being determined. Substitutional dopant atoms have recently been imaged for the first time. In biology the method is limited by radiation damage; however by summing many images of identical randomly oriented macromolecules, tomographic density maps can be reconstructed at subnanometre resolution from hydrated proteins which cannot be crystallized (e.g. membrane proteins). This section reviews the theoretical principles of high-resolution electron microscopy, including few-beam and structure image formation, effects of electron-optical lens aberrations, partial coherence, resolution-limiting factors, image-simulation methods, dynamical effects, and a summary of super-resolution schemes. Keywords: absorption; atomic scattering amplitudes; atomic scattering factors; Bethe theory; Bloch-wave method; core-electron spectroscopy; core-loss spectroscopy; crystal thickness; crystalline solids; dielectric description; dynamical diffraction; dynamical wave amplitudes; EELS; elastic scattering; electron diffraction; electron energy-loss spectroscopy; electron microscopy; electrons; free-electron gas; high-resolution electron microscopy; HREM; hyper-resolution; inelastic scattering; lamellar textures; lattice-fringe images; molecular scattering factors; multiple scattering; multislice method; oriented texture patterns; plasmons; real solids; relativistic effects; STEM; scanning transmission electron microscopes; scattering factors; solid-state effects; spectrometers; structure factors; surface plasmons; texture; wave amplitudes; X-ray diffraction. |
The most important interaction of electrons with crystalline matter is the interaction with the electrostatic potential field. The scattering into sharp, Bragg reflections is considered in terms of the interaction of an incident plane wave with a time-independent, averaged, periodic potential field which may be written where is the unit-cell volume and the Fourier coefficients, , may be referred to as the structure amplitudes corresponding to the reciprocal-lattice vectors h. In conformity with the crystallographic sign convention used throughout this volume [see also Volume B (IT B, 2001)], we choose a free-electron approximation for the incident electron beam of the form and the interaction is represented by inserting the potential (4.3.1.1) in the Schrödinger wave equation where eE is the kinetic energy of the incident beam, is the magnitude of the wavevector for the incident electrons, and σ is an `interaction constant' defined by where h is Planck's constant. Relativistic values of m and λ are assumed (see Subsection 4.3.1.4).
The solution of equation (4.3.1.2), subject to the boundary conditions imposed by the need to fit the waves in the crystal with the incoming and outgoing waves in vacuum at the crystal surfaces, then allows the directions and amplitudes of the diffracted beams to be obtained in terms of the crystal periodicities and the Fourier coefficients, , of by the eigenvalue or Bloch-wave method (Bethe, 1928).
Alternatively, the scattered amplitudes may be obtained from the integral form of (4.3.1.2), where represents the incident beam, K = σ/λ, and the integral is taken over the space of the variable, . An iterative solution of (4.3.1.4) leads to the Born series, where and for . Terms of the series for may be considered to represent the contributions from single, double and multiple scattering of the incident electron beam. This method has been applied to the diffraction from crystals by Fujiwara (1959).
A further formulation of the scattering problem in integral form is that due to Cowley & Moodie (1957) who considered the progressive modification of an incident plane wave as it passed through successive thin slices of a crystal. The effect of the nth slice on the incident electron wave is that of a phase-grating so that the wavefunction is modified by multiplication by a transmission function, where is the projection of the potential distribution within the slice in the direction of the incident beam, taken to be the z axis; Propagation of the wave between the centres of slices is represented by convolution with a propagation function, p(xy), so that the wave entering the (n + 1)th slice may be written In the small-angle approximation, the function is given by the usual Fresnel diffraction theory as
In the limit that the slice thickness, , tends to zero, the iteration of (4.3.1.8) gives an exact account of the diffraction by the crystal.
On the basis of the above-mentioned and other related formulations of the diffraction problem, several computing methods have been devised for calculation of the amplitudes and intensities of the many diffracted beams of appreciable intensity that may emerge simultaneously from a crystal (see Section 4.3.6). In this way, a degree of accuracy may be achieved in the calculation of the intensities of spots in diffraction patterns or of the contrast in electron-microscope images of crystals (Section 4.3.8).
All such calculations require a knowledge of the potential distribution, , or its Fourier coefficients, . It is usually convenient to express the potential distribution in terms of the sum of contributions of individual atoms centred at the positions . Thus: or, in terms of the Fourier transforms, , of the
As a first approximation, the functions may be identified with the potential distributions for individual, isolated atoms or ions, with the usual spreading due to thermal motion. The interatomic binding and the interactions of ions that are thereby neglected may have important effects on diffraction intensities in some cases.
In this approximation, the Fourier transforms for individual atoms may be written where and the are the Born electron scattering amplitudes, as conventionally defined, in units of Å. Here is half the scattering angle and, again, K = σ/λ. Some values of listed in the accompanying Tables 4.3.1.1 and 4.3.1.2 are obtained from the atomic potentials for isolated, spherically symmetrical atoms or ions by the relation
|
|
By the use of Poisson's equation relating the potential and charge-density distributions, it is possible to derive the Mott–Bethe formula for in terms of the atomic scattering factors for X-rays, : where is the permittivity of vacuum, or [for λ in Å, in Å, and in electron units]. This was used for the other listed values.
It has been shown by Fujiwara (1961) that, at least for electron energies up to 1 MeV or so, the relativistic effects on diffraction amplitudes and geometry are adequately described by the use of relativisitically corrected values for the mass and wavelength of the electrons; where is the rest mass, is the Compton wavelength , and λ is given in Å if E is in volts. Consequently, σ varies with the incident electron energy as , or Values of λ, β = ν/c, and σ are listed for various values of the accelerating voltage, E, in Table 4.3.2.1 with λ in Å and E in volts.
|
Any scattering process, whether elastic or inelastic, which removes energy from the set of diffracted beams being considered, may be said to constitute an absorption process. For example, for a measurement of the intensities of the elastically scattered, sharp Bragg reflections from a crystal, any process which gives diffuse background scattering or results in a detectable loss of energy gives rise to absorption.
The diffracted amplitudes in such cases may be calculated, at least as a first approximation, in terms of a complex potential, , containing an imaginary part due to an `absorption function' and a small added real part . Then under the crystallographic sign convention, . Correspondingly, for a centrosymmetric crystal, the structure amplitude becomes complex and may be written Under the appropriate conditions of observation, important contributions to the imaginary and real additions to the structure amplitudes may be given by the excitation of phonons, plasmons, or electron transitions, or by diffuse scattering due to crystal defects or disorder.
The additional terms and , however, are not invariant properties of the crystal structure but depend on the conditions of the diffraction experiment, such as the accelerating voltage and orientation of the incident beam, the aperture or resolution of the recording system, and the use of energy filtering or discrimination. In spite of this, it may often be convenient to treat them as being produced by phenomenological complex potentials, defined for a limited range of experimental conditions.
Tables 4.3.1.1 and 4.3.1.2 list values of in Å for all neutral atoms and most chemically significant ions, respectively. The values have been given by Doyle & Turner (1968) for several cases, denoted by RHF using the relativistic Hartree–Fock atomic potentials of Coulthard (1967). For all other atoms and ions, has been found using the Mott–Bethe formula [equation (4.3.1.15)] for , and the X-ray scattering factors of Table 2.2A of IT IV (1974). Thus all other neutral atoms except hydrogen are based on the relativistic Hartree–Fock wavefunctions of Mann (1968). These are designated by *RHF. For H and for ions below Rb, denoted by HF, is ultimately based on the nonrelativistic Hartree–Fock wavefunctions of Mann (1968). For ions above Rb, denoted by *DS, modified relativistic Dirac–Slater wavefunctions calculated by Cromer & Waber (1974) are used.
For low values of s, the Mott formula becomes less accurate, since tends to zero with s for neutral atoms. Except for the RHF atoms, for s from 0.01 to 0.03 are omitted in Table 4.3.1.1 and for s from 0.04 to 0.11, only two decimal places are given. is then accurate to the figure quoted. For these atoms, was found using the formula given by Ibers (1958): where is the mean-square atomic radius.
For ionized atoms, . The values listed at s = 0 in Table 4.3.1.2 for RHF atoms were calculated by Doyle & Turner (1968) with in equation (4.3.1.13) replaced by , where Here, is the ionic charge. This approach omits the Coulomb field due to the excess or deficiency of charge on the nucleus. With the use of these values, the structure factor for forward scattering by a neutral unit cell containing ions may be found in the conventional way. Similar values are not available for other ions because the atomic potential data are lacking.
For computer applications, numerical approximations to the f(s) of these tables have been given by Doyle & Turner (1968) as sums of Gaussians for the range s = 0 to 2 Å−1. An alternative is to make Gaussian fits to X-ray scattering factors, then use the Mott formula to derive electron scattering factors. As discussed by Peng & Cowley (1988), this practice may lead to problems for small values of s. An additional problem occurs in high-resolution electron-microscopy (HREM) image-simulation programs, where it is usually necessary to have electron scattering factors for the range 0 to 6 Å−1. Fox, O'Keefe & Tabbernor (1989) point out that extrapolation of the Gaussian fits of Doyle & Turner (1968) to values past 2 Å−1 can be highly inaccurate. For the range of s from 2 to 6 Å−1, Fox et al. have used sums of polynomials to make accurate fits to the X-ray scattering factors of Doyle & Turner (1968) for many elements (Section 6.1.1 ), and electron scattering factors can be generated from these data by use of the Mott formula.
Recently, Rez, Rez & Grant (1994) have published new tables of X-ray scattering factors obtained using a multiconfiguration Dirac–Fock code and two parameterizations in terms of four Gaussians, one of higher accuracy over the range of about 2 Å−1 and the other of lower accuracy over the extended range of about 6 Å−1. These authors suggest that electron scattering factors may best be obtained from these X-ray scattering factors by using the Mott formula. They provide a table of values for the electron scattering factor values for zero scattering angle, , for many elements and ions, which may be of value for the calculation of mean inner potentials.
In order to calculate the Fourier coefficients V(h) of the potential distribution , for insertion in the formulae used to calculate intensities [such as (4.3.1.6), (4.3.1.20), (4.3.1.21)], or in the numerical methods for dynamical diffraction calculations, use where The values are obtained from Tables 4.3.1.1 and 4.3.1.2, and is the unit-cell volume in Å3. The V(h) and the tabulated are properties of the crystal structure and the isolated atoms, respectively, and are independent of the particular scattering theory assumed.
Expressions for the calculation of intensities in the kinematical approximation are given for powder patterns and oblique texture patterns in Section 2.5.4 , and for thin crystal plates in Section 2.5.2 of Volume B (IT B, 2001). Since the formulas for kinematical scattering, such as (4.3.1.19) and (4.3.1.20), include the parameter K = σ /λ, which varies with the energy of the electron beam through relativistic effects, it may be considered that the electron scattering factors for kinematical calculations should be multiplied by relativistic factors.
For high-energy electrons, the relativistic variations of the electron mass, the electron wavelength and the interaction constant, σ, become significant. The relations are where is the rest mass, is the Compton wavelength, , and . Consequently, varies with the incident electron energy as
For the calculation of intensities in the kinematical approximation, the values of listed in Tables 4.3.1.1 and 4.3.1.2, which were calculated using , must be multiplied by for electrons of velocity v. Values of λ, 1/λ, , β = v/c, and σ are listed for various values of the accelerating voltage, E, in Table 4.3.2.1.
For computer applications, numerical approximations to the f(s) of Tables 4.3.1.1 or 4.3.1.2 are usually preferred and various approximations as sums of Gaussians have been proposed. The initial Gaussian fits were given by Doyle & Turner (1968) for the range s = 0 to 2 Å−1. However, for some purposes, as in the image-simulation programs for high-resolution electron microscopy, atomic scattering factors are needed for higher s values, up to 6 Å−1, and, as pointed out by Fox, O'Keefe & Tabbernor (1989), extrapolation of the Gaussian fits of Doyle & Turner to values above 2 Å−1 can be highly inaccurate.
An alternative approach to obtaining numerical values for the electron scattering factors is to make use of the polynomial fits to X-ray scattering factors of Fox et al. or the more recent tables of X-ray scattering factors produced by Rez, Rez & Grant (1994), who used a multiconfiguration Dirac–Fock code and two parameterizations in terms of four Gaussians, one of higher accuracy over the range of about 2 Å−1 and the other of lower accuracy over the extended range of about 6 Å−1. The electron scattering factors may then be derived from the X-ray scattering factors by use of the Mott formula (4.3.1.14). For small angles of scattering, the determination of electron scattering factors in this way may give problems, since the X-ray scattering factor tends to the atomic number, and both the numerator and denominator of (4.3.1.14) tend to zero. However, the electron scattering factor may be determined for zero scattering angle using equation (4.3.1.29) and Rez, Rez & Grant (1994) listed values of for many elements and ions.
Recently, Peng, Ren, Dudarev & Whelan (1996) have developed a new algorithm, based on a combined modified simulated-annealing and least-squares method, to parameterize both the elastic and absorptive scattering factors as sums of five Gaussians of the form where and are fitting parameters. The values of their fitting parameters for the range of s values from 0 to 2.0 for elastic electron scattering factors for all neutral atoms with atomic numbers up to 98 are given in Table 4.3.2.2 and the values obtained separately for these atoms for the range of s from 0 to 6.0 Å−1 are given in Table 4.3.2.3. For Table 4.3.2.2, the fitting was made to the values of f given in Table 4.3.1.1. For Table 4.3.2.3, the f values in the range of s from 2.0 to 6.0 Å−1 were those obtained by using the Mott formula to convert the X-ray scattering factors derived from the Dirac–Fock calculations of Rez, Rez & Grant (1994). Similar tables for atomic scattering factors of ions can be found in Peng (1998).
|
|
As an indication of the accuracy with which the parameterized f values of (4.3.2.1) reproduce the numerical values of the reference f values, Peng et al. (1996) computed values of , where σ is the square root of the mean square deviation, σ2, between the numerical and fitted scattering factors. The values of are typically in the range 0.02 to 0.05, and are consistently smaller (with a few exceptions) than the corresponding values given for the parameterizations of previous workers (Weickenmeier & Kohl, 1991; Bird & King, 1990; Doyle & Turner, 1968).
For the absorptive scattering factors, corresponding to the imaginary parts added to the real elastic scattering factors as a consequence of inelastic scattering processes, Peng et al. (1996) have tabulated values for particular elemental crystals and a selection of crystals of compounds having the zinc-blend structure. The main contribution to the absorptive scattering factors arises from the thermal vibrations of the atoms in the crystals so that the numerical values are not characteristic of the individual atom types but depend on the type of bonding of the atoms in the crystal, as indicated by the Debye–Waller factor, and must be calculated separately for each temperature. The authors offer copies of their computer programs, freely available via electronic mail, from which the parameterization of the absorptive scattering factors can be derived for other materials and temperatures, given the values of the atomic numbers of the elements, the Debye–Waller factor and the electron accelerating voltage.
This section includes tables of scattering factors of interest for gas-phase electron diffraction from atoms and molecules in the keV energy region. In addition to the tables and a description of their uses, a discussion of the theoretical uncertainties related to the material in the tables is also provided. The tables give scattering factors for elastic and inelastic scattering from free atoms. The theory of molecular scattering based on these atomic quantities is also discussed.
It has long been known that the first Born approximation provides an inadequate description at the 4% accuracy level for elastic and total differential cross sections in the 40 keV energy range for atoms heavier than Ne (Schomaker & Glauber, 1952; Glauber & Schomaker, 1953). Results of early experimental work have been confirmed for both atomic and molecular scattering (Kimura, Schomaker, Smith & Weinstock, 1968; Bartell & Brockway, 1953; Hanson, 1962; Fink & Kessler, 1966; Geiger, 1964; Kessler, 1959; Seip, 1965; Schäfer & Seip, 1967; Kohl & Bonham, 1967; Bonham & Cox, 1967; Seip & Stølevik, 1966a,b; Seip & Seip, 1966; Arnesen & Seip, 1966; McClelland & Fink, 1985; Coffman, Fink & Wellenstein, 1985). New partial wave scattering factors based on relativistic Hartree–Fock fields (Biggs, Mendelsohn & Mann, 1975) are presented here at a number of energies (Table 4.3.3.1). Because of the availability of these partial wave results, first Born approximation results are no longer needed for gas-phase work in this energy range.
|
The scattering of keV electrons from atoms is calculated in the central-field approximation in which the potential of the target is averaged over the angular coordinates and the resulting spherically symmetric potential V(r) is used in the computation. In order to take the relativistic effects properly into account, the Dirac equation has been used. In addition to the straightforward correction for the electron mass, spin-polarization effects are also included in these calculations. The scattering wavefunctions are four-component spinors that can be reduced to two components as shown by Mott & Massey (1965). This reduction leads to two decoupled second-order differential equations in Schrödinger form: where and where j is the total angular momentum for the lth partial wave including the two spin directions. Asymptotic solutions are available when is small relative to the centrifugal term, . If this term is taken into account, but is neglected, approaches with a similar limit holding for the solution. These limits lead directly to the scattering factors and the elastic differential scattering cross section where A, B and describe the direction and degree of spin polarization of the incoming electrons. The latter term is equal to 0 when unpolarized electrons (A = B = 1) are used in the scattering experiment.
The results printed in the tables were obtained in three steps. First, atomic wavefunctions were calculated and transformed into centrosymmetric potentials via Poisson's equation. Second, a sufficient number of phases, and , were computed in order to calculate the scattering factors f and g by performing the partial wave sums. Finally, the results were smoothed, because small oscillations were seen between nearest neighbours in the second difference function. These oscillations were only of the order of 0.1% of the data, so smoothing only had an effect in the third or fourth significant figure.
For the scattering potentials, we used relativistic Hartree–Fock wavefunctions calculated by Biggs, Mendelsohn & Mann (1975). The wavefunctions were used to calculate the potentials and their derivatives since they are needed for to solve the appropriate Dirac equation.
In order to solve the second-order differential equation, one must take advantage of the known asymptotic solutions. Following the procedure developed by Numerov (Numerov, 1924; Melkanov, Sawada & Raynal, 1966), an auxiliary function, is introduced: where and Now is computed. Starting with and , the integration of following the Numerov procedure is given by This recurrence relation is carried through for steps, where a is the asymptotic limit. At the asymptotic limit is The proportionality factor is eliminated by matching the logarithmic derivative of to the same derivative of at r = a. From this equality, the partial wave phase shifts are calculated as follows: Solving for leads to where and
It is straightforward to calculate the scattering amplitudes by partial wave summation since stable numerical methods are readily available for the spherical Bessel functions, , the Neumann functions, , and the Legendre polynomials, (Yates, 1971).
Particular attention was given to the choices of the integration step size, , and the matching radius, a. Both were varied to ensure the stability of the scattering factors to 0.1% for light atoms and to 0.3% for heavier atoms and higher incident energies. The results of the sensitivity calculations are summarized elsewhere (Ross & Fink, 1986).
Smoothing was carried out by the following procedure: Sixteen data points, quarter s units apart, were least-squares fit to a cubic polynomial and the eighth point was changed to lie on this analytical curve. This procedure was repeated in running point average mode for Å−1. The points for Å−1 were left unchanged since no oscillations were seen. Smoothed and unsmoothed data in quarter s units for f and g are available on tape at cost from the authors.
Total inelastic scattering in the first Born approximation (Bonham & Fink, 1974) is obtained by including all possible excitation processes: where is the nuclear electron vector, , , is the energy loss of the incident electron upon excitation of the scatterer to the nth state, is the Bragg angle, signifies a sum over all bound states and an integration over the continuum, and N is equal to the number of electrons in the atom. The sum is carried out over all states for which is less than the incident electron energy. The Morse approximation is obtained by making three assumptions: (1) that the incident energy is so high that , i.e. that all states are accessible; (2) that the ratio is unity for all inelastic processes of any importance; and (3) that may be replaced by its elastic value s. With these approximations, closure may be used to obtain (Morse, 1932; Heisenberg, 1931; Bethe, 1930): The function S(s) is the X-ray incoherent scattering factor (Wang, Sagar, Schmider & Smith, 1993) and is related to the inelastic electron scattering cross section by Inelastic scattering factors for X-rays and electrons are given in Table 4.3.3.2 in the Morse (1932) approximation for elements Z = 1 to Z = 92 with HF wave functions (Bunge, Barrientos & Bunge, 1993; McLean & McLean, 1981).
|
There are two kinds of relativistic correction that can be made on inelastic scattering factors. The first is for relativistic effects on the atomic field and has been neglected. This should not be too serious since HF wavefunctions are used and the corrections are only large for the heavier atoms where the contribution to the total scattering for 3–4 Å−1 tends to be negligible. The other correction is for effects in the scattering process, which can be significant above 40 keV, but again these corrections tend to be localized to the small-angle region ( 3 Å−1) (Yates, 1970). Hence the tables of inelastic scattering factors given here are based on HF atomic fields since these appear to be the most accurate results presently available.
The inelastic scattering equations must be modified in order to compare theory with experiment. First, the Morse theory is corrected to ensure that both energy and momentum are conserved in the scattering process. In the description of the elastic scattering process, no transformation is required from the centre-of-mass system (CMS), where the scattering factors are calculated, to the laboratory system (LS), where data are taken, since the nuclei are heavy compared with the incident electrons. In the inelastic channels, a similar argument holds for scattering involving the bound states. However, for ionizing processes, the interaction can be assumed to take place between the incident electron and the ejected electron, so that the CMS is entirely different from the LS. Considering the atomic electrons as free particles and considering only the ionization process, the transformation between the CMS and the LS is possible and leads to the Bethe modification (Tavard & Bonham, 1969) for inelastic scattering. The inelastic cross section can now be given by for and by for .
Another modification is necessary because the average energy of inelastically scattered electrons varies with energy and is given from approximate conservation of energy and momentum for a fast incident particle by . This means that for 30 Å−1 at 40 keV the average energy of inelastically scattered electrons may be around 30 keV and the fact that the response of the detector may be different for the 40 keV inelastically scattered electrons and the elastic ones may have to be considered (Fink, Bonham, Lee & Ng, 1969).
In addition to the values given in Table 4.3.3.2, a few calculations of S(s) have been carried out with very exact wavefunctions that include more than 85% of the correlation energy (Kohl & Bonham, 1967; Bartell & Gavin, 1964; Peixoto, Bunge & Bonham, 1969; Thakkar & Smith, 1978; Wang, Esquivel, Smith & Bunge, 1995).
Errors in the inelastic scattering factors from the three approximations made in the Morse theory have been investigated (Tavard & Bonham, 1969; Bonham, 1965b). The Morse theory breaks down at very large scattering angles , and is incorrect at small angles. Investigations carried out so far indicate that the small-angle failure is not serious outside s = 1 Å−1. It must be stressed that these uncertainties do not introduce important errors into the analysis of molecular structure using theoretical atomic scattering amplitudes. This is mainly because such deviations are smooth compared with molecular features and thus do not interfere with the analysis of molecular structure.
The simplest theory of molecular scattering assumes that a molecule consists of spherical atoms and that each electron is scattered by only one atom in the molecule. If only single scattering is allowed within each atom, the molecular intensity can be written as where M is the number of constituent atoms in the molecule, and are the coherent and incoherent X-ray scattering factors, and is the probability of finding atom i at a distance r from atom j at the temperature T (Bonham & Su, 1966; Kelley & Fink, 1982b; Mawhorter, Fink & Archer, 1983; Mawhorter & Fink, 1983; Miller & Fink, 1985; Hilderbrandt & Kohl, 1981; Kohl & Hilderbrandt, 1981). The constant is proportional to the product of the intensities of the electron and molecular beams and R is the distance from the point of scattering to the detector. The single sum is the atomic intensity and the double sum is the molecular intensity . This expression, referred to here as the independent atom model (IAM), may be improved by replacing the atomic elastic electron scattering factors by their partial wave counterparts. This modification is necessary to explain the failure of the Born approximation observed in molecules containing light and heavy atoms in proximity (Schomaker & Glauber, 1952; Seip, 1965), and may be written as This is the most commonly used expression for the interpretation of molecular gas electron-diffraction patterns in the keV energy range. If it is necessary to consider relativistic effects in the scattering intensity, equation (4.3.3.2) becomes (Yates & Bonham, 1969) where and refer to the scattering-factor magnitude and phase for electrons that have changed their electron spin state during the scattering process and and refer to retention of spin orientation. The incident electron beam is assumed to be unpolarized and no attempt has been made to consider relativistic effects on the inelastic scattering cross section, which is usually negligible in the structural s range.
If it is necessary to consider binding effects, the first Born approximation may usually be used in describing molecular scattering, since binding effects are largest for molecules containing small atoms where the Born approximation is most valid.
The exact expression for I(s) in the first Born approximation can be written as (Bonham & Fink, 1974; Tavard & Roux, 1965; Tavard, Rouault & Roux, 1965; Iijima, Bonham & Ando, 1963; Bonham, 1967) where and The brackets denote averaging over the vibrational motion, is the Dirac delta function, and is the molecular wavefunction. Binding effects appear to be proportional to the ratio of the number of electrons involved in binding to the total number of electrons in the system (Kohl & Bonham, 1967; Bonham & Iijima, 1965) so that binding effects in molecules containing mainly heavy atoms should be quite small.
The intensities, I(s), for many small molecules have been calculated based on molecular Hartree–Fock wavefunctions. In most cases, a distinctive minimum has been found at about s = 3–4 Å−1 and a much small maximum at s = 8–10 Å−1 in the cross-sectional difference curve between the IAM and the molecular HF results (Pulay, Mawhorter, Kohl & Fink, 1983; Kohl & Bartell, 1969; Liu & Smith, 1977; Epstein & Stewart, 1977; Sasaki, Konaka, Iijima & Kimura, 1982; Shibata, Hirota, Kakuta & Muramatsu, 1980; Horota, Kakuta & Shibata, 1981; Xie, Fink & Kohl, 1984). Further studies using correlated wavefunctions (accounting for up to 60% of the correlation energy) showed that in the elastic channel the binding effects are only weakly modified; only the maximum at s = 8–10 Å−1 is further reduced. However, strong effects are seen in the inelastic channel, deepening the minimum at s = 3–4 Å−1 significantly (Breitenstein, Endesfelder, Meyer, Schweig & Zittlau, 1983; Breitenstein, Endesfelder, Meyer & Schweig, 1984; Breitenstein, Mawhorter, Meyer & Schweig, 1984; Wang, Tripathi & Smith, 1994). Detailed calculations on CO2 and H2O averaging over many internuclear distances and applying the pair distribution functions showed that vibrational effects do not alter the binding effects (Breitenstein, Mawhorter, Meyer & Schweig, 1986). For CO2, the calculations have been confirmed in essence by an experimental set of data (McClelland & Fink, 1985). However, more molecules and more detailed analysis will be available in the future. The binding effects make it desirable to avoid the small-angle-scattering range when structural information is the main goal of a diffraction analysis.
The problem of intramolecular multiple scattering may necessitate corrections to the molecular intensity when three or more closely spaced heavy atoms are present. This correction (Karle & Karle, 1950; Hoerni, 1956; Bunyan, 1963; Gjønnes, 1964; Bonham, 1965a, 1966) appears to be more serious for three atoms in a right triangular configuration than for a collinear arrangement of three atoms. A case study by Kohl & Arvedson (1980) on SF6 showed the importance of multiple scattering. However, their approach is too cumbersome to be used in routine structure work. A very good approximate technique is available utilizing the Glauber approximation (Bartell & Miller, 1980; Bartell & Wong, 1972; Wong & Bartell, 1973; Bartell, 1975); Kohl's results are reproduced quite well using the atomic scattering factors only. Several applications of the multiple scattering routines showed that the internuclear distances are rather insensitive to this perturbation, but the mean amplitudes of vibration can easily change by 10% (Miller & Fink, 1981; Kelley & Fink, 1982a; Ketkar & Fink, 1985).
Among the different spectroscopies available for investigating the electronic excitation spectrum of solids, inelastic electron scattering experiments are very useful because the range of accessible energy and momentum transfer is very large, as illustrated in Fig. 4.3.4.1 taken from Schnatterly (1979). Absorption measurements with photon beams follow the photon dispersion curve, because it is impossible to vary independently the energy and the momentum of a photon. In a scattering experiment, a quasi-parallel beam of monochromatic particles is incident on the specimen and one measures the changes in energy and momentum that can be attributed to the creation of a given excitation in the target. Inelastic neutron scattering is the most powerful technique for the low-energy range 0.1 eV). On the other hand, inelastic X-ray scattering is well suited for the study of high momentum and large energy transfers because the energy resolution is limited to ∼1 eV and the cross section increases with momentum transfer. In the intermediate range, inelastic electron scattering [or electron energy-loss spectroscopy (EELS)] is the most useful technique. For recent reviews on different aspects of the subject, the reader may consult the texts by Schnatterly (1979), Raether (1980), Colliex (1984), Egerton (1986), and Spence (1988a).
The importance of inelastic scattering as a function of energy and momentum transfer is governed by a double differential cross section: where d corresponds to the solid angle of acceptance of the detector and d(ΔE) to the energy window transmitted by the spectrometer. The experimental conditions must therefore be defined before any interpretation of the data is possible. Integrations of the cross section over the relevant angular and energy-loss domains correspond to partial or total cross sections, depending on the feature measured. For instance, the total inelastic cross section corresponds to the probability of suffering any energy loss while being scattered into all solid angles. The discrimination between elastic and inelastic signal is generally defined by the energy resolution of the spectrometer, that is the minimum energy loss that can be unambiguously distinguished from the zero-loss peak; it is therefore very dependent on the instrumentation used.
The kinematics of a single inelastic event can be described as shown in Fig. 4.3.4.2 . In contrast to the elastic case, there is no simple relation between the scattering angle and the transfer of momentum . One has also to take into account the energy loss ΔE. Combining both equations of conservation of momentum and energy, and one obtains where the fundamental units = Bohr radius and = Rydberg energy are used to introduce dimensionless quantities. In this kinematical description, one deals only with factors concerning the primary or the scattered particle, without considering specifically the information on the ejected electron. For a core-electron excitation of an atom, one distinguishes q (the momentum exchanged by the incident particle) and χ (the momentum gained by the excited electron), the difference being absorbed by the recoil of the target nucleus (Maslen & Rossouw, 1983).
The strong coupling potential between the primary electron and the solid target is responsible for the occurrence of multiple inelastic events (and of mixed inelastic–elastic events) for thick specimens. To describe the interaction of a primary particle with an assembly of randomly distributed scattering centres (with a density N per unit volume), a useful concept is the mean free path defined as for the cross section σ. The ratio t/Λ measures the probability of occurrence of the event associated with the cross section σ when the incident particle travels a given length t through the material. This is true in the single scattering case, that is when .
For increased thicknesses, one must take into account all multiple scattering events and this contribution begins to be non-negligible for a few tens of nanometres. Multiple scattering is responsible for a broadening of the angular distribution of the scattering electrons – mostly due to single or multiple elastic events – and of an important fraction of inelastic electrons suffering more than one energy loss. The probability of having n inelastic processes of mean free path Λ is governed by the Poisson distribution: Multiple losses introduce additional peaks in the energy-loss spectrum; they are also responsible for an increased background that complicates the detection of single characteristic core-loss signals. Consequently, when the specimen thickness is not very small (i.e. for for 100 keV primary electrons), it is necessary to retrieve the single scattering profile that is truly representative of the electronic and chemical properties of the specimen.
Deconvolution techniques have therefore been developed to remove the effects of plural scattering from the low-loss spectrum (up to 100 eV) or from ionization edges but very few treatments deal with both angle and energy-loss distributions. Batson & Silcox (1983) have made a detailed study of the inelastic scattering properties of incident 75 keV electrons through a ~100 nm thick polycrystalline aluminium film, over the full range of transferred wavevectors and energy losses . Schattschneider (1983) has proposed a matrix approach that is sufficiently elaborate to handle angle-resolved energy-loss data. Simpler deconvolution schemes have been proposed and used for processing multiple losses without specific consideration of angular truncation effects. They are valid when the data have been recorded over sufficiently large angles of collection so that all appreciable inelastic scattering is included. They are generally based on Fourier transform techniques, except for the iterative approach of Daniels, Festenberg, Raether & Zeppenfeld (1970), which has been used for energy losses up to about 60 eV (Colliex, Gasgnier & Trebbia, 1976). The most accurate current methods are the Fourier-log method of Johnson & Spence (1974) and Spence (1979), and the Fourier-ratio method of Swyt & Leapman (1982), which only applies to the core-loss region. The first is far more complete and accurate and applies to any spectral range when one has recorded a full spectrum in unsaturated conditions.
4.3.4.1.4. Classification of the different types of excitations contained in an electron energy-loss spectrum
Figs. 4.3.4.3 and 4.3.4.4 display examples of electron energy-loss spectra over large energy domains, typically from 1 to about 2000 eV. With one instrument, all elementary excitations from the near infrared to the X-ray domain can be investigated. Apart from the main beam or zero-loss peak, two major families of electronic transitions can be distinguished in such spectra:
The non-characteristic background is due to the superposition of several contributions: the high-energy tail of valence-electron scattering, the tails of core losses with lower binding energy, Bremsstrahlung energy losses, plural scattering, etc. It is therefore rather difficult to model its behaviour, although some efforts have been made along this direction using Monte Carlo simulation of multiple scattering (Jouffrey, Sevely, Zanchi & Kihn, 1985).
When one monochromatizes the natural energy width of the primary beam to much smaller values (about a few meV) than its natural width, one has access to the infrared part of the electromagnetic spectrum. An example is provided in Fig. 4.3.4.6 for a specimen of germanium in the energy-loss range 0 up to 500 meV. In this case, one can investigate phonon modes, or the bonding states of impurities on surfaces. This field has been much less extensively studied than the higher-energy-loss range [for references see Ibach & Mills (1982)].
Energy-loss spectrum, in the meV region, of an evaporated germanium film (thickness 25 nm). Primary electron energy 25 keV. Scattering angle < 10−4. One detects the contributions of the phonon excitation, of the Ge—O bonding, and of intraband transitions [courtesy of Schröder & Geiger (1972)]. |
Generally, EELS techniques can be applied to a large variety of specimens. However, for the following review to remain of limited size, it is restricted to electron energy-loss spectroscopy on solids and surfaces in transmission and reflection. It omits some important aspects such as electron energy-loss spectroscopy in gases with its associated information on atomic and molecular states. In this domain, a bibliography of inner-shell excitation studies of atoms and molecules by electrons, photons or theory is available from Hitchcock (1982).
In a dedicated instrument for electron inelastic scattering studies, one aims at the best momentum and energy resolution with a well collimated and monochromatized primary beam. This is achieved at the cost of poor spatial localization of the incident electrons and one assumes the specimens to be homogeneous over the whole irradiated volume. In a sophisticated instrument such as that built by Fink & Kisker (1980), the energy resolution can be varied from 0.08 to 0.7 eV, and the momentum transfer resolution between 0.03 and 0.2 Å−1, but typical values for the electron-beam diameter are about 0.2 to 1 mm. Nowadays, many energy-analysing devices are coupled with an electron microscope: consequently, an inelastic scattering study involves recording for a primary intensity , the current I(r, , ΔE) scattered or transmitted at the position r on the specimen, in the direction with respect to the primary beam, and with an energy loss ΔE. Spatial resolution is achieved either with a focused probe or by a selected area method, angular acceptance is defined by an aperture, and energy width is controlled by a detector function after the spectrometer. It is not possible from signal-to-noise considerations to reduce simultaneously all instrumental widths to very small values. One of the parameters (r, or ΔE) is chosen for signal integration, another for selection, and the last is the variable. Table 4.3.4.1 classifies these different possibilities for inelastic scattering studies.
|
Because of the great variety of possible EELS experiments, it is impossible to build an optimum spectrometer for all applications. For instance, the design of a spectrometer for low-energy incident electrons and surface studies is different from that for high-energy incident electrons and transmission work. In the latter category, instruments built for dedicated EELS studies (Killat, 1974; Gibbons, Ritsko & Schnatterly, 1975; Fink & Kisker, 1980; etc.) are different from those inserted within an electron-microscope environment, in which case it is possible to investigate the excitation spectrum from a specimen area well characterized in image and diffraction [see the reviews by Colliex (1984) and Egerton (1986)].
The literature on dispersive electron–optical systems (equivalent to optical prisms) is very large. For example, the theory of uniform field magnets, which constitute an important family of analysing devices, has been extensively developed for the components in high-energy particle accelerators (Enge, 1967; Livingood, 1969). As for EELS spectrometers, they can be classified as:
|
Fig. 4.3.4.7 defines the basic parameters of a `general' energy-loss spectrometer: a region of electrostatic E and/or magnetic B fields transforms a distribution of electrons in the object plane of coordinate along the principal trajectory, into a distribution of electrons in the object plane of coordinate , coincident with the detector plane (or optically conjugate to it). The transverse coordinates are labelled as (x, y), the angular ones as (t, u), and ρ = Δp/p = ΔE/2E is the relative change in absolute momentum value associated with the energy loss.
Schematic drawing of a uniform magnetic sector spectrometer with induction B normal to the plane of the figure. Definition of the coordinates used in the text (the object plane at coordinate z0 along the mean trajectory coincides with the specimen, and the image plane at z1 coincides with the dispersion plane and the detector level). |
Common properties of such systems are:
The spectrometer performance can be evaluated with the following parameters:
|
From this simplified discussion, one deduces that there is generally competition between large angular acceptance for the inelastic signal, which is very important for obtaining a high signal-to-noise ratio (SNR) for core-level excitations, and good energy resolution. Two solutions have been used to remedy this limitation. The first is to improve spectrometer design, for example by correcting second-order aberrations in a homogeneous magnetic prism (Crewe, 1977a; Parker, Utlaut & Isaacson, 1978; Egerton, 1980b; Krivanek & Swann, 1981; etc). This can enhance the figure of merit by at least a factor of 100. The second possibility is to transform the distribution of electrons to be analysed at the exit surface of the specimen into a more convenient distribution at the spectrometer entrance. This can be done by introducing versatile transfer optics (see Crewe, 1977b; Egerton, 1980a; Johnson, 1980; Craven & Buggy, 1981; etc.). As a final remark on the energy resolution of a spectrometer, it is meaningless to define it without reference to the size and the angular aperture of the analysed beam.
Historically, many types of spectrometer have been used since the first suggestion by Wien (1897) that an energy analyser could be designed by employing crossed electric and magnetic fields. Reviews have been published by Klemperer (1965), Metherell (1971), Pearce-Percy (1978), and Egerton (1986). Nowadays, two configurations are mostly used and have become commercially available on modern electron microscopes: these are spectrometers on TEM/STEM instruments and filters on CTEM ones. In the first case, homogeneous magnetic sectors are the simplest and most widely used devices. Recent instrumental developments by Shuman (1980), Krivanek & Swann (1981), and Scheinfein & Isaacson (1984) have given birth to a generation of spectrometers with the following major characteristics: double focusing, correction for second-order aberrations, dispersion plane perpendicular to the trajectory. This has been made possible by a suitable choice of several parameters, such as the tilt angles and the radius of curvature for the entrance and exit faces and the correct choice of the object source position. As an example, for a 100 keV STEM equipped with a field emission gun, the coupling illustrated in Fig. 4.3.4.9 leads to an energy resolution of 0.35 eV for β0 = 7.5 mrad on the specimen as visible on the elastic peak, and 0.6 eV for α0 = 25 mrad as checked on the fine structures on core losses. Krivanek, Manoubi & Colliex (1985) demonstrated a sub-eV energy resolution over the whole range of energy losses up to 1 or 2 keV.
A very competitive solution is the Wien filter, which consists of uniform electric and magnetic fields crossed perpendicularly, see Fig. 4.3.4.10 . An electron travelling along the axis with a velocity such that is not deflected, the net force on it being null. All electrons with different velocities, or at some angle with respect to the optical axis, are deflected. The dispersion of the system is greatly enhanced by decelerating the electrons to about 100 eV within the filter, in which case a few 100 µm/eV. A presently achievable energy resolution of 150 meV at a spectrometer collection half-angle of 12.5 mrad has been demonstrated by Batson (1986, 1989). It allows the study of the detailed shape of the energy distribution of the electrons emitted from a field emission source and the taking of it into account in the investigation of band-gap states in semiconductors (Batson, 1987).
Principle of the Wien filter used as an EELS spectrometer: the trajectories are shown in the two principal (dispersive and focusing) sections. |
Filtering devices have been designed to form an energy-filtered image or diffraction pattern in a CTEM. The first solution, reproduced in Fig. 4.3.4.11 , is due to Castaing & Henry (1962). It consists of a double magnetic prism and a concave electrostatic mirror biased at the potential of the microscope cathode. The system possesses two pairs of stigmatic points that may coincide with a diffraction plane and an image plane of the electron-microscope column. One of these sets of points is achromatic and can be used for image filtering. The other is strongly chromatic and is used for spectrum analysis. Zanchi, Sevely & Jouffrey (1977) and Rose & Plies (1974) have proposed replacing this system, which requires an extra source of high voltage for the mirror, by a purely magnetic equivalent device. Several solutions, known as the α and filters, with three or four magnets, have thus been built, both on very high voltage microscopes (Zanchi, Perez & Sevely, 1975) and on more conventional ones (Krahl & Herrmann, 1980), the latest version now being available from one EM manufacturer (Zeiss EM S12).
The final important component in EELS is the detector that measures the electron flux in the dispersion plane of the spectrometer and transfers it through a suitable interface to the data storage device for further computer processing. Until about 1990, all systems were operated in a sequential acquisition mode. The dispersed beam was scanned in front of a narrow slit located in the spectrometer dispersion plane. Electrons were then generally recorded by a combination of scintillator and photomultiplier capable of single electron counting.
This process is, however, highly inefficient: while the counts are measured in one channel, all information concerning the other channels is lost. These requirements for improved detection efficiency have led to the consideration of possible solutions for parallel detection of the EELS spectrum. They use a multiarray of detectors, the position, the size and the number of which have to be adapted to the spectral distribution delivered by the spectrometer. In most cases with magnetic type devices, auxiliary electron optics has to be introduced between the spectrometer and the detector so that the dispersion matches the size of the individual detection cells. Different systems have been proposed and tested for recording media, the most widely used solutions at present being the photodiode and the charge-coupled diode arrays described by Shuman & Kruit (1985), Krivanek, Ahn & Keeney (1987), Strauss, Naday, Sherman & Zaluzec (1987), Egerton & Crozier (1987), Berger & McMullan (1989), etc. Fig. 4.3.4.12 shows a device, now commercially available from Gatan, that is made of a convenient combination of these different components. This progress in detection has led to significant improvements in many areas of EELS: enhanced detection limits, reduced beam damage in sensitive materials, data of improved quality in terms of both SNR and resolution, and access to time-resolved spectroscopy at the ms time scale (chronospectra). Several of these important consequences are illustrated in the following sections.
Most inelastic interaction of fast incident electrons is with outer atomic shells in atoms, or in solids with valence electrons (referred to as conduction electrons in metals). These involve excitations in the 0–50 eV range, but, in a few cases, interband transitions from low-binding-energy shells may also contribute.
The basic concept introduced by the many-body theory in the interacting free electron gas is the volume plasmon. In a condensed material, the assembly of loosely bound electrons behaves as a plasma in which collective oscillations can be induced by a fast external charged particle. These eigenmodes, known as volume plasmons, are longitudinal charge-density fluctuations around the average bulk density in the plasma n 1028 e−/m3). Their eigen frequency is given, in the free electron gas, as The corresponding energy, measured in an energy-loss spectrum (see the famous example of the plasmon in aluminium in Fig. 4.3.4.3), is the plasmon energy, for which typical values in a selection of pure solid elements are gathered in Table 4.3.4.2. The accuracies of the measured values depend on several instrumental parameters. Moreover, they are sensitive to the specimen crystalline state and to its degree of purity. Consequently, there exist slight discrepancies between published values. Numbers listed in Table 4.3.4.2 must therefore be accepted with a 0.1 eV confidence. Some specific cases require comments: amorphous boron, when prepared by vacuum evaporation, is not a well defined specimen. Carbon exists in several allotropic varieties. The selection of the diamond type in the table is made for direct comparison with the other tetravalent specimens (Si, Ge, Sn). The results for lead (Pb) are still subject to confirmation. The volumic mass density is an important factor (through n) in governing the value of the plasmon energy. It varies with temperature and may be different in the crystal, in the amorphous, and in the liquid phases. In simple metals, the amorphous state is generally less dense than the crystalline one, so that its plasmon energy shifts to lower energies.
|
The above description applies only to very small scattering vectors q. In fact, the plasmon energy increases with scattering angle (and with momentum transfer ). This dependence is known as the dispersion relation, in which two distinct behaviours can be described:
|
Plasmon lifetime is inversely proportional to the energy width of the plasmon peak . Even for Al, with one of the smallest plasmon energy widths ( eV), the lifetime is very short: after about five oscillations, their amplitude is reduced to 1/e. Such a damping demonstrates the strength of the coupling of the collective modes with other processes. Several mechanisms compete for plasmon decay:
|
Within this free-electron-gas description, the differential cross section for the excitation of bulk plasmons by incident electrons of velocity v is given by where N is the density of atoms per volume unit and is the characteristic inelastic angle defined as in the non-relativistic description and as {with } in the relativistic case. The angular dependence of the differential cross section for plasmon scattering is shown in Fig. 4.3.4.14 . The integral cross section up to an angle is The total plasmon cross section is calculated for . Converted into mean free path, this becomes and
Measured angular dependence of the differential cross section dσ/dΩ for the 15 eV plasmon loss in Al (dots) compared with a calculated curve by Ferrell (solid curve) and with a sharp cut-off approximation at θc (dashed curved). Also shown along the scattering angle axis, θE = characteristic inelastic angle defined as ΔE/2E0, = median inelastic angle defined by , and = average inelastic angle defined by [courtesy of Egerton (1986)]. |
The behaviour of as a function of the primary electron energy is shown in Fig. 4.3.4.15 .
The description of the bulk plasmon in the free-electron gas can be extended to any type of condensed material by introducing the dielectric response function , which describes the frequency and wavevector-dependent polarizability of the medium; cf. Daniels et al. (1970). One associates, respectively, the and functions with the propagation of transverse and longitudinal EM modes through matter. In the small-q limit, these tend towards the same value: As transverse dielectric functions are only used for wavevectors close to zero, the T and L indices can be omitted so that: The transverse solution corresponds to the normal propagation of EM waves in a medium of dielectric coefficient , i.e. to For longitudinal fields, the only solution is , which is basically the dispersion relation for the bulk plasmon.
In the framework of the Maxwell description of wave propagation in matter, it has been shown by several authors [see, for instance, Ritchie (1957)] that the transfer of energy between the beam electron and the electrons in the solid is governed by the magnitude of the energy-loss function , so that One can deduce (4.3.4.14) by introducing a δ function at energy loss for the energy-loss function: As a consequence of the causality principle, a knowledge of the energy-loss function over the complete frequency (or energy-loss) range enables one to calculate by Kramers–Kronig analysis: where PP denotes the principal part of the integral. For details of efficient practical evaluation of the above equation, see Johnson (1975).
The dielectric functions can be easily calculated for simple descriptions of the electron gas. In the Drude model, i.e. for a free-electron plasma with a relaxation time τ, the dielectric function at long wavelengths is with , as above. The behaviour of the different functions, the real and imaginary terms in , and the energy-loss function are shown in Fig. 4.3.4.16 . The energy-loss term exhibits a sharp Lorentzian profile centred at and of width 1/τ. The narrower and more intense this plasmon peak, the more the involved valence electrons behave like free electrons.
Dielectric and optical functions calculated in the Drude model of a free-electron gas with ħωp = 16 eV and τ = 1.64 × 10−16 s. R is the optical reflection coefficient in normal incidence, i.e. R = [(n − 1)2 + k2]/(n + 1)2 + k2] with n and k the real and imaginary parts of . The effective numbers and are defined in Subsection 4.3.4.5 [courtesy of Daniels et al. (1970)]. |
In the Lorentz model, i.e. for a gas of bound electrons with one or several excitation eigenfrequencies , the dielectric function is where denotes the density of electrons oscillating with the frequency and is the associated relaxation time. The characteristic , , and behaviours are displayed in Fig. 4.3.4.17 : a typical `interband' transition (in solid-state terminology) can be revealed as a maximum in the function, simultaneous with a `plasmon' mode associated with a maximum in the energy-loss function and slightly shifted to higher energies with respect to the annulation conditions of the function.
Same as previous figure, but for a Lorentz model with an oscillator of eigenfrequency ħω0 = 10 eV and relaxation time τ0 = 6.6 × 10−16 s superposed on the free-electron term [courtesy of Daniels et al. (1970)]. |
In most practical situations, there coexist a family of free electrons (with plasma frequency and one or several families of bound electrons (with eigenfrequencies . The influence of bound electrons is to shift the plasma frequency towards lower values if and to higher values if . As a special case, in an insulator, and all the electrons have a binding energy at least equal to the band gap , giving .
This description constitutes a satisfactory first step into the world of real solids with a complex system of valence and conduction bands between which there is a strong transition rate of individual electrons under the influence of photon or electron beams. In optical spectroscopy, for instance, this transition rate, which governs the absorption coefficient, can be deduced from the calculation of the factor as where is the matrix element for the transition from the occupied level j in the valence band to the unoccupied level in the conduction band, both with the same k value (which means for a vertical transition). is the joint density of states (JDOS) with the energy difference . This formula is also valid for small-angle-scattering electron inelastic processes. When parabolic bands are used to represent, respectively, the upper part of the valence band and the lower part of the conduction band in a semiconductor, the dominant JDOS term close to the onset of the interband transitions takes the form where is the band-gap energy. This concept has been successfully used by Batson (1987) for the detection of gap energy variations between the bulk and the vicinity of a single misfit dislocation in a GaAs specimen. The case of non-vertical transitions involving integration over k-space has also been considered (Fink et al., 1984; Fink & Leising, 1986).
The dielectric constants of many solids have been deduced from a number of methods involving either primary photon or electron beams. In optical measurements, one obtains the values of and from a Krakers–Kronig analysis of the optical absorption and reflection curves, while in electron energy-loss measurements they are deduced from Kramers–Kronig analysis of energy-loss functions.
Fig. 4.3.4.18 shows typical behaviours of the dielectric and energy-loss functions.
|
Čerenkov radiation is emitted when the velocity v of an electron travelling through a medium exceeds the speed of light for a particular frequency in this medium. The criterion for Čerenkov emission is
In an insulator, is positive at low energies and can considerably exceed unity, so that a `radiation peak' can be detected in the corresponding energy-loss range (between 2 and 4 eV in Si, Ge, III–V compounds, diamond, ); see Von Festenberg (1968), Kröger (1970), and Chen & Silcox (1971). The associated scattering angle, for high-energy electrons, is very small and this contribution can only be detected using a limited forward-scattering angular acceptance.
In an anisotropic crystal, the dielectric function has the character of a tensor, so that the energy-loss function is expressed as
If it is transformed to its orthogonal principal axes , and if the q components in this system are , the above expression simplifies to
In a uniaxial crystal, such as a graphite, and (i.e. parallel to the c axis): where is the angle between q and the c axis. The spectrum depends on the direction of q, either parallel or perpendicular to the c axis, as shown in Fig. 4.3.4.19 from Venghaus (1975). These experimental conditions may be achieved by tilting the graphite layer at 45° with respect to the incident axis, and recording spectra in two directions at with respect to it (see Fig. 4.3.4.20 ).
Dielectric functions in graphite derived from energy losses for E ⊥ c (i.e. the electric field vector being in the layer plane) and for E||c [from Daniels et al. (1970)]. The dashed line represents data extracted from optical reflectivity measurements [from Taft & Philipp (1965)]. |
Volume plasmons are longitudinal waves of charge density propagating through the bulk of the solid. Similarly, three exist longitudinal waves of charge density travelling along the surface between two media A and B (one may be a vacuum): these are the surface plasmons (Kliewer & Fuchs, 1974). Boundary conditions imply that The corresponding charge-density fluctuation is contained within the (x) boundary plane, z being normal to the surface: and the associated electrostatic potential oscillates in space and time as The characteristic energy of this surface mode is estimated in the free electron case as:
In the spherical interface case: (metal sphere in vacuum – the modes are now quantified following the l quantum number in spherical geometry); (spherical void within metal).
Thin-film geometry: (metal layer of thickness t embedded in dielectric films of constant ). The two solutions result from the coupling of the oscillations on the two surfaces, the electric field being symmetric for the mode and antisymmetric for the .
In a real solid, the surface plasmon modes are determined by the roots of the equation for vacuum coating [or for dielectric coating].
The probability of surface-loss excitation is mostly governed by the energy-loss function, which is analogous for surface modes to the bulk energy-loss function. In normal incidence, the differential scattering cross section is zero in the forward direction, reaches a maximum for , and decreases as at large angles. In non-normal incidence, the angular distribution is asymmetrical, goes through a zero value for momentum transfer in a direction perpendicular to the interface, and the total probability increases as where is the incidence angle between the primary beam and the normal to the surface. As a consequence, the probability of producing one (and several) surface losses increases rapidly for grazing incidences.
As for any core-electron spectroscopy, EELS spectroscopy at higher energy losses mostly deals with the excitation of well defined atomic electrons. When considering solid specimens, both initial and final states in the transition are actually eigenstates in the solid state. However, the initial wavefunction can be considered as purely atomic for core excitations. As a first consequence, one can classify these transitions as a function of the parameters of atomic physics: Z is the atomic number of the element; n, l, and j = l + s are the quantum numbers describing the subshells from which the electron has been excited. The spectroscopy notation used is shown in Fig. 4.3.4.21 . The list of major transitions is displayed as a function of Z and in Fig. 4.3.4.22 .
Definition of electron shells and transitions involved in core-loss spectroscopy [from Ahn & Krivanek (1982)]. |
Chart of edges encountered in the 50 eV up to 3 keV energy-loss range with symbols identifying the types of shapes [see Ahn & Krivanek (1982) for further comments]. |
Core excitations appear as edges superimposed, from the threshold energy upwards, above a regularly decreasing background. As explained below, the basic matrix element governing the probability of transition is similar for optical absorption spectroscopy and for small-angle-scattering EELS spectroscopy. Consequently, selection rules for dipole transitions define the dominant transitions to be observed, i.e. This major rule has important consequences for the edge shapes to be observed: approximate behaviours are also shown in Fig. 4.3.4.22. A very useful library of core edges can be found in the EELS atlas (Ahn & Krivanek, 1982), from which we have selected the family of edges gathered in Fig. 4.3.4.23 . They display the following typical profiles:
|
4.3.4.4.2. Bethe theory for inelastic scattering by an isolated atom (Bethe, 1930; Inokuti, 1971; Inokuti, Itikawa & Turner, 1978, 1979)
As a consequence of the atomic nature of the excited wavefunction in core-loss spectroscopy, the first step involves deriving a useful theoretical expression for inelastic scattering by an isolated atom. The differential cross section for an electron of wavevector k to be scattered into a final plane wave of vector k′, while promoting one atomic electron from to , is given in a one-electron excitation description by see, for instance, Landau & Lifchitz (1966) and Mott & Massey (1952). The potential V(r) corresponds to the Coulomb interaction with all charges (both in the nucleus and in the electron cloud) of the atom. The momentum change in the scattering event is . The final-state wavefunction is normalized per unit energy range. The orthogonality between initial- and final-state wavefunctions restricts the inelastic scattering to the only interactions with atomic electrons:
The first part of the above expression has the form of Rutherford scattering. γ is introduced to deal, to a first approximation, with relativistic effects. The ratio k′/k is generally assumed to be equal to unity. This kinematic scattering factor is modified by the second term, or matrix element, which describes the response of the atomic electrons: where the sum extends over all atomic electrons at positions . The dimensionless quantity is known as the inelastic form factor.
For a more direct comparison with photoabsorption measurements, one introduces the generalized oscillator strength (GOS) as for transitions towards final states in the continuum [ΔE is then the energy difference between the core level and the final state of kinetic energy above the Fermi level, scaled in energy to the Rydberg energy (R)]. Also, for transition towards bound states. In this case, is the energy difference between the two states involved.
The generalized oscillator strength is a function of both the energy ΔE and the momentum transferred to the atom. It is displayed as a three-dimensional surface known as the Bethe surface (Fig. 4.3.4.24 ), which embodies all information concerning the inelastic scattering of charged particles by atoms. The angular dependence of the cross section is proportional to at a given energy loss ΔE.
Bethe surface for K-shell ionization, calculated using a hydrogenic model. The generalized oscillator strength is zero for energy loss E below the threshold EK. The horizontal coordinate is related to scattering angle through q [from Egerton (1979)]. |
In the small-angle limit , where is the average radius of the initial orbital), the GOS reduces to the optical oscillator strength and where u is the unit vector in the q direction. When one is concerned with a given orbital excitation, the sum over reduces to a single term r for this electron. With some elementary calculations, the resulting cross section is
The major angular dependence is contained, as in the low-loss domain, in the Lorentzian factor , with the characteristic inelastic angle being again equal to . Over this reduced scattering-angle domain, known as the dipole region, the GOS is approximately constant and the inner-shell EELS spectrum is directly proportional to the photoabsorption cross section , whose data can be used to test the results of single-atom calculations. For larger scattering angles, Fig. 4.3.4.24 exhibits two distinct behaviours for energy losses just above the edge (df/dΔE drops regularly to zero), and for energy losses much greater than the core-edge threshold. In the latter case, the oscillator strength is mostly concentrated in the Bethe ridge, the maximum of which occurs for:
This contribution at large scattering angles is equivalent to direct knock-on collisions of free electrons, i.e. to the curve lying in the middle of the valence-electron–hole excitations continuum (see Fig. 4.3.4.13). The non-zero width of the Bethe ridge can be used as an electron Compton profile to analyse the momentum distribution of the atomic electrons [see also §4.3.4.4.4(c)].
The energy dependence of the cross section, responsible for the various edge shapes discussed in §4.3.4.4.1, is governed by i.e. it corresponds to sections through the Bethe surface at constant q. Within the general theory described above, various models have been developed for practical calculations of energy differential cross sections.
The hydrogenic model due to Egerton (1979) is an extension of the quantum-mechanical calculations for a hydrogen atom to inner-shell electron excitations in an atom Z by introduction of some useful parametrization (effective nuclear charge, effective threshold energy). It is applied in practice for K and shells.
In the Hartree–Slater (or Dirac–Slater) description, one calculates the final continuum-state wavefunction in a self-consistent central field atomic potential (Leapman, Rez & Mayers, 1980; Rez, 1989). The radial dependence of these wavefunctions is given by the solution of a Schrödinger equation with an effective potential: where is the centrifugal potential, which is important for explaining the occurrence of delayed maxima in spectra involving final states of higher . This approach is now useful for any major edge, as illustrated by Ahn & Rez (1985) and more specifically in rare-earth elements by Manoubi, Rez & Colliex (1989).
These differential cross sections can be integrated over the relevant angular and energy domains to provide data comparable with experimental measurements. In practice, one records the energy spectral distribution of electrons scattered into all angles up to the acceptance value β of the collection aperture. The integration has therefore to be made from for the zero scattering-angle limit, up to . Fig. 4.3.4.25 shows how such calculated profiles can be used for fitting experimental data.
A novel technique for simulating an energy-loss spectrum with two distinct edges as a superposition of theoretical contributions (hydrogenic saw-tooth for O K, Lorentzian white lines and delayed continuum for Fe L2,3 calculated with the Hartree–Slater description). The best fit between the experimental and the simulated spectra is shown; it can be used to evaluate the relative concentration of the two elements [see Manoubi et al. (1990)]. |
Setting β = π [or equal to an effective upper limit corresponding to the criterion , the integral cross section is the total cross section for the excitation of a given core level. These ionization cross sections are required for quantification in all analytical techniques using core-level excitations and de-excitations, such as EELS, Auger electron spectroscopy, and X-ray microanalysis (see Powell, 1976, 1984). A convenient way of comparing total cross sections is to rewrite the Bethe asymptotic cross section as when the result is given in cm2, is the total cross section per atom or molecule or ionization of the nl subshell with edge energy , is the number of electrons on the nl level, and is the overvoltage defined as . and are two parameters representing phenomenologically the average number of electrons involved in the excitation and their average energy loss (one finds for the major K and edges 0.6–0.9 and 0.5–0.7). These values are in practice estimated from plots of curves as a function of , known as Fano plots. From least-squares fits to linear regions, one can evaluate the values of (slope of the curves) and of (coordinate at the origin) for various elements and shells. However, it has been shown more recently (Powell, 1989) that the interpretation of Fano plots is not always simple, since they typically display two linear regions. It is only in the linear region for the higher incident energies that the plots show the asymptotic Bethe dependence with the slope directly related to the optical data. At lower incident energies, another linear region is found with a slope typically 10–20% greater. Despite great progress over the last two decades, more cross-section data, either theoretical or experimental, are still required to improve to the 1% level the accuracy in all techniques using these signals.
The characteristic core edges recorded from solid specimens display complex structures different from those described in atomic terms. Moreover, their detailed spectral distributions depend on the type of compound in which the element is present (Leapman, Grunes & Fejes, 1982; Grunes, Leapman, Wilker, Hoffmann & Kunz, 1982; Colliex, Manoubi, Gasgnier & Brown, 1985). Modifications induced by the local solid-state environment concern (see Fig. 4.3.4.26 ) the following:
|
|
Since the early work of Hillier & Baker (1944), EELS spectroscopy has established itself as a prominent technique for investigating various aspects of the electronic structure of solids. As a fundamental application, it is now possible to construct a self-consistent set of data for a substance by combination of optical or energy-loss functions over a wide spectral range (Altarelli & Smith, 1974; Shiles, Sazaki, Inokuti & Smith, 1980: Hagemann, Gudat & Kunz, 1975). Sum-rule tests provide useful guidance in selecting the best values from the available measurements. The Thomas–Reiche–Kuhn f-sum rule can be expressed in a number of equivalent forms, which all require the knowledge of a function describing dissipative processes over all frequencies: One defines the effective number density of electrons contributing to these various absorption processes at an energy by the partial f sums: As an example, the values of from the infrared to beyond the K-shell excitation energy for metallic aluminium are shown in Fig. 4.3.4.32 . In this case, the conduction and core-electron contributions are well separated. One sees that the excitation of conduction electrons is virtually completed above the plasmon resonance only, but the different behaviour of the integrands below this value is a consequence of the fact that they describe different properties of matter: is a measure of the rate of energy dissipation from an electromagnetic wave, describes the decrease in amplitude of the wave, and is related to the energy loss of a fast electron. The above curve shows some exchange of oscillator strength from core to valence electrons, arising from the Pauli principle, which forbids transitions to occupied states for the deeper electrons.
Values of neff for metallic aluminium based on composite optical data [courtesy of Shiles et al. (1980)]. |
More practically, in the microanalytical domain, the combination of high performance attained by using EELS with parallel detection (i.e. energy resolution below 1 eV, spatial resolution below 1 nm, minimum concentration below 10−3 atom, time resolution below 10 ms) makes it a unique tool for studying local electronic properties in solid specimens.
The formation of textures in specimens for diffraction experiments is a natural consequence of the tendency for crystals of a highly anisotropic shape to deposit with a preferred orientation. The corresponding diffraction patterns may present some special advantages for the solution of problems of phase and structure analysis. Lamellar textures composed of crystals with the most fully developed face parallel to a plane but randomly rotated about its normal are specially important. The ease of interpretation of patterns of such textures when oriented obliquely to the primary beam (OT patterns) is a valuable property of the electron-diffraction method (Pinsker, 1953; Vainshtein, 1964; Zvyagin, 1967; Zvyagin, Vrublevskaya, Zhukhlistov, Sidorenko, Soboleva & Fedotov, 1979). Texture patterns (T patterns) are also useful in X-ray diffraction (Krinary, 1975; Mamy & Gaultier, 1976; Plançon, Rousseaux, Tchoubar, Tchoubar, Krinari & Drits, 1982).
If in the plane of orientation (the texture basis) the crystal has a two-dimensional cell a, b, γ, the c* axis of the reciprocal cell will be the texture axis. Reciprocal-lattice rods parallel to c* intersect the plane normal to them (the ab plane of the direct lattice) in the positions hk of a two-dimensional net that has periods and with an angle γ′ = π − γ between them, whatever the direction of the c axis in the direct lattice. The latter is defined by the absolute value c and the normal projection on the ab plane, with components , along the axes a, b. In the triclinic case, (Zvyagin et al., 1979). The lattice points of each rod with constant hk and integer l are at intervals of , but their real positions, described by their distances from the plane ab, depend on the projections of the axes a* and b* on c* (see Fig. 4.3.5.1 ), the equations being satisfied.
The relative orientations of the direct and the reciprocal axes and their projections on the plane ab, with indication of the distances Bhk and Dhkl that define the positions of reflections in lamellar texture patterns. |
The reciprocal-space representation of a lamellar texture is formed by the rotation of the reciprocal lattice of a single crystal about the c* axis. The rods hk become cylinders and the lattice points become circles lying on the cylinders. In the case of high-energy electron diffraction (HEED), the wavelength of the electrons is very short, and the Ewald sphere, of radius 1/λ, is so great that it may be approximated by a plane passing through the origin of reciprocal space and normal to the incident beam. The patterns differ in their geometry, depending on the angle through which the specimen is tilted from perpendicularity to the primary beam. At , the pattern consists of hk rings. When it contains a two-dimensional set of reflections hkl falling on hk ellipses formed by oblique sections of the hk cylinders. In the limiting case of , the ellipses degenerate into pairs of parallel straight lines theoretically containing the maximum numbers of reflections. The reflection positions are defined by two kinds of distances: (1) between the straight lines hk (length of the short axes of the ellipses hk): and (2) from the reflection hkl to the line of the short axes: In patterns obtained under real conditions , accelerating voltage V proportional to , distance L between the specimen and the screen), these values are presented in the scale of , also being proportional to with maximum value for the registrable reflections. The values of and , determined by the unit cells and the indices hkl, are the objects of the geometrical analysis of the OT patterns. When the symmetry is higher than triclinic, the expression for and are much simpler.
Such OT patterns are very informative, because the regular two-dimensional distribution of the hkl reflections permits definite indexing, cell determination, and intensity measurements. For low-symmetry and fine-grained substances, they present unique advantages for phase identification, polytypism studies, and structure analysis.
In the X-ray study of textures, it is impossible to neglect the curvature of the Ewald sphere and the number of reflections recorded is restricted to larger d values. However, there are advantages in that thicker specimens can be used and reflections with small values of , especially the 00l reflections, can be recorded. Such patterns are obtained in usual powder cameras with the incident beam parallel to the platelets of the oriented aggregate and are recorded on photographic film in the form of hkl reflection sequences along hk lines, as was demonstrated by Mamy & Gaultier (1976). The hk lines are no longer straight, but have the shapes described by Bernal (1926) for rotation photographs. It is difficult, however, to prepare good specimens. Other arrangements have been developed recently with advantages for precise intensity measurements. The reflections are recorded consecutively by means of a powder diffractometer fitted with a goniometer head. The relation between the angle of tilt and the angle of diffraction (twice the Bragg angle) depends on the reciprocal-lattice point to be recorded. If the latter is defined by a vector of length and by the angle between the vector and the plane of orientation (texture basis), the relation permits scanning of reciprocal space along any trajectory by proper choice of consecutive values of or . In particular, if is constant, the trajectory is a straight line passing through the origin at an angle to the plane of orientation (Krinary, 1975). Using additional conditions , , Plançon et al. (1982) realized the recording and the measurement of intensities along the cylinder-generating hk rods for different shapes of the misorientation function N(α).
In the course of development of electron diffractometry, a deflecting system has been developed that permits scanning the electron diffraction pattern across the fixed detector along any direction over any interval (Fig. 4.3.5.2 ). The intensities are measured point by point in steps of variable length. This system is applicable to any kind of two-dimensional intensity pattern, and in particular to texture patterns (Zvyagin, Zhukhlistov & Plotnikov, 1996). Electron diffractometry provides very precise intensity measurements and very reliable structural data (Zhukhlistov et al., 1997).
(a) Part of the OTED pattern of the clay mineral kaolinite and (b) the intensity profile of a characteristic quadruplet of reflections recorded with the electron diffractometry system. The scanning direction is indicated in (a). |
If the effective thickness of the lamellae is very small, of the order of the lattice parameter c, the diffraction pattern generates into a combination of broad but recognizably distinct 00l reflections and broad asymmetrical hk bands (Warren, 1941). The classical treatments of the shape of the bands were given by Méring (1949) and Wilson (1949) [for an elementary introduction see Wilson (1962)].
A fibre texture occurs when the crystals forming the specimen have a single direction in common. Each point of the reciprocal lattice describes a circle lying in a plane normal to the texture axis. The pattern, considered as plane sections of the reciprocal-lattice representation, resembles rotation diagrams of single crystals and approximates to the patterns given by cylindrical lattices (characteristic, for example, of tubular crystals).
If the a axis is the texture axis, the hk rods are at distances from the texture axis and from the plane normal to the texture axis (the zero plane b*c*). On rotation, they intersect the plane normal to the incident beam and pass through the texture axis in layer lines at distances from the zero line, while the reflection positions along these lines are defined by their distances from the textures axis (see Fig. 4.3.5.3 ):
The projections of the reciprocal axes on the plane ab of the direct lattice, with indications of the distances B and D of the hk rows from the fibre-texture axes a or [hk]. |
If the texture axis forms an angle with the a axis and with the projection of a* on the plane ab, then The relation between the angles δ, , and the direction [hk] of the texture axis is given by the expression The layer lines with constant h that coincide when are split when according to the sign of k, since then and and defining the reflection positions along the layer line take other values. Such peculiarities have been observed by means of selected-area electron diffraction for tabular particles and linear crystal aggregates of some phyllosilicates in the simple case of γ = π/2 (Gritsaenko, Zvyagin, Boyarskaya, Gorshkov, Samotoin & Frolova, 1969).
When fibres or linear aggregates are deposited on a film (for example, in specimens for high-resolution electron diffraction) with one direction parallel to a plane, they form a texture that is intermediate between lamellar and fibre. The points of the reciprocal lattice are subject to two rotations: around the fibre axis and around the normal to the plane. The first rotation results in circles, the second in spherical bands of different widths, depending on the position of the initial point relative to the texture axis and the zero plane normal to it. The diffraction patterns correspond to oblique plane sections of reciprocal space, and consist of arcs having intensity maxima near their ends; in some cases, the arcs close to form complete circles. In particular, when the particle elongation is in the a direction, the angular range of the arcs decreases with h and increases with k (Zvyagin, 1967).
The above treatment, though general, had layer silicates primarily in view. Texture studies are particularly important for metal specimens that have been subjected to cold work or other treatments; the phenomena and their interpretation occupy several chapters of the book by Barrett & Massalski (1980). Similarly, Kakudo & Kasai (1972) devote much space to texture in polymer specimens, and Guinier (1956) gives a good treatment of the whole subject. The mathematical methods for describing and analysing textures of all types have been described by Bunge (1982; the German edition of 1969 was revised in many places and a few errors were corrected for the English translation).
The calculation of very large numbers of diffracted orders, i.e. more than 100 and often several thousand, requires the multislice procedure. This occurs because, for N diffracted orders, the multislice procedure involves the manipulation of arrays of size N, whereas the scattering matrix or the eigenvalue procedures involve manipulation of arrays of size N by N.
The simplest form of the multislice procedure presumes that the specimen is a parallel-sided plate. The surface normal is usually taken to be the z axis and the crystal structure axes are often chosen or transformed such that the c axis is parallel to z and the a and b axes are in the xy plane. This can often lead to rather unconventional choices for the unit-cell parameters. The maximum tilt of the incident beam from the surface normal is restricted to be of the order of 0.1 rad. For the calculation of wave amplitudes for larger tilts, the structure must be reprojected down an axis close to the incident-beam direction. For simple calculations, other crystal shapes are generally treated by the column approximation, that is the crystal is presumed to consist of columns parallel to the z axis, each column of different height and tilt in order to approximate the desired shape and variation of orientation.
The numerical procedure involves calculation of the transmission function through a thin slice, calculation of the vacuum propagation between centres of neighbouring slices, followed by evaluation in a computer of the iterated equation in order to obtain the scattered wavefunction, , emitted from slice n, i.e. for crystal thickness ; the symbol * indicates the operation `convolution' defined by and is the propagation function in the small-angle approximation between slice n − 1 and slice n over the slice spacing . For simplicity, the equation is given for orthogonal axes and h′′, k′′ are the usually non-integral intercepts of the Laue circle on the reciprocal-space axes in units of (1/a), (1/b). The excitation errors, , can be evaluated using The transmission function for slice n is where F denotes Fourier transformation from real to reciprocal space, and and where W is the beam voltage, v is the relativistic velocity of the electron, c is the velocity of light, and λ is the relativistic wavelength of the electron.
The operation * in (4.3.6.1) is most effectively carried out for large N by the use of the convolution theorem of Fourier transformations. This efficiency presumes that there is available an efficient fast-Fourier-transform subroutine that is suitable for crystallographic computing, that is, that contains the usual crystallographic normalization factors and that can deal with a range of values for h, k that go from negative to positive. Then, where F denotes and denotes where , , and are the sampling intervals in the unit cell. The array sizes used in the calculations of the Fourier transforms are commonly powers of 2 as is required by many fast Fourier subroutines. The array for is usually defined over the central portion of the reserved computer array space in order to avoid oscillation in the Fourier transforms (Gibbs instability). It is usual to carry out a beam calculation in an array of , hence the critical timing interval in a multislice calculation is that interval taken by a fast Fourier transform for 4N coefficients. If the number of beams, N, is such that there is still appreciable intensity being scattered outside the calculation aperture, then it is usually necessary to impose a circular aperture on the calculation in order to prevent the symmetry of the calculation aperture imposing itself on the calculated wavefunction. This is most conveniently achieved by setting all p(h, k) coefficients outside the desired circular aperture to zero.
It is clear that the iterative procedure of (4.3.6.1) means that care must be taken to avoid accumulation of error due to the precision of representation of numbers in the computer that is to be used. Practical experience indicates that a precision of nine significant figures (decimal) is more than adequate for most calculations. A precision of six to seven (decimal) figures (a common 32-bit floating-point representation) is only barely satisfactory. A computer that uses one of the common 64-bit representations (12 to 16 significant figures) is satisfactory even for the largest calculations currently contemplated.
The choice of slice thickness depends upon the maximum value of the projected potential within a slice and upon the validity of separation of the calculation into transmission function and propagation function. The second criterion is not severe and in practice sets an upper limit to slice thickness of about 10 Å. The first criterion depends upon the atomic number of atoms in the trial structure. In practice, the slice thickness will be too large if two atoms of medium to heavy atomic weight are projected onto one another. It is not necessary to take slices less than one atomic diameter for calculations for fast electron (acceleration voltages greater than 50 keV) diffraction or microscopy. If the trial structure is such that the symmetry of the diffraction pattern is not strongly dependent upon the structure of the crystal parallel to the slice normal, then the slices may be all identical and there is no requirement to have a slice thickness related to the periodicity of the structure parallel to the surface normal. This is called the `no upper-layer-line' approximation. If the upper-layer lines are important, then the slice thickness will need to be a discrete fraction of the c axis, and the contents of each slice will need to reflect the actual atomic contents of each slice. Hence, if there were four slices per unit cell, then there would need to be four distinct q(h, k), each taken in the appropriate order as the multislice operation proceeds in thickness.
The multislice procedure has two checks that can be readily performed during a calculation. The first is applied to the transmission function, q(h, k), and involves the evaluation of a unitarity test by calculation of for all h, k, where q* denotes the complex conjugate of q, and δ(h,k) is the Kronecker delta function. The second test can be applied to any calculation for which no phenomenological absorption potential has been used in the evaluation of the q(h, k). In that case, the sum of intensities of all beams at the final thickness should be no less than 0.9, the incident intensity being taken as 1.0. A value of this sum that is less than 0.9 indicates that the number of beams, N, has been insufficient. In some rare cases, the sum can be greater than 1.0; this is usually an indication that the number of beams has been allowed to come very close to the array size used in the convolution procedure. This last result does not occur if the convolution is carried out directly rather than by use of fast-Fourier-transform methods.
A more complete discussion of the multislice procedure can be obtained from Cowley (1975) and Goodman & Moodie (1974). These references are not exhaustive, but rather an indication of particularly useful articles for the novice in this subject.
Bloch waves, familiar in solid-state valence-band theory, arise as the basic wave solutions for a periodic structure. They are thus always implicit and often explicit in dynamical diffraction calculations, whether applied in perfect crystals, in almost perfect crystals with slowly varying defect strain fields or in more general structures that (see Subsection 4.3.6.1) can always, for computations, be treated by periodic continuation.
The Schrödinger wave equation in a periodic structure, can be applied to high-energy, relativistic electron diffraction, taking as the relativistically corrected electron wave number (see Subsection 4.3.1.4). The Fourier coefficients in the expression for the periodic potential are defined at reciprocal-lattice points g by the expression where is the Born scattering amplitude (see Subsection 4.3.1.2) of the jth atom at position in the unit cell of volume and is the Debye–Waller factor.
The simplest solution to (4.3.6.6) is a single Bloch wave, consisting of a linear combination of plane-wave beams coupled by Bragg reflection. In practice, only a limited number of terms N, corresponding to the most strongly excited Bragg beams, is included in (4.3.6.8). Substitution in (4.3.6.6) then yields N simultaneous equations for the wave amplitudes Usually, χ and the two tangential components and are fixed by matching to the incident wave at the crystal entrance surface. then emerges as a root of the determinant of coefficients appearing in (4.3.6.9).
Numerical solution of (4.3.6.9) is considerably simplified (Hirsch, Howie, Nicholson, Pashley & Whelan, 1977a) in cases of transmission high-energy electron diffraction where all the important reciprocal-lattice points lie in the zero-order Laue zone and . The equations then reduce to a standard matrix eigenvalue problem (for which efficient subroutines are widely available): where and is the distance, measured in the z direction, of the reciprocal-lattice point g from the Ewald sphere.
There will in general be N distinct eigenvalues corresponding to N possible values , , each with its eigenfunction defined by N wave amplitudes . The waves are normalized and orthogonal so that In simple transmission geometry, the complete solution for the total coherent wavefunction is Inelastic and thermal-diffuse-scattering processes cause anomalous absorption effects whereby the amplitude of each component Bloch wave decays with depth z in the crystal from its initial value . The decay constant is computed using an imaginary optical potential with Fourier coefficients (for further details of these see Humphreys & Hirsch, 1968, and Subsection 4.3.1.5 and Section 4.3.2). The Bloch-wave, matrix-diagonalization method has been extended to include reciprocal-lattice points in higher-order Laue zones (Jones, Rackham & Steeds, 1977) and, using pseudopotential scattering amplitudes, to the case of low-energy electrons (Pendry, 1974).
The Bloch-wave picture may be compared with other variants of dynamical diffraction theory, which, like the multislice method (Subsection 4.3.6.1), for example, employ plane waves whose amplitudes vary with position in real space and are determined by numerical integration of first-order coupled differential equations. For cases with beams in perfect crystals or in crystals containing localized defects such as stacking faults or small point-defect clusters, the Bloch-wave method offers many advantages, particularly in thicker crystals with t > 1000 Å. For high-resolution image calculations in thin crystals where the periodic continuation process may lead to several hundred diffracted beams, the multislice method is more efficient. For cases of defects with extended strain fields or crystals illuminated at oblique incidence, coupled plane-wave integrations along columns in real space (Howie & Basinski, 1968) can be the most efficient method.
The general advantage of the Bloch-wave method, however, is the picture it affords of wave propagation and scattering in both perfect and imperfect crystals. For this purpose, solutions of equations (4.3.6.9) allow dispersion surfaces to be plotted in k space, covering with several sheets j all the wave points for a given energy E. Thickness fringes and other interference effects then arise because of interference between waves excited at different points . The average current flow at each point is normal to the dispersion surface and anomalous-absorption effects can be understood in terms of the distribution of Bloch-wave current within the unit cell. Detailed study of these effects, and the behaviour of dispersion surfaces as a function of energy, yields accurate data on scattering amplitudes via the critical-voltage effect (see Section 4.3.7). Static crystal defects induce elastic scattering transitions on sheets of the same dispersion surface. Transitions between points on dispersion surfaces of different energies occur because of thermal diffuse scattering, generation of electronic excitations or the emission of radiation by the fast electron. The Bloch-wave picture and the dispersion surface are central to any description of these phenomena. For further information and references, the reader may find it helpful to consult Section 5.2.10 of Volume B (IT B, 2001).
4.3.7. Measurement of structure factors and determination of crystal thickness by electron diffraction
Current advances in quantitative electron diffraction are connected with improved experimental facilities, notably the combination of convergent-beam electron diffraction (CBED) with new detection systems. This is reflected in extended applications of electron diffraction intensities to problems in crystallography, ranging from valence-electron distributions in crystals with small unit cells to structure determination of biological molecules in membranes. The experimental procedures can be seen in relation to the two main principles for measurement of diffracted intensities from crystals:
|
Integrated intensities are not easily defined in the most common type of electron-diffraction pattern, viz the selected-area (SAD) spot pattern. This is due to the combination of dynamical scattering and the orientation and thickness variations usually present within the typically micrometre-size illuminated area. This combination leads to spot pattern intensities that are poorly defined averages over complicated scattering functions of many structure factors. Convergent-beam electron diffraction is a better alternative for intensity measurements, especially for inorganic structures with small-to-moderate unit cells. In CBED, a fine beam is focused within an area of a few hundred ångströms, with a divergence of the order of a tenth of a degree. The diffraction pattern then appears in the form of discs, which are essentially two-dimensional rocking curves from a small illuminated area, within which thickness and orientation can be regarded as constant. These intensity distributions are obtained under well defined conditions and are well suited for comparison with theoretical calculations. The intensity can be recorded either photographically, or with other parallel recording systems, viz YAG screen/CCD camera (Krivanek, Mooney, Fan, Leber & Meyer, 1991) or image plates (Mori, Oikawa & Harada, 1990) – or sequentially by a scanning system. The inelastic background can be removed by an energy filter (Krahl, Pätzold & Swoboda, 1990; Krivanek, Gubbens, Dellby & Meyer, 1991). Detailed intensity profiles in one or two dimensions can then be measured with high precision for low-order reflections from simple structures. But there are limitations also with the CBED technique: the crystal should be fairly perfect within the illuminated area and the unit cell relatively small, so that overlap between discs can be avoided. The current development of electron diffraction is therefore characterized by a wide range of techniques, which extend from the traditional spot pattern to two-dimensional, filtered rocking curves, adapted to the structure problems under study and the specimens that are available.
Spot-pattern intensities are best for thin samples of crystals with light atoms, especially organic and biological materials. Dorset and co-workers (Dorset, Jap, Ho & Glaeser, 1979; Dorset, 1991) have shown how conventional crystallographic techniques (`direct phasing') can be applied in ab initio structure determination of thin organic crystals from spot intensities in projections. Two main complications were treated by them: bending of the crystal and dynamical scattering. Thin crystals will frequently be bent; this will give some integration of the reflection, but may also produce a slight distortion of the structure, as pointed out by Cowley (1961), who proposed a correction formula. The thickness range for which a kinematical approach to intensities is valid was estimated theoretically by Dorset et al. (1979). For organic crystals, they quoted a few hundred ångströms as a limit for kinematical scattering in dense projections at 100 kV.
Radiation damage is a problem, but with low-dose and cryo-techniques, electron-microscopy methods can be applied to many organic crystals, as shown by several recent investigations. Voigt-Martin, Yan, Gilmore, Shankland & Bricogne (1994) collected electron-diffraction intensities from a beam-sensitive dione and constructed a 1.4 Å Fourier map by a direct method based on maximum entropy. Large numbers of electron-diffraction intensities have been collected from biological molecules crystallized in membranes. The structure amplitudes can be combined with phases extracted from high-resolution micrographs, following Henderson Unwin's (1975) early work. Kühlbrandt, Wang & Fujiyoshi (1994) collected about 18 000 amplitudes and 15 000 phases for a protein complex in an electron cryomicroscope operating at 4.2 K (Fujiyoshi et al., 1991). Using these data, they determined the structure from a three-dimensional Fourier map calculated to 3.4 Å resolution. The assumption of kinematical scattering in such studies has been investigated by Spargo (1994), who found the amplitudes to be kinematic within 4% but with somewhat larger deviations for phases.
For inorganic structures, spot-pattern intensities are less useful because of the stronger dynamical interactions, especially in dense zones. Nevertheless, it may be possible to derive a structure and refine parameters from spot-pattern intensities. Andersson (1975) used experimental intensities from selected projections for comparison with dynamical calculations, including an empirical correction factor for orientation spread, in a structure determination of V14O8. Recently, Zou, Sukharev & Hovmöller (1993) combined spot-pattern intensities read from film by the program ELD with image processing of high-resolution micrographs for structure determination of a complex perovskite.
A considerable improvement over the spot pattern has been obtained by the elegant double-precession technique devised by Vincent & Midgley (1994). They programmed scanning coils above and below the specimen in the electron microscope so as to achieve simultaneous precession of the focused incident beam and the diffraction pattern around the optical axis. The net effect is equivalent to a precession of the specimen with a stationary incident beam. Integrated intensities can be obtained from reflections out to a Bragg angle equal to the precession angle for the zeroth Laue zone. In addition, reflections in the first and second Laue zones appear as broad concentric rings. Dynamical effects are reduced appreciably by this procedure, especially in the non-zero Laue zones. The experimental integrated intensities, Ig, must be multiplied with a geometrical factor analogous to the Lorentz factor in X-ray diffraction, viz where nh is the reciprocal spacing between the zeroth and nth layers. The intensities can be used for structure determination by procedures taken over from X-ray crystallography, e.g. the conditional Patterson projections that are used by the Bristol group (Vincent, Bird & Steeds, 1984). The precession method may be seen as intermediate between the spot pattern and the CBED technique. Another intermediate approach was proposed by Goodman (1976) and used later by Olsen, Goodman & Whitfield (1985) in the structure determination of a series of selenides. CBED patterns from thin crystals were taken in dense zones; intensities were measured at corresponding points in the discs, e.g. at the zone-axis position. Structure parameters were determined by fitting the observed intensities to dynamical calculations.
Higher precision and more direct comparisons with dynamical scattering calculations are achieved by measurements of intensity distributions within the CBED discs, i.e. one- or two-dimensional rocking curves. An up-to-date review of these techniques is found in the recent book by Spence & Zuo (1992), where all aspects of the CBED technique, theory and applications are covered, including determination of lattice constants and strains, crystal symmetry, and fault vectors of defects. Refinement of structure factors in crystals with small unit cells are treated in detail. For determination of bond charges, the structure factors (Fourier potentials) should be determined to an accuracy of a few tenths of a percent; calculations must then be based on many-beam dynamical scattering theory, see Chapter 8.8 . Removal of the inelastic background by an energy filter will improve the data considerably; analytical expressions for the inelastic background including multiple-scattering contributions may be an alternative (Marthinsen, Holmestad & Høier, 1994).
Early CBED applications to the determination of structure factors were based on features that can be related to dynamical effects in the two-beam case. Although insufficient for most accurate analyses, the two-beam expression for the intensity profile may be a useful guide. In its standard form, where Ug and sg are Fourier potential and excitation error for the reflection g, k wave number and t thickness. The expression can be rewritten in terms of the eigenvalues γ(i, j) that correspond to the two Bloch-wave branches, i, j: where Note that the minimum separation between the branches i, j or the gap at the dispersion surface is where g is an extinction distance. The two-beam form is often found to be a good approximation to an intensity profile Ig(sg) even when other beams are excited, provided an effective potential , which corresponds to the gap at the dispersion surface, is substituted for Ug. This is suggested by many features in CBED and Kikuchi patterns and borne out by detailed calculations, see e.g. Høier (1972). Approximate expressions for have been developed along different lines; the best known is the Bethe potential Other perturbation approaches are based on scattering between Bloch waves, in analogy with the `interband scattering' introduced by Howie (1963) for diffuse scattering; the term `Bloch-wave hybridization' was introduced by Buxton (1976). Exact treatment of symmetrical few-beam cases is possible (see Fukuhara, 1966; Kogiso & Takahashi, 1977). The three-beam case (Kambe, 1957; Gjønnes & Høier, 1971) is described in detail in the book by Spence & Zuo (1992).
Many intensity features can be related to the structure of the dispersion surface, as represented by the function γ(kx, ky). The gap [equation (4.3.7.4)] is an important parameter, as in the four-beam symmetrical case in Fig. 4.3.7.1 . Intensity measurements along one dimension can then be referred to three groups, according to the width of the gap, viz:
|
A small gap at the dispersion surface implies that the two-beam-like rocking curve above approaches a kinematical form and can be represented by an integrated intensity. Within a certain thickness range, this intensity may be proportional to , with an angular width inversely proportional to gt. Several schemes have been proposed for measurement of relative integrated intensities for reflections in the outer, high-angle region, where the lines are narrow and can be easily separated from the background. Steeds (1984) proposed use of the HOLZ (high-order Laue-zone) lines, which appear in CBED patterns taken with the central disc at the zone-axis position. Along a ring that defines the first-order Laue zone (FOLZ), reflections appear as segments that can be associated with scattering from strongly excited Bloch waves in the central ZOLZ part into the FOLZ reflections. Vincent, Bird & Steeds (1984) proposed an intensity expression for integrated intensity for a line segment associated with scattering from (or into) the ZOLZ Bloch wave j. is here the excitation coefficient and β(j) the matrix element for scattering between the Bloch wave j and the plane wave g. μ(j) and μ are absorption coefficients for the Bloch wave and plane wave, respectively; t is the thickness. From measurements of a number of such FOLZ (or SOLZ) reflections, they were able to carry out ab initio structure determinations using so-called conditional Patterson projections and coordinate refinement. Tanaka & Tsuda (1990) have refined atomic positions from zone-axis HOLZ intensities. Ratios between HOLZ intensities have been used for determination of the Debye–Waller factor (Holmestad, Weickenmeier, Zuo, Spence & Horita, 1993).
Another CBED approach to integrated intensities is due to Taftø & Metzger (1985). They measured a set of high-order reflections along a systematic row with a wide-aperture CBED tilted off symmetrical incidence. A number of high-order reflections are then simultaneously excited in a range where the reflections are narrow and do not overlap. Gjønnes & Bøe (1994) and Ma, Rømming, Lebech & Gjønnes (1992) applied the technique to the refinement of coordinates and thermal parameters in high-Tc superconductors and intermetallic compounds. The validity and limitation of the kinematical approximation and dynamic potentials in this case has been discussed by Gjønnes & Bøe (1994).
Zero gap at the dispersion surface corresponds to zero effective Fourier potential or, to be more exact, an accidental degeneracy, γ(i) = γ(j), in the Bloch-wave solution. This is the basis for the critical-voltage method first shown by Watanabe, Uyeda & Fukuhara (1969). From vanishing contrast of the Kikuchi line corresponding to a second-order reflection 2g, they determined a relation between the structure factors Ug and U2g. Gjønnes & Høier (1971) derived the condition for the accidental degeneracy in the general centrosymmetrical three-beam case 0,g,h, expressed in terms of the excitation errors sg,h and Fourier potentials Ug,h,g−h, viz where m and m0 are the relativistic and rest mass of the incident electron. Experimentally, this condition is obtained at a particular voltage and diffraction condition as vanishing line contrast of a Kikuchi or Kossel line – or as a reversal of a contrast feature. The second-order critical-voltage effect is then obtained as a special case, e.g. by the mass ratio: Measurements have been carried out for a number of elements and alloy phases; see the review by Fox & Fisher (1988) and later work on alloys by Fox & Tabbernor (1991). Zone-axis critical voltages have been used by Matsuhata & Steeds (1987). For analytical expressions and experimental determination of non-systematic critical voltages, see Matsuhata & Gjønnes (1994).
Large gaps at the dispersion surface are associated with strong inner reflections – and a strong dynamical effect of two-beam-like character. The absolute magnitude of the gap – or its inverse, the extinction distance – can be obtained in different ways. Early measurements were based on the split of diffraction spots from a wedge, see Lehmpfuhl (1974), or the corresponding fringe periods measured in bright- and dark-field micrographs (Ando, Ichimiya & Uyeda, 1974). The most precise and applicable large-gap methods are based on the refinement of the fringe pattern in CBED discs from strong reflections, as developed by Goodman & Lehmpfuhl (1967) and Voss, Lehmpfuhl & Smith (1980). In recent years, this technique has been developed to high perfection by means of filtered CBED patterns, see Spence & Zuo (1992) and papers referred to therein. See also Chapter 8.8 .
The gap at the dispersion surface can also be obtained directly from the split observed at the crossing of a weak Kikuchi line with a strong band. Gjønnes & Høier (1971) showed how this can be used to determine strong low-order reflections. High voltage may improve the accuracy (Terasaki, Watanabe & Gjønnes, 1979). The sensitivity of the intersecting Kikuchi-line (IKL) method was further increased by the use of CBED instead of Kikuchi patterns (Matsuhata, Tomokiyo, Watanabe & Eguchi, 1984; Taftø & Gjønnes, 1985). In a recent development, Høier, Bakken, Marthinsen & Holmestad (1993) have measured the intensity distribution in the CBED discs around such intersections and have refined the main structure factors involved.
Two-dimensional rocking curves collected by CBED patterns around the axis of a dense zone are complicated by extensive many-beam dynamical interactions. The Bristol–Bath group (Saunders, Bird, Midgley & Vincent, 1994) claim that the strong dynamic effects can be exploited to yield high sensitivity in refinement of low-order structure factors. They have also developed procedures for ab initio structure determination based on zone-axis patterns (Bird & Saunders, 1992), see Chapter 8.8 .
Determination of phase invariants. It has been known for some time (e.g. Kambe, 1957) that the dynamical three-beam case contains information about phase. As in the X-ray case, measurement of dynamical effects can be used to determine the value of triplets (Zuo, Høier & Spence, 1989) and to determine phase angles to better than one tenth of a degree (Zuo, Spence, Downs & Mayer, 1993) which is far better than any X-ray method. Bird (1990) has pointed out that the phase of the absorption potential may differ from the phase of the real potential.
Thickness is an important parameter in electron-diffraction experiments. In structure-factor determination based on CBED patterns, thickness is often included in the refinement. Thickness can also be determined directly from profiles connected with large gaps at the dispersion surface (Goodman & Lehmpfuhl, 1967; Blake, Jostsons, Kelly & Napier, 1978; Glazer, Ramesh, Hilton, & Sarikaya, 1985). The method is based on the outer part of the fringe profile, which is not so sensitive to the structure factor. The intensity minimum of the ith fringe in the diffracted disc occurs at a position corresponding to the excitation error si and expressed as where ni is a small integer describing the order of the minimum. This equation can be arranged in two ways for graphic determination of thickness. The commonest method appears to be to plot (si/ni)2 against 1/ni2and then determine the thickness from the intersection with the ordinate axis (Kelly, Jostsons, Blake & Napier, 1975). Glazer et al. (1985) claim that the method originally proposed by Ackermann (1948), where is plotted against ni and the thickness is taken from the slope, is more accurate. In both cases, the outer part of the rocking curve is emphasized; exact knowledge of the gap is not necessary for a good determination of thickness, provided the assumption of a two-beam-like rocking curve is valid.
For the crystallographic study of real materials, high-resolution electron microscopy (HREM) can provide a great deal of information that is complementary to that obtainable by X-ray and neutron diffraction methods. In contrast to the statistically averaged information that these other methods provide, the great power of HREM lies in its ability to elucidate the detailed atomic arrangements of individual defects and the microcrystalline structure in real crystals. The defects and inhomogeneities of real crystals frequently exert a controlling influence on phase-transition mechanisms and more generally on all the electrical, mechanical, and thermal properties of solids. The real-space images that HREM provides (such as that shown in Fig. 4.3.8.1 ) can give an immediate and dramatic impression of chemical crystallography processes, unobtainable by other methods. Their atomic structure is of the utmost importance for an understanding of the properties of real materials. The HREM method has proven powerful for the determination of the structure of such defects and of the submicrometre-sized microcrystals that constitute many polyphase materials.
Atomic resolution image of a tantalum-doped tungsten trioxide crystal (pseudo-cubic structure) showing extended crystallographic shear-plane defects (C), pentagonal-column hexagonal-tunnel (PCHT) defects (T), and metallization of the surface due to oxygen desorption (JEOL 4000EX, crystal thickness less than 200 Å, 400 kV, Cs = 1 mm). Atomic columns are black. [Smith, Bursill & Wood (1985).] |
In summary, HREM should be considered the technique of choice where a knowledge of microcrystal size, shape or morphology is required. In addition, it can be used to reveal the presence of line and planar defects, inclusions, grain boundaries and phase boundaries, and, in favourable cases, to determine atomic structure. Surface atomic structure and reconstruction have also been studied by HREM. However, meaningful results in this field require accurately controlled ultra-high-vacuum conditions. The determination of the atomic structure of point defects by HREM so far has proven extremely difficult, but this situation is likely to change in the near future.
The following sections are not intended to review the applications of HREM, but rather to provide a summary of the main theoretical results of proven usefulness in the field, a selected bibliography, and recommendations for good experimental practice. At the time of writing (1997), the point resolution of HREM machines lies between 1 and 2 Å.
The function of the objective lens in an electron microscope is to perform a Fourier synthesis of the Bragg-diffracted electron beams scattered (in transmission) by a thin crystal, in order to produce a real-space electron image in the plane r. This electron image intensity can be written where represents the complex amplitude of the diffracted wave after diffraction in the crystal as a function of the reciprocal-lattice vector u [magnitude in the plane perpendicular to the beam, so that the wavevector of an incident plane wave is written . Following the convention of Section 2.5.2 in IT B (2001), we write . The function χ(u) is the phase factor for the objective-lens transfer function and P(u) describes the effect of the objective aperture:
For a periodic object, the image wavefunction is given by summing the contributions from the set of reciprocal-lattice points, g, so that For atomic resolution, with 1 Å−1, it is apparent that, for all but the simplest structures and smallest unit cells, this synthesis will involve many hundreds of Bragg beams. A scattering calculation must involve an even larger number of beams than those that contribute resolvable detail to the image, since, as described in Section 2.5.2 in IT B (2001), all beams interact strongly through multiple coherent scattering. The theoretical basis for HREM image interpretation is therefore the dynamical theory of electron diffraction in the transmission (or Laue) geometry [see Chapter 5.2 in IT B (2001)]. The resolution of HREM images is limited by the aberrations of the objective electron lens (notably spherical aberration) and by electronic instabilities. An intuitive understanding of the complicated effect of these factors on image formation from multiply scattered Bragg beams is generally not possible. To provide a basis for understanding, therefore, the following section treats the simplified case of few-beam `lattice-fringe' images, in order to expose the relationship between the crystal potential, its structure factors, electron-lens aberrations, and the electron image.
Image formation in the transmission electron microscope is conventionally treated by analogy with the Abbe theory of coherent optical imaging. The overall process is subdivided as follows. (a) The problem of beam–specimen interaction for a collimated kilovolt electron beam traversing a thin parallel-sided slab of crystal in a given orientation. The solution to this problem gives the elastically scattered dynamical electron wavefunction , where r is a two-dimensional vector lying in the downstream surface of the slab. Computer algorithms for dynamical scattering are described in Section 4.3.6. (b) The effects of the objective lens are incorporated by multiplying the Fourier transform of by a function T(u), which describes both the wavefront aberration of the lens and the diffraction-limiting effects of any apertures. The dominant aberrations are spherical aberration, astigmatism, and defect of focus. The image intensity is then formed from the modulus squared of the Fourier transform of this product. (c) All partial coherence effects may be incorporated by repeating this procedure for each of the component energies and directions that make up the illumination from an extended electron source, and summing the resulting intensities. Because this procedure requires a separate dynamical calculation for each component direction of the incident beam, a number of useful approximations of restricted validity have been developed; these are described in Subsection 4.3.8.4. This treatment of partial coherence assumes that a perfectly incoherent effective source can be identified. For field-emission HREM instruments, a coherent sum (over directions) of complex image wavefunctions may be required.
General treatments of the subject of HREM can be found in the texts by Cowley (1981) and Spence (1988b). The sign conventions used throughout the following are consistent with the standard crystallographic convention of Section 2.5.2 of IT B (2001), which assumes a plane wave of form and so is consistent with X-ray usage.
We consider few-beam lattice images, in order to understand the effects of instrumental factors on electron images, and to expose the conditions under which they faithfully represent the scattering object. The case of two-beam lattice images is instructive and contains, in simplified form, most of the features seen in more complicated many-beam images. These fringes were first observed by Menter (1956) and further studied in the pioneering work of Komoda (1964) and others [see Spence (1988b) for references to early work]. The electron-microscope optic-axis orientation, the electron beam, and the crystal setting are indicated in Fig. 4.3.8.2 . If an objective aperture is used that excludes all but the two beams shown from contributing to the image, equation (4.3.8.2) gives the image intensity along direction g for a centrosymmetric crystal of thickness t as
Imaging conditions for few-beam lattice images. For three-beam axial imaging shown in (c), the formation of half-period fringes is also shown. |
The Bragg-diffracted beams have complex amplitudes . The lattice-plane period is in direction g [Miller indices (hkl)]. The lens-aberration phase function, including only the effects of defocus Δf and spherical aberration (coefficient ), is given by The effects of astigmatism and higher-order aberrations have been ignored. The defocus, Δf, is negative for the objective lens weakened (i.e. the focal length increased, giving a bright first Fresnel-edge fringe). The magnitude of the reciprocal-lattice vector , where is the Bragg angle. If these two Bragg beams were the only beams excited in the crystal (a poor approximation for quantitative work), their amplitudes would be given by the `two-beam' dynamical theory of electron diffraction as where is the two-beam extinction distance, is a Fourier coefficient of crystal potential, is the excitation error (see Fig. 4.3.8.2), , and the interaction parameter σ is defined in Section 2.5.2 of IT B (2001).
The two-beam image intensity given by equation (4.3.8.3) therefore depends on the parameters of crystal thickness (t), orientation , structure factor , objective-lens defocus Δf, and spherical-aberration constant . We consider first the variation of lattice fringes with crystal thickness in the two-beam approximation (Cowley, 1959; Hashimoto, Mannami & Naiki, 1961). At the exact Bragg condition , equations (4.3.8.5) and (4.3.8.3) give If we consider a wedge-shaped crystal with the electron beam approximately normal to the wedge surface and edge, and take x and g parallel to the edge, this equation shows that sinusoidal lattice fringes are expected whose contrast falls to zero (and reverses sign) at thicknesses of . This apparent abrupt translation of fringes (by d/2 in the direction x) at particular thicknesses is also seen in some experimental many-beam images. The effect of changes in focus (due perhaps to variations in lens current) is seen to result in a translation of the fringes (in direction x), while time-dependent variations in the accelerating voltage have a similar effect. Hence, time-dependent variations of the lens focal length or the accelerating voltage result in reduced image contrast (see below). If the illumination makes a small angle with the optic axis, the intensity becomes For a uniformly intense line source subtending a semiangle , the total lattice-fringe intensity is The resulting fringe visibility is proportional to , where . The contrast falls to zero for β = π, so that the range of focus over which fringes are expected is . This is the approximate depth of field for lattice images due to the effects of the finite source size alone.
The case of three-beam fringes in the axial orientation is of more practical importance [see Fig. 4.3.8.2(b)]. The image intensity for and is The lattice image is seen to consist of a constant background plus cosine fringes with the lattice spacing, together with cosine fringes of half this spacing. The contribution of the half-spacing fringes is independent of instrumental parameters (and therefore of electronic instabilities if ). These fringes constitute an important HREM image artifact. For kinematic scattering, and only the half-period fringes will then be seen if , or for focus settings Fig. 4.3.8.2(c) indicates the form of the fringes expected for two focus settings with differing half-period contributions. As in the case of two-beam fringes, dynamical scattering may cause to be severely attenuated at certain thicknesses, resulting also in a strong half-period contribution to the image.
Changes of 2π in in equation (4.3.8.7) leave I(x, t) unchanged. Thus, changes of defocus by amounts or changes in by yield identical images. The images are thus periodic in both Δf and . This is a restricted example of the more general phenomenon of n-beam Fourier imaging discussed in Subsection 4.3.8.3.
We note that only a single Fourier period will be seen if is less than the depth of field Δz. This leads to the approximate condition , which, when combined with the Bragg law, indicates that a single period only of images will be seen when adjacent diffraction discs just overlap.
The axial three-beam fringes will coincide with the lattice planes, and show atom positions as dark if and . This total phase shift of −π between and the scattered beams is the desirable imaging condition for phase contrast, giving rise to dark atom positions on a bright background. This requires as a condition for identical axial three-beam lattice images for . This family of lines has been plotted in Fig. 4.3.8.3 for the (111) planes of silicon. Dashed lines denote the locus of `white-atom' images (reversed contrast fringes), while the dotted lines indicate half-period images. In practice, the depth of field is limited by the finite illumination aperture , and few-beam lattice-image contrast will be a maximum at the stationary-phase focus setting, given by
A summary of three- (or five-) beam axial imaging conditions. Here, Δff is the Fourier image period, Δ f0 the stationary-phase focus, Cs(0) the image period in Cs, and a scattering phase of −π/2 is assumed. The lines are drawn for the (111) planes of silicon at 100 kV with θc = 1.4 mrad. |
This choice of focus ensures for , and thus ensures the most favourable trade-off between increasing and loss of fringe contrast for lattice planes g. Note that is not equal to the Scherzer focus (see below). This focus setting is also indicated on Fig. 4.3.8.3, and indicates the instrumental conditions which produce the most intense (111) three- (or five-) beam axial fringes in silicon. For three-beam axial fringes of spacing d, it can be shown that the depth of field is approximately This depth of field, within which strong fringes will be seen, is indicated as a boundary on Fig. 4.3.8.3. Thus, the finer the image detail, the smaller is the focal range over which it may be observed, for a given illumination aperture .
Fig. 4.3.8.4 shows an exact dynamical calculation for the contrast of three-beam axial fringes as a function of Δf in the neighbourhood of . Both reversed contrast and half-period fringes are noted. The effects of electronic instabilities on lattice images are discussed in Subsection 4.3.8.3. It is assumed above that is sufficiently small to allow the neglect of any changes in diffraction conditions (Ewald-sphere orientation) within . Under a similar approximation but without the approximations of transfer theory, Desseaux, Renault & Bourret (1977) have analysed the effect of beam divergence on two-dimensional five-beam axial lattice fringes.
The contrast of few-beam lattice images as a function of focus in the neighbourhood of the stationary-phase focus [see Olsen & Spence (1981)]. |
When two-dimensional patterns of fringes are considered, the Fourier imaging conditions become more complex (see Subsection 4.3.8.3), but half-period fringe systems and reversed-contrast images are still seen. For example, in a cubic projection, a focus change of results in an image shifted by half a unit cell along the cell diagonal. It is readily shown that if when n, m are integers and a and b are the two dimensions of any orthogonal unit cell that can be chosen for . Thus, changes in focus by produce identical images in crystals for which such a cell can be chosen, regardless of the number of beams contributing (Cowley & Moodie, 1960).
For closed-form expressions for the few-beam (up to 10 beams) two-dimensional dynamical Bragg-beam amplitudes in orientations of high symmetry, the reader is referred to the work of Fukuhara (1966).
We define a crystal structure image as a high-resolution electron micrograph that faithfully represents a projection of a crystal structure to some limited resolution, and which was obtained using instrumental conditions that are independent of the structure, and so require no a priori knowledge of the structure. The resolution of these images is discussed in Subsection 4.3.8.6, and their variation with instrumental parameters in Subsection 4.3.8.4.
Equation (4.3.8.2) must now be modified to take account of the finite electron source size used and of the effects of the range of energies present in the electron beam. For a perfect crystal we may write, as in equation (2.5.2.36) in IT B (2001), for the total image intensity due to an electron source whose normalized distribution of wavevectors is , where has components , and which extends over a range of energies corresponding to the distribution of focus . If χ is also assumed to vary linearly across and changes in the diffraction conditions over this range are assumed to make only negligible changes in the diffracted-beam amplitude , the expression for a Fourier coefficient of the total image intensity becomes where γ(h) and β(g) are the Fourier transforms of and , respectively.
For the imaging of very thin crystals, and particularly for the case of defects in crystals, which are frequently the objects of particular interest, we give here some useful approximations for HREM structure images in terms of the continuous projected crystal potential where the projection is taken in the electron-beam direction. A brief summary of the use of these approximations is included in Section 2.5.2 of IT B (2001) and computing methods are discussed in Subsection 4.3.8.5 and Section 4.3.4.
The projected-charge-density (PCD) approximation (Cowley & Moodie, 1960) gives the HREM image intensity (for the simplified case where ) as where is the projected charge density for the specimen (including the nuclear contribution) and is related to through Poisson's equation. Here, is the specimen dielectric constant. This approximation, unlike the weak-phase-object approximation (WPO), includes multiple scattering to all orders of the Born series, within the approximation that the component of the scattering vector is zero in the beam direction (a `flat' Ewald sphere). Contrast is found to be proportional to defocus and to . The failure conditions of this approximation are discussed by Lynch, Moodie & O'Keefe (1975); briefly, it fails for (and hence if , Δf or becomes large) or for large thicknesses t (t 7 nm is suggested for specimens of medium atomic weight and λ = 0.037 Å). The PCD result becomes increasingly accurate with increasing accelerating voltage for small .
The WPO approximation has been used extensively in combination with the Scherzer-focus condition (Scherzer, 1949) for the interpretation of structure images (Cowley & Iijima, 1972). This approximation neglects multiple scattering of the beam electron and thereby allows the application of the methods of linear transfer theory from optics. The image intensity is then given, for plane-wave illumination, by where denotes Fourier transform, * denotes convolution, and u and v are orthogonal components of the two-dimensional scattering vector u. The function S(x, y) is sharply peaked and negative at the `Scherzer focus'and the optimum objective aperture size It forms the impulse response of an electron microscope for phase contrast. Contrast is found to be proportional to and to the interaction parameter σ, which increases very slowly with accelerating voltage above about 500 keV. The point resolution [see Subsection 2.5.2.9 of IT B (2001) and Subsection 4.3.8.6] is conventionally defined from equation (4.3.8.15b) as , or
The occurrence of appreciable multiple scattering, and therefore of the failure of the WPO approximation, depends on specimen thickness, orientation, and accelerating voltage. Detailed comparisons between accurate multiple-scattering calculations, the PCD approximation, and the WPO approximation can be found in Lynch, Moodie & O'Keefe (1975) and Jap & Glaeser (1978). As a very rough guide, equation (4.3.8.14) can be expected to fail for light elements at 100 keV and thicknesses greater than about 5.0 nm. Multiple-scattering effects have been predicted within single atoms of gold at 100 keV.
The WPO approximation may be extended to include the effects of an extended source (partial spatial coherence) and a range of incident electron-beam energies (temporal coherence). General methods for incorporating these effects in the presence of multiple scattering are described in Subsection 4.3.8.5. Under the approximations of linear imaging outlined below, it can be shown (Wade & Frank, 1977; Fejes, 1977) that in equation (4.3.8.14) may be replaced by if astigmatism is absent. Here, and . In addition, is the Fourier transform of the source intensity distribution (assumed Gaussian), so that is small in regions where the slope of is large, resulting in severe attenuation of these spatial frequencies. If the illuminating beam divergence is chosen as the angular half width for which the distribution of source intensity falls to half its maximum value, then The quantity q is defined by where T2 expresses a coupling between the effects of partial spatial coherence and temporal coherence. This term can frequently be neglected under HREM conditions [see Wade & Frank (1977) for details]. The damping envelope due to chromatic effects is described by the parameter where and are the variances in the statistically independent fluctuations of accelerating voltage and objective-lens current . The r.m.s. value of the high voltage fluctuation is equal to the standard deviation . The full width at half-maximum height of the energy distribution of electrons leaving the filament is Here, is the chromatic aberration constant of the objective lens.
Equations (4.3.8.14) and (4.3.8.17) indicate that under linear imaging conditions the transfer function for HREM contains a chromatic damping envelope more severely attenuating than a Gaussian of width which is present in the absence of any objective aperture P(u). The resulting resolution limit is known as the information resolution limit (see Subsection 4.3.8.6) and depends on electronic instabilities and the thermal-energy spread of electrons leaving the filament. The reduction in the contribution of particular diffracted beams to the image due to limited spatial coherence is minimized over those extended regions for which is small, called passbands, which occur when The Scherzer focus corresponds to n = 0. These passbands become narrower and move to higher u values with increasing n, but are subject also to chromatic damping effects. The passbands occur between spatial frequencies and , where Their use for extracting information beyond the point resolution of an electron microscope is further discussed in Subsection 4.3.8.6.
Fig. 4.3.8.5 shows transfer functions for a modern instrument for n = 0 and 1. Equations (4.3.8.14) and (4.3.8.17) provide a simple, useful, and popular approach to the interpretation of HREM images and valuable insights into resolution-limiting factors. However, it must be emphasized that these results apply only (amongst other conditions) for (in crystals) and therefore do not apply to the usual case of strong multiple electron scattering. Equation (4.3.8.13b) does not make this approximation. In real space, for crystals, the alignment of columns of atoms in the beam direction rapidly leads to phase changes in the electron wavefunction that exceed π/2, leading to the failure of equation (4.3.8.14). Accurate quantitative comparisons of experimental and simulated HREM images must be based on equation (4.3.8.13a), or possibly (4.3.8.13b), with obtained from many-beam dynamical calculations of the type described in Subsection 4.3.8.5.
(a) The transfer function for a 400 kV electron microscope with a point resolution of 1.7 Å at the Scherzer focus; the curve is based on equation (4.3.8.17). In (b) is shown a transfer function for similar conditions at the first `passband' focus [n = 1 in equation (4.3.8.22)]. |
For the structure imaging of specific types of defects and materials, the following references are relevant. (i) For line defects viewed parallel to the line, d'Anterroches & Bourret (1984); viewed normal to the line, Alexander, Spence, Shindo, Gottschalk & Long (1986). (ii) For problems of variable lattice spacing (e.g. spinodal decomposition), Cockayne & Gronsky (1981). (iii) For point defects and their ordering, in tunnel structures, Yagi & Cowley (1978); in semiconductors, Zakharov, Pasemann & Rozhanski (1982); in metals, Fields & Cowley (1978). (iv) For interfaces, see the proceedings reported in Ultramicroscopy (1992), Vol. 40, No. 3. (v) For metals, Lovey, Coene, Van Dyck, Van Tendeloo, Van Landuyt & Amelinckx (1984). (vi) For organic crystals, Kobayashi, Fujiyoshi & Uyeda (1982). (vii) For a general review of applications in solid-state chemistry, see the collection of papers reported in Ultramicroscopy (1985), Vol. 18, Nos. 1–4. (viii) Radiation-damage effects are observed at atomic resolution by Horiuchi (1982).
The instrumental parameters that affect HREM images include accelerating voltage, astigmatism, optic-axis alignment, focus setting Δf, spherical-aberration constant , beam divergence , and chromatic aberration constant . Crystal parameters influencing HREM images include thickness, absorption, ionicity, and the alignment of the crystal zone axis with the beam, in addition to the structure factors and atom positions of the sample. The accurate measurement of electron wavelength or accelerating voltage has been discussed by many workers, including Uyeda, Hoier and others [see Fitzgerald & Johnson (1984) for references]. The measurement of Kikuchi-line spacings from crystals of known structure appears to be the most accurate and convenient method for HREM work, and allows an overall accuracy of better than 0.2% in accelerating voltage. Fluctuations in accelerating voltage contribute to the chromatic damping term Δ in equation (4.3.8.19) through the variance . With the trend toward the use of higher accelerating voltages for HREM work, this term has become especially significant for the consideration of the information resolution limit [equation (4.3.8.21)].
Techniques for the accurate measurement of astigmatism and chromatic aberration are described by Spence (1988b). The displacement of images of small crystals with beam tilt may be used to measure ; alternatively, the curvature of higher-order Laue-zone lines in CBED patterns has been used. The method of Budinger & Glaeser (1976) uses a similar dark-field image-displacement method to provide values for both Δf and , and appears to be the most convenient and accurate for HREM work. The analysis of optical diffractograms initiated by Thon and co-workers from HREM images of thin amorphous films provides an invaluable diagnostic aid for HREM work; however, the determination of by this method is prone to large errors, especially at small defocus. Diffractograms provide a rapid method for the determination of focus setting (see Krivanek, 1976) and in addition provide a sensitive indicator of specimen movement, astigmatism, and the damping-envelope constants Δ and .
Misalignment of the electron beam , optic axis, and crystal axis in bright-field HREM work becomes increasingly important with increasing resolution and specimen thickness. The first-order effects of optical misalignment are an artifactual translation of spatial frequencies in the direction of misalignment by an amount proportional to the misalignment and to the square of spatial frequency. The corresponding phase shift is not observable in diffractograms. The effects of astigmatism on transfer functions for inclined illumination are discussed in Saxton (1978).
The effects of misalignment of the beam with respect to the optic axis are discussed in detail by Smith, Saxton, O'Keefe, Wood & Stobbs (1983), where it is found that all symmetry elements (except a mirror plane along the tilt direction) may be destroyed by misalignment. The maximum allowable misalignment for a given resolution δ in a specimen of thickness t is proportional to Misalignment of a crystalline specimen with respect to the beam may be distinguished from misalignment of the optic axis with respect to the beam by the fact that, in very thin crystals, the former does not destroy centres of symmetry in the image.
The use of known defect point-group symmetry (for example in stacking faults) to identify a point in a HREM image with a point in the structure and so to resolve the black or white atomic contrast ambiguity has been described (Olsen & Spence, 1981). Structures containing screw or glide elements normal to the beam are particularly sensitive to misalignment, and errors as small as 0.2 mrad may substantially alter the image appearance.
A rapid comparison of images of amorphous material with the beam electronically tilted into several directions appears to be the best current method of aligning the beam with the optic axis, while switching to convergent-beam mode appears to be the most effective method of aligning the beam with the crystal axis. However, there is evidence that the angle of incidence of the incident beam is altered by this switching procedure.
The effects of misalignment and choice of beam divergence on HREM images of crystals containing dynamically forbidden reflections are reviewed by Nagakura, Nakamura & Suzuki (1982) and Smith, Bursill & Wood (1985). Here the dramatic example of rutile in the [001] orientation is used to demonstrate how a misalignment of less than 0.2 mrad of the electron beam with respect to the crystal axis can bring up a coarse set of fringes (4.6 Å), which produce an image of incorrect symmetry, since these correspond to structure factors that are forbidden both dynamically and kinematically.
Crystal thickness is most accurately determined from images of planar faults in known orientations, or from crystal morphology for small particles. It must otherwise be treated as a refinement parameter. Since small crystals (such as MgO smoke particles, which form as perfect cubes) provide such an independent method of thickness determination, they provide the most convincing test of dynamical imaging theory. The ability to match the contrast reversals and other detailed changes in HREM images as a function of either thickness or focus (or both) where these parameters have been measured by an independent method gives the greatest confidence in image interpretation. This approach, which has been applied in rather few cases [see, for example, O'Keefe, Spence, Hutchinson & Waddington (1985)] is strongly recommended. The tendency for n-beam dynamical HREM images to repeat with increasing thickness in cases where the wavefunction is dominated by just two Bloch waves has been analysed by several workers (Kambe, 1982).
Since electron scattering factors are proportional to the difference between atomic number and X-ray scattering factors, and inversely proportional to the square of the scattering angle (see Section 4.3.1), it has been known for many years that the low-order reflections that contribute to HREM images are extremely sensitive to the distribution of bonding electrons and so to the degree of ionicity of the species imaged. This observation has formed the basis of several charge-density-map determinations by convergent-beam electron diffraction [see, for example, Zuo, Spence & O'Keefe (1988)]. Studies of ionicity effects on HREM imaging can be found in Anstis, Lynch, Moodie & O'Keefe (1973) and Fujiyoshi, Ishizuka, Tsuji, Kobayashi & Uyeda (1983).
The depletion of the elastic portion of the dynamical electron wavefunction by inelastic crystal excitations (chiefly phonons, single-electron excitations, and plasmons) may have dramatic effects on the HREM images of thicker crystals (Pirouz, 1974). For image formation by the elastic component, these effects may be described through the use of a complex `optical' potential and the appropriate Debye–Waller factor (see Section 2.5.1 ). However, existing calculations for the absorption coefficients derived from the imaginary part of this potential are frequently not applicable to lattice images because of the large objective apertures used in HREM work. It has been suggested that HREM images formed from electrons that suffer small energy losses (and so remain `in focus') but large-angle scattering events (within the objective aperture) due to phonon excitation may contribute high-resolution detail to images (Cowley, 1988). For measurements of the imaginary part of the optical potential by electron diffraction, the reader is referred to the work of Voss, Lehmpfuhl & Smith (1980), and references therein. All evidence suggests, however, that for the crystal thicknesses generally used for HREM work ( 200 Å) the effects of `absorption' are small.
In summary, the general approach to the matching of computed and experimental HREM images proceeds as follows (Wilson, Spargo & Smith, 1982). (i) Values of Δ, , and are determined by careful measurements under well defined conditions (electron-gun bias setting, illumination aperture size, specimen height as measured by focusing-lens currents, electron-source size, etc). These parameters are then taken as constants for all subsequent work under these instrumental conditions (assuming also continuous monitoring of electronic instabilities). (ii) For a particular structure refinement, the parameters of thickness and focus are then varied, together with the choice of atomic model, in dynamical computer simulations until agreement is obtained. Every effort should be made to match images as a function of thickness and focus. (iii) If agreement cannot be obtained, the effects of small misalignments must be investigated (Smith et al., 1985). Crystals most sensitive to these include those containing reflections that are absent due to the presence of screw or glide elements normal to the beam.
The general formulations for the dynamical theory of electron diffraction in crystals have been described in Chapter 5.2 of IT B (2001). In Section 4.3.6, the computing methods used for calculating diffraction-beam amplitudes have been outlined.
Given the diffracted-beam amplitudes, , the image is calculated by use of equations (4.3.8.2), including, when appropriate, the modifications of (4.3.8.13b).
The numerical methods that can be employed in relation to crystal-structure imaging make use of algorithms based on (i) matrix diagonalization, (ii) fast Fourier transforms, (iii) real-space convolution (Van Dyck, 1980), (iv) Runge-Kutta (or similar) methods, or (v) power-series evaluation. Two other solutions, the Cowley–Moodie polynomial solution and the Feynman path-integral solution, have not been used extensively for numerical work. Methods (i) and (ii) have proven the most popular, with (ii) (the multislice method) being used most extensively for HREM image simulations. The availability of inexpensive array processors has made this technique highly efficient. A comparison of these two N-beam methods is given by Self, O'Keefe, Buseck & Spargo (1983), who find the multislice method to be faster (time proportional to ) than the diagonalization method (time proportional to ) for N 16. Computing space increases roughly as for the diagonalization method, and as N for the multislice. The problem of steeply inclined boundary conditions for multislice computations has been discussed by Ishizuka (1982).
In the Bloch-wave formulation, the lattice image is given by where and are the eigenvector elements and eigenvalues of the structure matrix [see Hirsch, Howie, Nicholson, Pashley & Whelan (1977a) and Section 4.3.4].
Using modern personal computers or workstations, it is now possible to build efficient single-user systems that allow interactive dynamical structure-image calculations. Either an image intensifier or a cooled scientific grade charge-coupled device and single-crystal scintillator screen may be used to record the images, which are then transferred into a computer (Daberkow, Herrmann, Liu & Rau, 1991). This then allows for the possibility of automated alignment, stigmation and focusing to the level of accuracy needed at 0.1 nm point resolution (Krivanek & Mooney, 1993). An image-matching search through trial structures, thickness and focus parameters can then be completed rapidly. Where large numbers of pixels, large dynamic range and high sensitivity are required, the Image Plate has definite advantages and so should find application in electron holography and biology (Shindo, Hiraga, Oikawa & Mori, 1990).
For the calculation of images of defects, the method of periodic continuation has been used extensively (Grinton & Cowley, 1971). Since, for kilovolt electrons traversing thin crystals, the transverse spreading of the dynamical wavefunction is limited (Cowley, 1981), the complex image amplitude at a particular point on the specimen exit face depends only on the crystal potential within a cylinder a few ångströms in diameter, erected about that point (Spence, O'Keefe & Iijima, 1978). The width of this cylinder depends on accelerating voltage, specimen thickness, and focus setting (see above references). Thus, small overlapping `patches' of exit-face wavefunction may be calculated in successive computations, and the results combined to form a larger area of image. The size of the `artificial superlattice' used should be increased until no change is found in the wavefunction over the central region of interest. For most defects, the positions of only a few atoms are important and, since the electron wavefunction is locally determined (for thin specimens at Scherzer focus), it appears that very large calculations are rarely needed for HREM work. The simulation of profile images of crystal surfaces at large defocus settings will, however, frequently be found to require large amounts of storage.
A new program should be tested to ensure that (a) under approximate two-beam conditions the calculated extinction distances for small-unit-cell crystals agree roughly with tabulated values (Hirsch et al., 1977b), (b) the simulated dynamical images have the correct symmetry, (c) for small thickness, the Scherzer-focus images agree with the projected potential, and (d) images and beam intensities agree with those of a program known to be correct. The damping envelope (product representation) [equation (4.3.8.17)] should only be used in a thin crystal with ; in general, the effects of partial spatial and temporal coherence must be incorporated using equation (4.3.8.13a) or (4.3.8.13b), depending on whether variations in diffraction conditions over are important. Thus, a separate multislice dynamical-image calculation for each component plane wave in the incident cone of illumination may be required, followed by an incoherent sum of all resulting images.
The outlook for obtaining higher resolution at the time of writing (1997) is broadly as follows. (1) The highest point resolution currently obtainable is close to 0.1 nm, and this has been obtained by taking advantage of the reduction in electron wavelength that occurs at high voltage [equation (4.3.8.16)]. A summary of results from these machines can be found in Ultramicroscopy (1994), Vol. 56, Nos. 1–3, where applications to fullerenes, glasses, quasicrystals, interfaces, ceramics, semiconductors, metals and oxides and other systems may be found. Fig. 4.2.8.6 shows a typical result. High cost, and the effects of radiation damage (particularly at larger thickness where defects with higher free energies are likely to be found), may limit these machines to a few specialized laboratories in the future. The attainment of higher resolution through this approach depends on advances in high-voltage engineering. (2) Aberration coefficients may be reduced if higher magnetic fields can be produced in the pole piece, beyond the saturation flux of the specialized iron alloys currently used. Research into superconducting lenses has therefore continued for many years in a few laboratories. Fluctuations in lens current are also eliminated by this method. (3) Electron holography was originally developed for the purpose of improving electron-microscope resolution, and this approach is reviewed in the following section. (4) Electron–optical correction of aberrations has been under study for many years in work by Scherzer, Crewe, Beck, Krivanek, Lanio, Rose and others – results of recent experimental tests are described in Haider & Zach (1995) and Krivanek, Dellby, Spence, Camps & Brown (1997). The attainment of 0.1 nm point resolution is considered feasible. Aberration correctors will also provide benefits other than increased resolution, including greater space in the pole piece for increased sample tilt and access to X-ray detectors, etc.
Structure image of a thin lamella of the 6H polytype of SiC projected along [110] and recorded at 1.2 MeV. Every atomic column (darker dots) is separately resolved at 0.109 nm spacing. The central horizontal strip contains a computer-simulated image; the structure is sketched at the left. [Courtesy of H. Ichinose (1994).] |
The need for resolution improvement beyond 0.1 nm has been questioned – the structural information retrievable by a single HREM image is always limited by the fact that a projection is obtained. (This problem is particularly acute for glasses.) Methods for combining different projected images (particularly of defects) from the same region (Downing, Meisheng, Wenk & O'Keefe, 1990) may now be as important as the search for higher resolution.
Since the resolution of an instrument is a property of the instrument alone, whereas the ability to distinguish HREM image features due to adjacent atoms depends on the scattering properties of the atoms, the resolution of an electron microscope cannot easily be defined [see Subsection 2.5.2.9 in IT B (2001)]. The Rayleigh criterion was developed for the incoherent imaging of point sources and cannot be applied to coherent phase contrast. Only for very thin specimens of light elements for which it can be assumed that the scattering phase is −π/2 can the straightforward definition of point resolution [equation (4.3.8.16)] be applied. In general, the dynamical wavefunction across the exit face of a crystalline sample bears no simple relationship to the crystal structure, other than to preserve its symmetry and to be determined by the `local' crystal potential. The use of a dynamical `R factor' between computed and experimental images of a known structure has been suggested by several workers as the basis for a more general resolution definition.
For weakly scattering specimens, the most satisfactory method of measuring either the point resolution or the information limit [see equation (4.3.8.21)] appears to be that of Frank (1975). Here two successive micrographs of a thin amorphous film are recorded (under identical conditions) and the superimposed pair used to obtain a coherent optical diffractogram crossed by fringes. The fringes, which result from small displacements of the micrographs, extend only to the band limit of information common to both micrographs, and cannot be extended by photographic processing, noise, or increased exposure. By plotting this band limit against defocus, it is possible to determine both Δ and . As an alternative, for thin crystalline samples of large-unit-cell materials, the parameters Δ, , and can be determined by matching computed and experimental images of crystals of known structure. It is the specification of these parameters (for a given electron intensity and wavelength) that is important in describing the performance of high-resolution electron microscopes. We note that certain conditions of focus or thickness may give a spurious impression of ultra-high resolution [see equations (4.3.8.7) and (4.3.8.8)].
Within the domain of linear imaging, implying, for the most part, the validity of the WPO approximation, many forms of image processing have been employed. These have been of particular importance for crystalline and non-crystalline biological materials and include image reconstruction [see Section 2.5.5 in IT B (2001)] and the derivation of three-dimensional structures from two-dimensional projections [see Section 2.5.6 in IT B (2001)]. For reviews, see also Saxton (1980a), Frank (1980), and Schiske (1975). Several software packages now exist that are designed for image manipulation, Fourier analysis, and cross correlation; for details of these, see Saxton (1980a) and Frank (1980). The theoretical basis for the WPO approximation closely parallels that of axial holography in coherent optics, thus much of that literature can be applied to HREM image processing. Gabor's original proposal for holography was intended for electron microscopy [see Cowley (1981) for a review].
The aim of image-processing schemes is the restoration of the exit-face wavefunction, given in equation (4.3.8.13a). The reconstruction of the crystal potential from this is a separate problem, since these are only simply related under the approximation of Subsection 4.3.8.3. For a non-linear method that allows the reconstruction of the dynamical image wavefunction, based on equation (4.3.8.13b), which thus includes the effects of multiple scattering, see Saxton (1980b).
The concept of holographic reconstruction was introduced by Gabor (1948, 1949) as a means of enhancing the resolution of electron microscopes. Gabor proposed that, if the information on relative phases of the image wave could be recorded by observing interference with a known reference wave, the phase modification due to the objective-lens aberrations could be removed. Of the many possible forms of electron holography (Cowley, 1994), two show particular promise of useful improvements of resolution. In what may be called in-line TEM holography, a through-focus series of bright-field images is obtained with near-coherent illumination. With reference to the relatively strong transmitted beam, the relative phase and amplitude changes due to the specimen are derived from the variations of image intensity (see Van Dyck, Op de Beeck & Coene, 1994). The tilt-series reconstruction method also shows considerable promise (Kirkland, Saxton, Chau, Tsuno & Kawasaki, 1995).
In the alternative off-axis approach, the reference wave is that which passes by the specimen area in vacuum, and which is made to interfere with the wave transmitted through the specimen by use of an electrostatic biprism (Möllenstedt & Düker, 1956). The hologram consists of a modulated pattern of interference fringes. The image wavefunction amplitude and phase are deduced from the contrast and lateral displacements of the fringes (Lichte, 1991; Tonomura, 1992). The process of reconstruction from the hologram to give the image wavefunction may be performed by optical-analogue or digital methods and can include the correction of the phase function to remove the effects of lens aberrations and the attendant limitation of resolution. The point resolution of electron microscopes has recently been exceeded by this method (Orchowski, Rau & Lichte, 1995).
The aim of the holographic reconstructions is the restoration of the wavefunction at the exit face of the specimen as given by equation (4.3.8.13a). The reconstruction of the crystal potential from this is a separate problem, since the exit-face wavefunction and are simply related only under the WPO approximations of Subsection 4.3.8.3. The possibility of deriving reconstructions from wavefunctions strongly affected by dynamical diffraction has been considered by a number of authors (for example, Van Dyck et al., 1994). The problem does not appear to be solvable in general, but for special cases, such as perfect thin single crystals in exact axial orientations, considerable progress may be possible.
Since a single atom, or a column of atoms, acts as a lens with negative spherical aberration, methods for obtaining super-resolution using atoms as lenses have recently been proposed (Cowley, Spence & Smirnov, 1997).
A number of non-conventional imaging modes have been found useful in electron microscopy for particular applications. In scanning transmission electron microscopy (STEM), powerful electron lenses are used to focus the beam from a very small bright source, formed by a field-emission gun, to form a small probe that is scanned across the specimen. Some selected part of the transmitted electron beam (part of the coherent convergent-beam electron diffraction pattern produced) is detected to provide the image signal that is displayed or recorded in synchronism with the incident-beam scan. The principle of reciprocity suggests that, for equivalent lenses, apertures and column geometry, the resolution and contrast of STEM and TEM images will be identical (Cowley, 1969). Practical considerations of instrumental convenience distinguish particularly useful STEM modes.
Crewe & Wall (1970) showed that, if an annular detector is used to detect all electrons scattered outside the incident-beam cone, dark-field images could be obtained with high efficiency and with a resolution better than that of the bright-field mode by a factor of about 1.4. If the inner radius of the annular detector is made large (of the order of 10−1 rad for 100 kV electrons), the strong diffracted beams occurring for lower angles do not contribute to the resulting high-angle annular dark-field (HAADF) image (Howie, 1979), which is produced mainly by thermal diffuse scattering. The HAADF mode has important advantages for particular purposes because the contrast is strongly dependent on the atomic number, Z, of the atoms present but is not strongly affected by dynamical diffraction effects and so shows near-linear variation with Z and with the atom-number density in the sample. Applications have been made to the imaging of small high-Z particles in low-Z supports, such as in supported metal catalysts (Treacy & Rice, 1989) and to the high-resolution imaging of individual atomic rows in semiconductor crystals, showing the variations of composition across planar interfaces (Pennycook & Jesson, 1991).
The STEM imaging modes may be readily correlated with microchemical analysis of selected specimen areas having lateral dimensions in the nanometre range, by application of the techniques of electron energy-loss spectroscopy or X-ray energy-dispersive analysis (Williams & Carter, 1996; Section 4.3.4). Also, diffraction patterns (coherent convergent-beam electron diffraction patterns) may be obtained from any chosen region having dimensions equal to those of the incident-beam diameter and as small as about 0.2 nm (Cowley, 1992). The coherent interference between diffracted beams within such a pattern may provide information on the symmetries, and, ultimately, the atomic arrangement, within the illuminated area, which may be smaller than the projection of the crystal unit cell in the beam direction. This geometry has been used to extend resolution for crystalline samples beyond even the information resolution limit, di (Nellist, McCallum & Rodenburg, 1995), and is the basis for an exact, non-perturbative inversion scheme for dynamical electron diffraction (Spence, 1998).
The detection of secondary radiations (light, X-rays, low-energy `secondary' electrons, etc.) in STEM or the detection of energy losses of the incident electrons, resulting from particular elementary excitations of the atoms in a crystal, in TEM or STEM, may be used to form images showing the distributions in a crystal structure of particular atomic species. In principle, this may be extended to the chemical identification of individual atom types in the projection of crystal structures, but only limited success has been achieved in this direction because of the relatively low level of the signals available. The formation of atomic resolution images using inner-shell excitations, for example, is complicated by the Bragg scattering of these inelastically scattered electrons (Endoh, Hashimoto & Makita, 1994; Spence & Lynch, 1982).
Reflection electron microscopy (REM) has been shown to be a powerful technique for the study of the structures and defects of crystal surfaces with moderately high spatial resolution (Larsen & Dobson, 1988), especially when performed in a specially built electron microscope having an ultra-high-vacuum specimen environment (Yagi, 1993). Images are formed by detecting strong diffracted beams in the RHEED patterns produced when kilovolt electron beams are incident on flat crystal surfaces at grazing incidence angles of a few degrees. The images suffer from severe foreshortening in the beam direction, but, in directions at right angles to the beam, resolutions approaching 0.3 nm have been achieved (Koike, Kobayashi, Ozawa & Yagi, 1989). Single-atom-high surface steps are imaged with high contrast, surface reconstructions involving only one or two monolayers are readily seen and phase transitions of surface superstructures may be followed.
The study of surface structure by use of high-resolution transmission electron microscopes has also been productive in particular cases. Images showing the structures of surface layers with near-atomic resolution have been obtained by the use of `forbidden' or `termination' reflections (Cherns, 1974; Takayanagi, 1984) and by phase-contrast imaging (Moodie & Warble, 1967; Iijima, 1977). The imaging of the profiles of the edges of thin or small crystals with clear resolution of the surface atomic layers has also been effective (Marks, 1986). The introduction of the scanning tunnelling microscope (Binnig, Rohrer, Gerber & Weibel, 1983) and other scanning probe microscopies has broadened the field of high-resolution surface structure imaging considerably.
For many materials of organic or biological origin, it is possible to obtain very thin crystals, only one or a few molecules thick, extending laterally over micrometre-size areas. These may give selected-area electron-diffraction patterns in electron microscopes with diffraction spots extending out to angles corresponding to d spacings as low as 0.1 nm. Because the materials are highly sensitive to electron irradiation, conventional bright-field images cannot be obtained with resolutions better than several nanometres. However, if images are obtained with very low electron doses and then a process of averaging over the content of a very large number of unit cells of the image is carried out, images showing detail down to the scale of 1 nm or less may be derived for the periodically repeated unit. From such images, it is possible to derive both the magnitudes and phases of the Fourier coefficients, the structure factors, out to some limit of d spacings, say . From the diffraction patterns, the magnitudes of the structure factors may be deduced, with greater accuracy, out to a much smaller limit, . By combination of the information from these two sources, it may be possible to obtain a greatly improved resolution for an enhanced image of the structure. This concept was first introduced by Unwin & Henderson (1975), who derived images of the purple membrane from Halobacterium halobium, with greatly improved resolution, revealing its essential molecular configuration.
Recently, several methods of phase extension have been developed whereby the knowledge of the relative phases may be extended from the region of the diffraction pattern covered by the electron-microscope image transform to the outer parts. These include methods based on the use of the tangent formula or Sayre's equation (Dorset, 1994; Dorset, McCourt, Fryer, Tivol & Turner, 1994) and on the use of maximum-entropy concepts (Fryer & Gilmore, 1992). Such methods have also been applied, with considerable success, to the case of some thin inorganic crystals (Fu et al., 1994). In this case, the limitation on the resolution set by the electron-microscope images may be that due to the transfer function of the microscope, since radiation-damage effects are not so limiting. Then, the resolution achieved by the combined application of the electron diffraction data may represent an advance beyond that of normal HREM imaging. Difficulties may well arise, however, because the theoretical basis for the phase-extension methods is currently limited to the WPO approximation. A summary of the present situation is given in the book by Dorset (1995).
Acknowledgements
The authors of Section 4.3.3 acknowledge with gratitude the contributions of Kenneth and Lise Hedberg, who made many helpful suggestions regarding the manuscript and carefully checked the numerical results for smoothness and consistency.
References
Ackermann, I. (1948). Observations on the dynamical interference phenomena in convergent electron beams. II. Ann. Phys. (Leipzig), 2, 41–54.Google ScholarAhn, C. C. & Krivanek, O. L. (1982). An EELS atlas. Available from Center for Solid State Science, Arizona State University, Tempe, Arizona 85287, USA.Google Scholar
Ahn, C. C. & Rez, P. (1985). Inner shell edge profiles in electron energy loss spectroscopy. Ultramicroscopy, 17, 105–116.Google Scholar
Alexander, H., Spence, J. C. H., Shindo, D., Gottschalk, H. & Long, N. (1986). Forbidden reflection lattice imaging for the determination of kink densities on partial dislocations. Philos. Mag. A53, 627–643.Google Scholar
Altarelli, M. & Smith, S. Y. (1974). Superconvergence and sum rules for the optical constants: physical meaning, comparison with experiment and generalization. Phys. Rev. B, 9, 1290–1298.Google Scholar
Andersson, B. (1975). Structure analysis of the γ-phase in the vanadium oxide system by electron diffraction studies. Acta Cryst. A31, 63–70.Google Scholar
Ando, Y., Ichimiya, A. & Uyeda, R. (1974). A determination of values and signs of the 111 and 222 structure factors of silicon. Acta Cryst. A30, 600–601.Google Scholar
Anstis, G. R., Lynch, D. F., Moodie, A. F. & O'Keefe, M. A. (1973). n-Beam lattice images. III. Upper limits of ionicity in W4Nb26P77. Acta Cryst. A29, 138–147.Google Scholar
d'Anterroches, C. & Bourret, A. (1984). Atomic structure of [011] and [001] near-coincident tilt boundaries in germanium and silicon. Philos. Mag. A49, 783–807.Google Scholar
Arnesen, S. P. & Seip, H. M. (1966). Studies on the failure of the first Born approximation in electron diffraction. V. Molybdenum- and tungsten hexacarbonyl. Acta Chem. Scand. 20, 2711–2727.Google Scholar
Barrett, C. S. & Massalski, T. B. (1980). Structure of metals, 3rd revised ed. Oxford: Pergamon Press.Google Scholar
Bartell, L. S. (1975). Modification of Glauber theory for dynamic scattering of electrons by polyatomic molecules. J. Chem. Phys. 63, 3750–3755.Google Scholar
Bartell, L. S. & Brockway, L. O. (1953). The investigation of electron distribution in atoms by electron diffraction. Phys. Rev. 90, 833–838.Google Scholar
Bartell, L. S. & Gavin, R. M. Jr (1964). Effects of electron correlation in X-ray and electron diffraction. J. Am. Chem. Soc. 86, 3493–3498.Google Scholar
Bartell, L. S. & Miller, B. (1980). Extension of Glauber theory to account for intratarget diffraction in multicenter scattering. J. Chem. Phys. 72, 800–807.Google Scholar
Bartell, L. S. & Wong, T. C. (1972). Three-atom scattering in gas-phase electron diffraction: a tractable limiting case. J. Chem. Phys. 56, 2364–2367.Google Scholar
Batson, P. E. (1986). High energy resolution electron spectrometer for a 1 nm spatial analysis. Rev. Sci. Instrum. 57, 43–48.Google Scholar
Batson, P. E. (1987). Spatially resolved interband spectroscopy. Physical aspects of microscopic characterization of materials, edited by J. Kirschner, K. Murata & J. A. Venables, pp. 189–195. Scanning Microscopy, Suppl. I.Google Scholar
Batson, P. E. (1989). High resolution energy-loss spectroscopy. Ultramicroscopy, 28, 32–39.Google Scholar
Batson, P. E. & Silcox, J. (1983). Experimental energy loss function, , for aluminium. Phys. Rev. B, 27, 5224–5239.Google Scholar
Berger, S. D. & McMullan, D. (1989). Parallel recording for an electron spectrometer on a scanning transmission electron microscope. Ultramicroscopy, 28, 122–125.Google Scholar
Bernal, J. D. (1926). On the interpretation of X-ray single-crystal rotation photographs. Proc. R. Soc. London Ser. A, 113, 117–160.Google Scholar
Bethe, H. A. (1928). Theorie der Beugung von Elektronen an Kristallen. Ann. Phys. Leipzig, 87, 55–129.Google Scholar
Bethe, H. A. (1930). Zur Theorie des Durchgangs schneller Korpuskularstrahlen durch Materie. Ann. Phys. (Leipzig), 5, 325–400.Google Scholar
Bianconi, A., Fritsch, E., Calas, G. & Petiau, J. (1985). X-ray absorption near edge structure of 3d transition elements in tetrahedral coordination: the effect of bond length variation. Phys. Rev. B, 32, 4292–4295.Google Scholar
Biggs, F., Mendelsohn, L. B. & Mann, J. B. (1975). Hartree–Fock Compton profiles for the elements. At. Data Nucl. Data Tables, 16, 210–309.Google Scholar
Binnig, G., Rohrer, H., Gerber, C. & Weibel, E. (1983). Direct imaging of semiconductor surfaces. Phys. Rev. Lett. 50, 120–123.Google Scholar
Bird, D. M. (1990). Absorption in high-energy electron diffraction from non-centrosymmetrical crystals. Acta Cryst. A46, 208–214.Google Scholar
Bird, D. M. & King, Q. A. (1990). Absorptive form factors for high-energy electron diffraction. Acta Cryst. A46, 202–208.Google Scholar
Bird, D. M. & Saunders, M. (1992). Inversion of convergent-beam electron diffraction patterns. Acta Cryst. A48, 555–562.Google Scholar
Blackman, M. (1939). On the intensities of electron diffraction rings. Proc. R. Soc. London Ser. A, 173, 68–82.Google Scholar
Blake, R. G., Jostsons, A., Kelly, P. M. & Napier, J. G. (1978). The determination of extinction distances and anomalous absorption coefficients by scanning electron microscopy. Philos. Mag. A37, 1–16.Google Scholar
Boersch, H. (1954). Experimentelle Bestimmung der Energieverteilung in thermisch ausgelosten Elektronenstrahlen. Z. Phys. 139, 115–146.Google Scholar
Bonham. R. A. (1965a). Multiple elastic intramolecular scattering in gas electron diffraction. J. Chem. Phys. 43, 1103–1109.Google Scholar
Bonham, R. A. (1965b). Corrections to the incoherent scattering factor for electrons and X-rays. J. Chem. Phys. 43, 1460–1464.Google Scholar
Bonham, R. A. (1966). Dynamic effects in gas electron diffraction. Trans. Am. Crystallogr. Assoc. 2, 165–172.Google Scholar
Bonham, R. A. (1967). Some new relations connecting molecular properties and electron and X-ray diffraction intensities. J. Phys. Chem. 71, 856–862.Google Scholar
Bonham, R. A. & Cox, H. L. Jr (1967). 40-kV electron scattering fom Ne, Ar, Kr, and Xe measured by the sector-microphotometer electron-diffraction method. J. Chem. Phys. 47, 3508–3517.Google Scholar
Bonham, R. A. & Fink, M. (1974). High energy electron scattering, Chaps. 5 and 6. New York: Van Nostrand Reinhold.Google Scholar
Bonham. R. A. & Iijima, T. (1965). Preliminary electron-diffraction study of H2 at small scattering angles. J. Chem. Phys. 42, 2612–2614.Google Scholar
Bonham, R. A. & Su, L. S. (1966). Use of Hellman–Feynman and hyperviral theorems to obtain anharmonic vibration–rotation expectation values and their application to gas diffraction. J. Chem. Phys. 45, 2827–2831.Google Scholar
Breitenstein, M., Endesfelder, A., Meyer, H. & Schweig, A. (1984). CI calculations of electron-scattering cross sections for some linear molecules. Chem. Phys. Lett. 108, 430–434.Google Scholar
Breitenstein, M., Endesfelder, A., Meyer, H., Schweig, A. & Zittlau, W. (1983). Electron correlation effects in electron scattering cross-section calculations of N2. Chem. Phys. Lett. 97, 403–409.Google Scholar
Breitenstein, M., Mawhorter, R. J., Meyer, H. & Schweig, A. (1984). Theoretical study of potential-energy differences from high-energy electron scattering cross sections of CO2. Phys. Rev. Lett. 53, 2398–2401.Google Scholar
Breitenstein, M., Mawhorter, R. J., Meyer, H. & Schweig, A. (1986). Vibrational effects on electron–molecule scattering for polyatom in the first Born approximation: H2O. Mol. Phys. 57, 81–88.Google Scholar
Bross, H. (1978a). Anisotropy of plasmon dispersion in Al. Phys. Lett. A, 64, 418–420.Google Scholar
Bross, H. (1978b). Pseudopotential theory of the dielectric function of Al – the volume plasmon dispersion. J. Phys. F, 8, 2631–2649.Google Scholar
Brydson, R., Sauer, H., Engle, W., Thomas, J. M., Zeitler, E., Kosugi, N. & Kuroda, H. (1989). Electron energy loss and X-ray absorption spectroscopy of rutile and anatase: a test of structural sensitivity. J. Phys. Condens. Matter, 1, 797–812.Google Scholar
Budinger, T. F. & Glaeser, R. M. (1976). Measurement of focus and spherical aberration of an electron microscope objective lens. Ultramicroscopy, 2, 31–41.Google Scholar
Bunge, C. F., Barrientos, J. & Bunge, A. V. (1993). Roothaan–Hartree–Fock ground-state atomic wave functions: Slater-type orbital expansions and expectation values. At. Data Nucl. Data Tables, 53, 113–162.Google Scholar
Bunge, H.-J. (1982). Texture analysis in materials science. London: Butterworth.Google Scholar
Bunyan, P. J. (1963). The effect of multiple elastic scattering in gas electron diffraction. Proc. Phys. Soc. 82, 1051–1057.Google Scholar
Buxton, B. F. (1976). Bloch waves in high order Laue zone effects in high energy electron diffraction. Proc. R. Soc. London Ser. A, 300, 335–361.Google Scholar
Castaing, R. & Henry, L. (1962). Filtrage magnétique des vitesses en microscopie électronique. C. R. Acad. Sci. Paris Sér. B, 255, 76–78.Google Scholar
Chen, C. H. & Silcox, J. (1971). Detection of optical surface guided modes in thin graphite films by high energy electron scattering. Phys. Rev. Lett. 35, 390–393.Google Scholar
Cherns, D. (1974). Direct resolution of surface steps by transmission electron microscopy. Philos. Mag. 30, 549–557.Google Scholar
Cockayne, D. J. H. & Gronsky, R. (1981). Lattice fringe imaging of modulated structures. Philos. Mag. A44, 159–175.Google Scholar
Coffman, D., Fink, M. & Wellenstein, H. (1985). Elastic small-angle electron scattering by He, Ne, and Ar at 35 keV. Phys. Rev. Lett. 55, 1392–1394.Google Scholar
Colliex, C. (1984). Electron energy loss spectroscopy in the electron microscope. Advances in optical and electron microscopy, Vol. 9, edited by V. E. Cosslett & R. Barer, pp. 65–177. London: Academic Press.Google Scholar
Colliex, C. (1985). An illustrated review on various factors governing the high spatial resolution capabilities in EELS microanalysis. Ultramicroscopy, 18, 131–150.Google Scholar
Colliex, C., Gasgnier, M. & Trebbia, P. (1976). Analysis of the electron excitation spectra in heavy rare earch metals, hydrides and oxides. J. Phys. (Paris), 27, 397–406.Google Scholar
Colliex, C., Manoubi, T., Gasgnier, M. & Brown, L. M. (1985). Near edge structures on EELS core-loss edges. Scanning Electron Microsc. 2, 489–512.Google Scholar
Colliex, C., Maurice, J. L. & Ugarte, D. (1989). Frontiers of analytical electron microscopy with special reference to cluster and interface problems. Ultramicroscopy, 29, 31–43.Google Scholar
Coulthard, M. A. (1967). A relativistic Hartree–Fock atomic field calculation. Proc. Phys. Soc. 91, 44–49.Google Scholar
Cowley, J. M. (1959). The electron-optical imaging of crystal lattices. Acta Cryst. 12, 367–375.Google Scholar
Cowley, J. M. (1961). Diffraction intensities from bent crystals. Acta Cryst. 14, 920–926.Google Scholar
Cowley, J. M. (1969). Image contrast in transmission scanning electron microscope. Appl. Phys. Lett. 15, 58–59.Google Scholar
Cowley, J. M. (1975). Diffraction physics. New York: North-Holland.Google Scholar
Cowley, J. M. (1981). Diffraction physics, 2nd ed. New York: North-Holland.Google Scholar
Cowley, J. M. (1988). Electron microscopy of crystals with time-dependent perturbations. Acta Cryst. A44, 847–853.Google Scholar
Cowley, J. M. (1992). Coherent convergent beam diffraction. Electron diffraction techniques, Vol. 1, edited by J. M. Cowley, pp. 439–464. Oxford University Press.Google Scholar
Cowley, J. M. (1994). Applications of electron holography. In Handbook of advanced materials testing, edited by N. P. Cheremisinoff & P. N. Cheremisinoff. New York: Marcel Dekker, Inc.Google Scholar
Cowley, J. M. & Iijima, S. (1972). Electron microscope image contrast for thin crystals. Z. Naturforsch. Teil A, 27, 445–451.Google Scholar
Cowley, J. M. & Moodie, A. F. (1957). The scattering of electrons by atoms and crystals. I. A new theoretical approach. Acta Cryst. 10, 609–619.Google Scholar
Cowley, J. M. & Moodie. A. F. (1960). Fourier images IV: the phase grating. Proc. Phys. Soc. (London), 76, 378–384.Google Scholar
Cowley, J. M., Spence, J. C. & Smirnov, V. V. (1997). The enhancement of electron microscope resolution by the use of atomic focusers. Ultramicroscopy, 68, 135–148.Google Scholar
Craven, A. J. & Buggy, T. W. (1981). Design considerations and performance of an analytical STEM. Ultramicroscopy, 7, 27–37.Google Scholar
Crewe, A. V. (1977a). Post specimen optics in the STEM. I. General information. Optik (Stuttgart), 47, 299–312.Google Scholar
Crewe, A. V. (1977b). Post specimen optics in the STEM. II. Optik (Stuttgart), 47, 371–380.Google Scholar
Crewe, A. V. & Wall, J. (1970). A scanning microscope with 5 Å resolution. J. Mol. Biol. 48, 375–393.Google Scholar
Cromer, D. T. & Waber, J. T. (1974). Atomic scattering factors for X-rays. In International tables for X-ray crystallography, Vol. IV, Section 2.2. Birmingham: Kynoch Press.Google Scholar
Daberkow, I., Herrmann, K.-H., Liu, L. & Rau, W. (1991). Performance of electron image converters with YAG and CCD. Ultramicroscopy, 38, 215–224.Google Scholar
Daniels, J., Festenberg, C. V., Raether, H. & Zeppenfeld, K. (1970). Optical constants of solids by electron spectroscopy. Springer tracts in modern physics, Vol. 54, pp. 78–135. New York: Springer-Verlag.Google Scholar
Desseaux, J., Renault, A. & Bourret, A. (1977). Multibeam lattice images from germanium oriented in (001). Philos. Mag. 35, 357–363.Google Scholar
Disko, M. M., Krivanek, O. L. & Rez, P. (1982). Orientation dependent extended fine structure in EELS. Phys. Rev. B, 25, 4252–4255.Google Scholar
Dorset, D. L. (1991). Is electron crystallography possible? The direct determination of organic crystal structures. Ultramicroscopy, 38, 23–40.Google Scholar
Dorset, D. L. (1994). Electron crystallography of organic molecules. Adv. Electron. Electron Phys. 88, 111–197.Google Scholar
Dorset, D. L. (1995). Editor. Structural electron crystallography. New York/London: Plenum Press.Google Scholar
Dorset, D. L., Jap, B. K., Ho, M. M. & Glaeser, R. M. (1979). Direct phasing of electron diffraction data from organic crystals: the effect of n-beam dynamical scattering. Acta Cryst. A35, 1001–1009.Google Scholar
Dorset, D. L., McCourt, M. P., Fryer, J. R., Tivol, W. F. & Turner, J. N. (1994). The tangent formula in electron crystallography. Phase determination of copper perchlorophthalocyanine. Microsc. Soc. Am. Bull. 24, 398–404.Google Scholar
Downing, K. H., Meisheng, H., Wenk, H. R. & O'Keefe, M. A. (1990). Resolution of oxygen atoms in staurolite by three dimensional transmission electron microscopy. Nature (London), 348, 525.Google Scholar
Doyle, P. A. & Turner, P. S. (1968). Relativistic Hartree–Fock X-ray and electron scattering factors. Acta Cryst. A24, 390–397.Google Scholar
Durham, J. P., Pendry, J. B. & Hodges, C. H. (1981). XANES: Determination of bond angles and multi-atom correlations in ordered and disordered systems. Solid State Commun. 8, 159–162.Google Scholar
Egerton, R. F. (1978). Formulae for light element analysis by electron energy loss spectrometry. Ultramicroscopy, 3, 243–351.Google Scholar
Egerton, R. F. (1979). K-shell ionization cross sections for use in microanalysis. Ultramicroscopy, 4, 169–179.Google Scholar
Egerton, R. F. (1980a). The use of electron lenses between a TEM specimen and an electron spectrometer. Optik (Stuttgart), 56, 363–376.Google Scholar
Egerton, R. F. (1980b). Design of an aberration-corrected electron spectrometer for the TEM. Optik (Stuttgart), 57, 229–242.Google Scholar
Egerton, R. F. (1986). Electron energy loss spectroscopy in the electron microscope. New York/London: Plenum.Google Scholar
Egerton, R. F. & Crozier, P. A. (1987). A compact parallel recording. J. Microsc. 148, 157.Google Scholar
Endoh, H., Hashimoto, H. & Makita, Y. (1994). Theoretical and observed images of impurity atoms formed by L-shell ionization. Ultramicroscopy, 56, 108–120.Google Scholar
Enge, H. A. (1967). Deflecting magnets. Focusing of charged particles, Vol. 2, edited by A. Septier, pp. 203–264. New York: Academic Press.Google Scholar
Engel, W., Sauer, H., Zeitler, E., Brydson, R., Williams, B. G. & Thomas, J. M. (1988). Electron energy loss spectroscopy and the crystal chemistry of rhodizite. J. Chem. Soc. Faraday Trans. 1, 84, 617–629.Google Scholar
Epstein, J. & Stewart, R. F. (1977). X-ray and electron scattering from diatomic molecules in the first Born approximation. J. Chem. Phys. 66, 4057–4064.Google Scholar
Fano, U. (1961). Effects of configuration interaction on intensities and phase shifts. Phys. Rev. 124, 1966–1978.Google Scholar
Fejes, P. L. (1977). Approximations for the calculation of high-resolution electron-microscope images of thin films. Acta Cryst. A33, 109–113.Google Scholar
Fields, P. M. & Cowley, J. M. (1978). Computed electron microscope images of atomic defects in f.c.c. metals. Acta Cryst. A34, 103–112.Google Scholar
Fink, J. & Kisker, E. (1980). A method for rapid calculation of electron trajectories in multielement electrostatic cylinder lenses. Rev. Sci. Instrum. 51, 918–920.Google Scholar
Fink, J. & Leising, G. (1986). Momentum-dependent dielectric functions of oriented trans-polyacetylene. Phys. Rev. B, 34, 5320–5328.Google Scholar
Fink, J., Müller-Heinzerling, T., Pflüger, J., Scheerer, B., Dischler, B., Koidl, P., Bubenzer, A. & Sah, R. E. (1984). Investigation of hydrocarbon-plasma-generated carbon films by EELS. Phys. Rev. B, 30, 4713–4718.Google Scholar
Fink, M., Bonham, R. A., Lee, J. S. & Ng, E. W. (1969). Large angle scattering from N2 with 40 keV electrons. Chem. Phys. Lett. 4, 347–351.Google Scholar
Fink, M. & Kessler, J. (1966). Absolute Wirkungsquerschnitte für Elektronenstreuung um Kleine Winkel. Experimente sum Gültigkeitsbereich der Ersten Bornschen Näherung. Z. Phys. 196, 1–15.Google Scholar
Fischer, D. W. (1970). Molecular orbital interpretation of the soft X-ray L23 emission and absorption spectra from some titanium and vanadium compounds. J. Appl. Phys. 41, 3561–3569.Google Scholar
Fitzgerald, J. D. & Johnson, A. W. S. (1984). A simplified method of electron microscope voltage measurement. Ultramicroscopy, 12, 231–236.Google Scholar
Fox, A. G. & Fisher, R. M. (1988). A summary of low-angle X-ray atomic scattering factors measured by the critical voltage effect in high energy electron diffraction. Aust. J. Phys. 41, 461–468.Google Scholar
Fox, A. G., O'Keefe, M. A. & Tabbernor, M. A. (1989). Relativistic Hartree–Fock X-ray and electron atomic scattering factors at high angles. Acta Cryst. A45, 786–793.Google Scholar
Fox, A. G. & Tabbernor, M. A. (1991). The bonding charge density of ′NiAl′′. Acta Metall. 39, 669–678.Google Scholar
Frank, J. (1975). A practical resolution criterion in optics and electron microscopy. Optik (Stuttgart), 43, 25–34.Google Scholar
Frank, J. (1980). The role of correlation techniques in computer image processing. Computer processing of electron microscope images. Topics in current physics, Vol. 13, edited by P. W. Hawkes, p. 187. Berlin/Heidelberg/New York: Springer.Google Scholar
Fryer, J. R. & Gilmore, C. J. (1992). Structure determination by electron crystallography. Trans. Am. Crystallogr. Assoc. 28, 57–75.Google Scholar
Fu, Z. Q., Huang, D. X., Li, F. H., Li, J. Q., Zhao, Z. X., Cheng, T. Z. & Fan, H. F. (1994). Incommensurate modulation in minute crystals revealed by combining high-resolution electron microscopy and electron diffraction. Ultramicroscopy, 54, 229–236.Google Scholar
Fujiwara, K. (1959). Application of higher-order Born approximation to multiple elastic scattering of electrons in crystals. J. Phys. Soc. Jpn, 14, 1513–1524.Google Scholar
Fujiwara, K. (1961). Relativistic dynamical theory of electron diffraction. J. Phys. Soc. Jpn, 16, 2226–2238.Google Scholar
Fujiyoshi, Y., Ishizuka, K., Tsuji, M., Kobayashi, T. & Uyeda, N. (1983). Charge density distribution from high resolution molecular images. Proceedings of the 17th International Conference on High Voltage Electron Microscopy, p. 21.Google Scholar
Fujiyoshi, Y., Mizusaki, T., Morikawa, K., Yamagishi, H., Aoki, Y., Kihara, H. & Harada, Y. (1991). Development of a superfluid helium stage for high-resolution electron microscopy. Ultramicroscopy, 38, 241–251.Google Scholar
Fukuhara, A. (1966). Many-ray approximation in the dynamical theory of electron diffraction. J. Phys. Soc. Jpn, 21, 2645–2662.Google Scholar
Gabor, D. (1948). A new microscope principle. Nature (London), 161, 777–778.Google Scholar
Gabor, D. (1949). Microscopy of reconstructed wavefronts. Proc. R. Soc. London Ser. A, 197, 454–487.Google Scholar
Geiger, J. (1964). Streuung von 25 keV-Elektronen an Gasen. II. Streuung an Neon, Argon, Krypton und Xenon. Z. Phys. 177, 138–145.Google Scholar
Gibbons, P. C., Ritsko, J. J. & Schnatterly, S. E. (1975). Inelastic electron scattering spectrometer. Rev. Sci. Instrum. 46, 1546–1554.Google Scholar
Gjønnes, J. (1964). A dynamic effect in electron diffraction by molecules. Acta Cryst. 17, 1075–1076.Google Scholar
Gjønnes, J. & Høier, R. (1971). The application of non-systematic many-beam dynamic effects to structure-factor determination. Acta Cryst. A27, 313–316.Google Scholar
Gjønnes, K. & Bøe, N. (1994). Refinement of temperature factors and charge distributions in YBa2Cu3O7 and YBa2(Cu,Co)3O7 from CBED intensities. Micron Microsc. Acta, 25, 29–44.Google Scholar
Glauber, R. & Schomaker, V. (1953). The theory of electron diffraction. Phys. Rev. 89, 667–671.Google Scholar
Glazer, J., Ramesh, R., Hilton, M. R. & Sarikaya, M. (1985). Comparison of convergent beam electron diffraction methods for determination of foil thickness. Philos. Mag. A52, 59–63.Google Scholar
Goodman, P. (1976). Examination of the graphite structure using convergent-beam electron diffraction. Acta Cryst. A32, 793–798.Google Scholar
Goodman, P. & Lehmpfuhl, G. (1967). Electron diffraction study of MgO h00 systematic interactions. Acta Cryst. 22, 14–24.Google Scholar
Goodman, P. & Moodie, A. F. (1974). Numerical evaluation of N-beam wave-functions in electron scattering by the multislice method. Acta Cryst. A30, 280–290.Google Scholar
Grinton, G. R. & Cowley, J. M. (1971). Phase and amplitude contrast in electron micrographs of biological materials. Optik (Stuttgart), 34, 221–233.Google Scholar
Gritsaenko, G. S., Zvyagin, B. B., Boyarskaya, R. V., Gorshkov, A. I., Samotoin, N. D. & Frolova, K. E. (1969). Methods of electron microscopy of minerals. Moscow: Nauka.Google Scholar
Grunes, L. A., Leapman, R. D., Wilker, C. N., Hoffmann, R. & Kunz, A. B. (1982). Oxygen K near-edge fine structure: an electron energy-loss investigation with comparisons to new theory for selected 3d transition-metal oxides. Phys. Rev. B, 25, 7157–7173.Google Scholar
Guinier, A. (1956). Théorie et technique de la radiocristallographie, 2nd ed. Paris: Dunod.Google Scholar
Hagemann, H. J., Gudat, W. & Kunz, C. (1975). Optical constants from the far infrared to the X-ray region: Mg, Al, Cu, Ag, Au, Bi, C, and Al2O3. J. Opt. Soc. Am. 65, 742–748.Google Scholar
Haider, M. & Zach, J. (1995). Multipole correctors. Proceedings of Microscopy and Microanalysis, edited by G. Bailey, pp. 596–567. New York: Jones and Bigell.Google Scholar
Hanson, H. P. (1962). Experimental f values and electron diffraction amplitudes for bromine. J. Chem. Phys. 36, 1043–1049.Google Scholar
Hartl, W. A. M. (1966). Die Filterlinse als Monochromator für schnelle Elektronen. Z. Phys. 191, 487–502.Google Scholar
Hashimoto, H., Mannami, M. & Naiki, T. (1961). Theory of lattice images. Philos. Trans. R. Soc. London, 253, 459–489.Google Scholar
Heine, V. (1980). Electronic structure from the point of view of the local atomic environment. Solid State Phys. 35, 1–127.Google Scholar
Heisenberg, W. (1931). Über die Inkohärente Streuung von Röntgenstrahlen. Phys. Z. 32, 737–740.Google Scholar
Henderson, R. & Unwin, P. N. T. (1975). Three-dimensional model of purple membrane obtained by electron microscopy. Nature, 257, 28–32.Google Scholar
Hilderbrandt, R. L. & Kohl, D. A. (1981). A variational treatment of the effects of vibrational anharmonicity on gas-phase electron diffraction intensities. Part I. Molecular scattering function. J. Mol. Struct. Theochem. 85, 25–36.Google Scholar
Hillier, J. & Baker, R. F. (1944). Microanalysis by means of electrons. J. Appl. Phys. 15, 663–675.Google Scholar
Hirsch, P. B., Howie, A., Nicholson, R. B., Pashley, D. W. & Whelan, M. J. (1977a). Electron microscopy of thin crystals. New York: Krieger.Google Scholar
Hirsch, P. B., Howie, A., Nicholson, R. B., Pashley, D. W. & Whelan, M. J. (1977b). Electron microscopy of thin crystals, p. 190. London: Butterworth.Google Scholar
Hitchcock, A. P. (1982). Bibliography of atomic and molecular inner-shell excitation studies. J. Electron Spectrosc. Relat. Phenom. 25, 245–275. [Updated copies of this bibliography are available from the author on request.]Google Scholar
Hoerni, J. A. (1956). Multiple elastic scattering in electron diffraction by molecules. Phys. Rev. 102, 1530–1533.Google Scholar
Hofer, F., Golob, P. & Brunegger, A. (1988). EELS quantification of the elements Sr to W by means of M45 edges. Ultramicroscopy, 25, 81–84.Google Scholar
Höhberger, H. J., Otto, A. & Petri, E. (1975). Plasmon resonance in Al, deviations from quadratic dispersion observed. Solid State Commun. 16, 175–179.Google Scholar
Høier, R. (1972). Displaced lines in Kikuchi patterns. Phys. Status Solidi A, 11, 597–610.Google Scholar
Høier, R., Bakken, L. N., Marthinsen, K. & Holmestad, R. (1993). Structure factor determination in non-centrosymmetrical crystals by a two-dimensional CBED-based multi-parameter refinement method. Ultramicroscopy, 49, 159–170.Google Scholar
Holmestad, R., Weickenmeier, A. L., Zuo, J. M., Spence, J. C. H. & Horita, Z. (1993). Debye–Waller factor measurement in TiAl from HOLZ reflections. Electron microscopy and analysis 1993, pp. 141–144. Bristol: IOP Publishing.Google Scholar
Horiuchi, S. (1982). Reduction in a niobium tungsten bronze. J. Appl. Cryst. 15, 323–329.Google Scholar
Horota, F., Kakuta, N. & Shibata, S. (1981). High energy electron scattering by diborane. J. Phys. B, 14, 3299–3304.Google Scholar
Howie, A. (1963). Inelastic scattering of electrons by crystals. I. The theory of small angle inelastic scattering. Proc. R. Soc. London Ser. A, 271, 268–287.Google Scholar
Howie, A. (1979). Image contrast and localized signal selection techniques. J. Micros. (Oxford), 117, 11–23.Google Scholar
Howie, A. & Basinski, Z. S. (1968). Approximations of the dynamical theory of diffraction contrast. Philos. Mag. 17, 1039–1063.Google Scholar
Humphreys, C. J. & Hirsch, P. B. (1968). Absorption parameters in electron diffraction theory. Philos. Mag. 18, 115–122.Google Scholar
Ibach, H. & Mills, D. L. (1982). Electron energy-loss spectroscopy and surface vibrations. New York: Academic Press.Google Scholar
Ibers, J. A. (1958). Atomic scattering amplitudes for electrons. Acta Cryst. 11, 178–183.Google Scholar
Iijima, S. (1977). High resolution electron microscopy of phase objects: observation of small holes and steps on graphite crystals. Optik (Stuttgart), 47, 437–452.Google Scholar
Iijima, T., Bonham, R. A. & Ando, T. (1963). The theory of electron scattering from molecules. I. Theoretical development. J. Phys. Chem. 67, 1472–1474.Google Scholar
Inokuti, M. (1971). Inelastic collisions of fast charged particles with atoms and molecules. The Bethe theory revisited. Rev. Mod. Phys. 43, 297–344.Google Scholar
Inokuti, M., Itikawa, Y. & Turner, J. E. (1978). Addenda: Inelastic collisions of fast charged particles with atoms and molecules. The Bethe theory revisited. Rev. Mod. Phys. 50, 23–26.Google Scholar
Inokuti, M. (1979). Electron scattering cross sections pertinent to electron microscopy. Ultramicroscopy, 3, 423–427.Google Scholar
International Tables for Crystallography (2001). Vol. B, 2nd ed. Dordrecht: Kluwer Academic Publishers.Google Scholar
International Tables for X-ray Crystallography (1974). Vol. IV. Birmingham: Kynoch Press.Google Scholar
Isaacson, M. (1972a). Interaction of 24 keV electrons with the nucleic acid bases, adenine, thymine and uracil. I. Outer shell excitation. J. Chem. Phys. 56, 1803–1812.Google Scholar
Isaacson, M. (1972b). Interaction of 25 keV electrons with the nucleic acid bases, adenine, thymine and uracil. II. Inner shell excitation and inelastic scattering cross section. J. Chem. Phys. 56, 1813–1818.Google Scholar
Isaacson, M. & Johnson, D. (1975). The microanalysis of light elements using transmitted energy-loss electrons. Ultramicroscopy, 1, 33–52.Google Scholar
Ishizuka, K. (1982). Multislice formula for inclined illumination. Acta Cryst. A38, 773–779.Google Scholar
Janssen, R. W. & Sankey, O. F. (1987). Ab initio linear combination of pseudo-atomic orbital scheme for the electronic properties of semiconductors. Results for ten materials. Phys. Rev. B, 36, 6520–6531.Google Scholar
Jap, B. K. & Glaeser, R. M. (1978). The scattering of high energy electrons. Acta Cryst. A34, 94–102.Google Scholar
Johnson, D. E. (1979). Basic aspects of energy-loss spectrometer systems. Ultramicroscopy, 3, 361–365.Google Scholar
Johnson, D. E. (1980). Post specimen optics for energy loss spectrometry. Scanning Electron. Microsc. 1, 33–40.Google Scholar
Johnson, D. W. (1975). A Fourier method for numerical Kramers–Kronig analysis. J. Phys. A, 8, 490–495.Google Scholar
Johnson, D. W. & Spence, J. C. H. (1974). Determination of the single scattering probability distribution from plural scattering data. J. Phys. D, 7, 771–780.Google Scholar
Jones, P. M., Rackham, G. M. & Steeds, J. W. (1977). High-order Laue zone electron diffraction. Proc. R. Soc. London Ser. A, 354, 192–222.Google Scholar
Jouffrey, B., Sevely, J., Zanchi, G. & Kihn, Y. (1985). Characteristic energy losses with high energy electrons up to 2.5 MeV. Scanning Electron Microsc. 3, 1063–1070.Google Scholar
Kakudo, M. & Kasai, N. (1972). X-ray diffraction by polymers. Tokyo: Kodanska; Amsterdam: Elsevier.Google Scholar
Kambe, K. (1957). Study of simultaneous reflections in electron diffraction by crystals. J. Phys. Soc. Jpn, 12, 13–36.Google Scholar
Kambe, K. (1982). Visualization of Bloch waves of high energy electrons in high resolution electron microscopy. Ultramicroscopy, 10, 223–228.Google Scholar
Karle, I. L. & Karle, J. (1950). Internal motion and molecular structure studies by electron diffraction. III. Structure of CH2CF2 and CF2CF2. J. Chem. Phys. 18, 963–971.Google Scholar
Keil, P. (1968). Elektronen-Energieverlustmessungen und Berechnung optischer Konstanten. I. Festes Xenon. Z. Phys. 214, 251–265.Google Scholar
Kelley, M. H. & Fink, M. (1982a). The molecular structure of dimolybdenum tetraacetate. J. Chem. Phys. 76, 1407–1416.Google Scholar
Kelley, M. H. & Fink, M. (1982b). The temperature dependence of the molecular structure parameters in SF6. J. Chem. Phys. 77, 1813–1817.Google Scholar
Kelly, P. M., Jostsons, A., Blake, R. G. & Napier, J. G. (1975). The determination of foil thickness by scanning transmission electron microscopy. Phys. Status Solidi A, 31, 771–780.Google Scholar
Kessler, J. (1959). Einzelstreuung Mittelschneller Elektronen an Schweren Atomkernen. Z. Phys. 155, 350–367.Google Scholar
Ketkar, S. N. & Fink, M. (1985). Structure of dichromium tetraacetate by gas-phase electron diffraction. J. Am. Chem. Soc. 107, 338–340.Google Scholar
Killat, U. (1974). Optical properties of C6H12, C6H10, C6H8, C6H6, C7H8, C6H5Cl and C5H5N in the solid and gaseous state derived from electron energy losses. J. Phys. C, 7, 2396–2408.Google Scholar
Kimura, M., Schomaker, V., Smith, D. & Weinstock, B. (1968). Electron-diffraction investigation of the hexafluorides of tungsten, osmium, iridium, uranium, neptunium, and plutonium. J. Chem. Phys. 48, 4001–4012.Google Scholar
Kirkland, A., Saxton, W., Chau, K., Tsuno, K. & Kawasaki, M. (1995). Super-resolution by aperture synthesis. Ultramicroscopy, 57, 355–374.Google Scholar
Klemperer, O. (1965). Electron beam spectroscopy. Rev. Prog. Phys. 28, 77–111.Google Scholar
Kliewer, K. & Fuchs, R. (1974). Theory of dynamical properties of dielectric surfaces. Adv. Chem. Phys. 27, 355–541.Google Scholar
Kobayashi, T., Fujiyoshi, Y. & Uyeda, N. (1982). The observation of molecular orientations in crystal defects and the growth mechanism of thin phthalocyanine films. Acta Cryst. A38, 356–362.Google Scholar
Kogiso, M. & Takahashi, H. (1977). Group-theoretical method in the many-beam theory in electron diffraction. J. Phys. Soc. Jpn, 42, 223–229.Google Scholar
Kohl, D. A. & Arvedson, M. (1980). Elastic electron scattering from molecular potentials. J. Chem. Phys. 73, 3818–3822.Google Scholar
Kohl, D. A. & Bartell, L. S. (1969). Electron densities from gas-phase electron diffraction intensities. II. Molecular Hartree–Fock cross sections. J. Chem. Phys. 51, 2896–2904.Google Scholar
Kohl, D. A. & Bonham, R. A. (1967). Effect of bond formation on electron scattering cross sections for molecules. J. Chem. Phys. 47, 1634–1646.Google Scholar
Kohl, D. A. & Hilderbrandt, R. L. (1981). A variational treatment of the effects of vibrational anharmonicity on gas-phase electron diffraction intensities. Part II. Temperature dependence. J. Mol. Struct. Theochem. 85, 325–335.Google Scholar
Koike, H., Kobayashi, K., Ozawa, S. & Yagi, K. (1989). High resolution reflection electron microscopy of Si(111) 7 × 7 surfaces using a high voltage electron microscope. Jpn. J. Appl. Phys. 28, 861–865.Google Scholar
Komoda, T. (1964). On the resolution of the lattice image in the electron microscope. Optik (Stuttgart), 21, 94–110.Google Scholar
Krahl, D. & Herrmann, K.-H. (1980). Experiments with an imaging energy filter in the CTEM. Micron, 11, 287–289.Google Scholar
Krahl, D., Pätzold, H. & Swoboda, M. (1990). An aberration-minimized imaging energy filter of simple design. Proceedings of 12th International Conference on Electron Microscopy 1990, Vol. 2, pp. 60–61.Google Scholar
Krinary, G. A. (1975). On the possibilities to use oriented specimens for recording of non-basal X-ray reflexions of fine-grained layer silicates. Crystal chemistry of minerals and geological problems, pp. 132–138. Moscow: Nauka.Google Scholar
Krivanek, O. L. (1976). A method for determining the coefficient of spherical aberration from a single electron micrograph. Optik (Stuttgart), 45, 97–101.Google Scholar
Krivanek, O. L., Ahn, C. C. & Keeney, R. B. (1987). Parallel detection electron spectrometer using quadrupole lens. Ultramicroscopy, 22, 103–115.Google Scholar
Krivanek, O. L., Dellby, N., Spence, A. J., Camps, R. A. & Brown, L. M. (1997). Aberration correction in the STEM. Proc EMAG 1997, edited by S. McVitie. London: Institute of Physics.Google Scholar
Krivanek, O. L., Gubbens, A. J., Dellby, N. & Meyer, C. E. (1991). Design and first applications of a post-column imaging filter. Microsc. Microanal. Microstruct. (France), 3, 187–199.Google Scholar
Krivanek, O. L., Manoubi, T. & Colliex, C. (1985). Sub 1 eV resolution EELS at energy losses greater than 1 keV. Ultramicroscopy, 18, 155–158.Google Scholar
Krivanek, O. L. & Mooney, P. E. (1993). Applications of slow-scan CCD cameras in HREM. Ultramicroscopy, 49, 95–108.Google Scholar
Krivanek, O. L., Mooney, P. E., Fan, G. Y. Leber, M. L. & Meyer, C. E. (1991). Slow-scan CCD cameras for transmission electron microscopy. Electron microscopy and analysis 1991, pp. 523–526. Bristol: IOP Publishing.Google Scholar
Krivanek, O. L. & Swann, P. R. (1981). An advanced electron energy loss spectrometer. Quantitative microanalysis with high spatial resolution, pp. 136–140. London: The Metals Society.Google Scholar
Kröger, E. Z. (1970). Transition radiation, Čerenkov radiation and energy losses of relativistic charged particles traversing thin foils at oblique incidence. Z. Phys. 235, 403–421.Google Scholar
Kühlbrandt, W., Wang, D. N. & Fujiyoshi, Y. (1994). Atomic model of plant light-harvesting complex by electron crystallography. Nature (London), 367, 614–621.Google Scholar
Kuyatt, C. E. & Simpson, J. A. (1967). Electron monochromator design. Rev. Sci. Instrum. 38, 103–111.Google Scholar
Landau, L. & Lifchitz, E. (1966). Mécanique quantique. Théorie non relativiste, pp. 632–690. Moscow: Editions Mir.Google Scholar
Larsen, P. K. & Dobson, P. J. (1988). Editors. Reflection high energy electron diffraction and reflection electron imaging of surfaces. NATO ASI series. New York/London: Plenum Press.Google Scholar
Leapman, R. D., Grunes, L. A. & Fejes, P. L. (1982). Study of the L23 edges in the 3d transition metals and their oxides by electron energy loss spectroscopy with comparisons to theory. Phys. Rev. B, 26, 614–635.Google Scholar
Leapman, R. D., Rez, P. & Mayers, D. F. (1980). K, L and M shell generalized oscillator strengths and ionization cross sections for fast electron collisions. J. Chem. Phys. 72, 1232–1243.Google Scholar
Leapman, R. D. & Swyt, C. R. (1988). Separation of overlapping core edges in EELS spectra by multiple least-squares fitting. Ultramicroscopy, 26, 393–404.Google Scholar
Lehmpfuhl, G. (1974). Dynamical interaction of electron waves in a perfect single crystal. Z. Naturforsch. Teil A, 27, 424–433.Google Scholar
Lichte, H. (1991). Electron image plane off-axis holography of atomic structures. Adv. Opt. Electron Microsc. 12, 25–91.Google Scholar
Lindhard, J. (1954). On the properties of a gas of charged particles. Dan. Vidensk. Selsk. Mater. Fys. Medd. 28, 1–57.Google Scholar
Lindner, T., Sauer, H., Engel, W. & Kambe, K. (1986). Near-edge structure in electron energy loss spectra of MgO. Phys. Rev. B, 33, 22–24.Google Scholar
Liu, J. W. & Smith, V. H. (1977). A critical study of high energy electron scattering from H2. Chem. Phys. Lett. 45, 59–63.Google Scholar
Livingood, J. J. (1969). The optics of dipole magnets. New York: Academic Press.Google Scholar
Lovey, F. C., Coene, W., Van Dyck, D., Van Tendeloo, G., Van Landuyt, J. & Amelinckx, S. (1984). HREM imaging conditions for stacking sequences in 18R martensite of Cu–Al alloys. Ultramicroscopy, 15, 345–356.Google Scholar
Lynch, D. F., Moodie, A. F. & O'Keefe, M. A. (1975). n-Beam lattice images. V. The use of the charge-density approximation in the interpretation of lattice images. Acta Cryst. A31, 300–307.Google Scholar
Lytle, F. W., Greegor, R. B. & Panson, A. Y. (1988). Discussion of X-ray absorption near edge structure: application to Cu in the high Tc superconductors La1.8Sr0.2Cu4 and YBa2Cu3O7. Phys. Rev. B, 37, 1550–1562.Google Scholar
Ma, Y., Rømming, C., Lebech, B. & Gjønnes, J. (1992). Structure refinement of Al3Zr using single-crystal X-ray diffraction, powder neutron diffraction and CBED. Acta Cryst. B48, 11–16.Google Scholar
McClelland, J. J. & Fink, M. (1985). Electron correlation and binding effects in measured electron-scattering cross sections of CO2. Phys. Rev. Lett. 54, 2218–2221.Google Scholar
MacGillavry, C. H. (1940). Zur Prufung der Dynamischen Theorie der Elektronenbeugung an Kristallgitter. Physica (Utrecht), 7, 329–343.Google Scholar
McLean, A. D. & McLean, R. S. (1981). Roothaan–Hartree–Fock atomic wave-functions: Slater basis-set expansions for Z = 55–92. At. Data Nucl. Data Tables, 26, 197–381.Google Scholar
Maher, D. M. (1979). Elemental analysis using inner-shell excitations: a microanalytical technique for materials characterization. Introduction to analytical electron microscopy, edited by J. J. Hren, J. I. Goldstein & D. C. Joy, pp. 259–294. New York: Plenum.Google Scholar
Mamy, J. & Gaultier, J.-P. (1976). Evolution structurale de la montmorillonite associée au phénomène de fixation irréversible du potassium. An. Agron. 27(1), 1–16.Google Scholar
Mann, J. B. (1968). Los Alamos Scientific Laboratory Report LA3691, p. 168.Google Scholar
Manoubi, T., Rez, P. & Colliex, C. (1989). Quantitative electron energy loss spectroscopy on M45 edges in rare earth oxides. J. Electron Spectrosc. Relat. Phenom. 50, 1–18.Google Scholar
Manoubi, T., Tence, M., Walls, M. G. & Colliex, C. (1990). Curve fitting methods for quantitative analysis in EELS. Microsc. Microanal. Microstruct. 1, 23–39.Google Scholar
Manzke, R. (1980). Wavevector dependence of the volume plasmon of GaAs and InSb. J. Phys. C, 13, 911–917.Google Scholar
Marks, L. (1986). High resolution electron microscopy of surfaces. In Topics in current physics, Vol. 41. Structure and dynamics of surfaces. I, edited by W. Schommers and P. Von Blackenhagen. Berlin/Heidelberg: Springer Verlag.Google Scholar
Marthinsen, K., Holmestad, R. & Høier, R. (1994). Analytical filtering of low-angle inelastic scattering contributions to CBED contrast. Ultramicroscopy, 55, 268–275.Google Scholar
Maslen, V. M. & Rossouw, C. J. (1983). The inelastic scattering matrix element and its application to electron energy loss spectroscopy. Philos. Mag. A47, 119–130.Google Scholar
Matsuhata, H. & Gjønnes, J. (1994). Bloch-wave degeneracies and non-systematic critical voltage: a method for structure-factor determination. Acta Cryst. A50, 107–115.Google Scholar
Matsuhata, H. & Steeds, J. W. (1987). Observation of accidental Bloch-wave degeneracies of zone-axis critical voltage. Philos. Mag. B55, 39–54.Google Scholar
Matsuhata, H., Tomokiyo, Y., Watanabe, H. & Eguchi, T. (1984). Determination of the structure factors of Cu and Cu3Au by the intersecting Kikuchi-line method. Acta Cryst. B40, 544–549.Google Scholar
Mawhorter, R. J. & Fink, M. (1983). The vibrationally averaged, temperature-dependent structure of polyatomic molecules. II. SO2. J. Chem. Phys. 79, 3292–3296.Google Scholar
Mawhorter, R. J., Fink, M. & Archer, B. T. (1983). The vibrationally averaged, temperature-dependent structure of polyatomic molecules. I. CO2. J. Chem. Phys. 79, 170–174.Google Scholar
Melkanov, M. A., Sawada, T. & Raynal, J. (1966). Nuclear optical model calculations. Methods Comput. Phys. 6, 1–80.Google Scholar
Menter, J. W. (1956). The resolution of crystal lattices. Proc. R. Soc. London Ser. A, 236, 119.Google Scholar
Méring, J. (1949). L'interférence des rayons X dans les systèmes à stratification désordonnée. Acta Cryst. 2, 371–377.Google Scholar
Metherell, A. J. F. (1971). Energy analysing and energy selecting electron microscopes. Adv. Opt. Electron Microsc. 4, 263–361.Google Scholar
Miller, B. R. & Fink, M. (1981). Mean amplitudes of vibration of SF6 and intramolecular multiple scattering. J. Chem. Phys. 75, 5326–5328.Google Scholar
Miller, B. R. & Fink, M. (1985). The vibrationally averaged, temperature-dependent structure of polyatomic molecules. III. NO2. J. Chem. Phys. 83, 939–944.Google Scholar
Möllenstedt, G. & Düker, H. (1956). Beobachtungen und Messungen an Biprisma-Interferenzen mit Elektronenwellen. Z. Phys. 145, 377–397.Google Scholar
Moodie, A. F. & Warble, C. E. (1967). The observation of primary step growth in magnesium oxide by direct transmission electron microscopy. Philos. Mag. 16, 891–904.Google Scholar
Mori, N., Oikawa, T & Harada, Y. (1990). Development of the imaging plate for the transmission electron microscope and its characteristics. J. Electron Microsc. (Japan), 39, 433–436.Google Scholar
Morse, P. M. (1932). Unelastische Streuung von Kathodenstrahlen. Phys. Z. 33, 443–445.Google Scholar
Mory, C. & Colliex, C. (1989). Elemental analysis near the single-atom detection level by processing sequences of energy-filtered images. Ultramicroscopy, 28, 339–346.Google Scholar
Mott, N. F. & Massey, H. S. W. (1952). The theory of atomic collisions, pp. 224–248. Oxford: Clarendon Press.Google Scholar
Mott, N. F. & Massey, H. S. W. (1965). The theory of atomic collisions, 3rd ed., Chap. IX, Section 4, equations (22) and (23). Oxford University Press.Google Scholar
Müller, J. E., Jepsen, O. & Wilkins, J. W. (1982). X-ray absorption spectra: K edges of 3d transition metals, L edges of 3d and 4d metals and M edges of palladium. Solid State Commun. 42, 365–368.Google Scholar
Nagakura, S., Nakamura, Y. & Suzuki, T. (1982). Forbidden reflection intensity in electron diffraction and its influence on the crystal structure image. Jpn. J. Appl. Phys. 21, L449–L451.Google Scholar
Nellist, P., McCallum, B. & Rodenburg, J. (1995). Resolution beyond the information limit in STEM. Nature (London), 374, 630–632.Google Scholar
Numerov, B. V. (1924). A method of extrapolation of perturbations. Mon. Not. R. Astron. Soc. 84, 592–601.Google Scholar
O'Keefe, M. A., Spence, J. C. H., Hutchinson, J. L. & Waddington, W. G. (1985). Proc. 43rd EMSA Meeting, p. 64. San Francisco: San Francisco Press. [See also H. Hashimoto in Ultramicroscopy (1985), 18, 19–32.]Google Scholar
Olsen, A., Goodman, P. & Whitfield, H. (1985). Tl3SbS3, Tl3SbSe3, Tl3Sb3−xSex and Tl3SbyAs1−ySe3. J. Solid State Chem. 60, 305–315.Google Scholar
Olsen, A. & Spence, J. C. H. (1981). Distinguishing dissociated glide and shuffle set dislocations by high resolution electron microscopy. Philos. Mag. A43, 945–965.Google Scholar
Orchowski, A., Rau, W. D. & Lichte, H. (1995). Electron holography surmounts resolution limit of electron microscopy. Phys. Rev. Lett. 74, 399.Google Scholar
Parker, N. W., Utlaut, M. & Isaacson, M. S. (1978). Design of magnetic spectrometers with second order aberrations corrected. Optik (Stuttgart), 51, 333–351.Google Scholar
Pearce-Percy. H. T. (1978). The design of spectrometers for energy loss spectroscopy. Scaning Electron Microsc. 1, 41–51.Google Scholar
Peixoto, E. M. A., Bunge, C. F. & Bonham, R. A. (1969). Elastic and inelastic electron scattering by He and Ne atoms in their ground states. Phys. Rev. 181, 322–328.Google Scholar
Pendry, J. B. (1974). Low energy electron diffraction. New York: Academic Press.Google Scholar
Peng, L.-M. (1998). Electron scattering factors of ions and their parameterization. Acta Cryst. A54, 481–485.Google Scholar
Peng, L. M. & Cowley, J. M. (1988). Errors arising from numerical use of the Mott formula in electron image simulation. Acta Cryst. A44, 1–5.Google Scholar
Peng, L.-M., Ren, G., Dudarev, S. L. & Whelan, M. J. (1996). Robust parameterization of elastic and absorptive electron atomic scattering factors. Acta Cryst. A52, 257–276.Google Scholar
Pennycook, S. J. & Jesson, D. E. (1991). High-resolution Z-contrast imaging of crystals. Ultramicroscopy, 37, 14–38.Google Scholar
Pinsker, Z. G. (1953). Electron diffraction. London: Butterworth.Google Scholar
Pirouz, P. (1974). Effects of absorption on lattice images. Optik (Stuttgart), 54, 69–74.Google Scholar
Plançon, A., Rousseaux, F., Tchoubar, D., Tchoubar, C., Krinari, G. & Drits, V. A. (1982). Recording and calculation of hk rod intensities in case of diffraction by highly oriented powders of lamellar samples. J. Appl. Cryst. 15, 509–512.Google Scholar
Powell, C. J. (1976). Cross sections for ionization of inner-shell electrons by electrons. Rev. Mod. Phys. 48, 33–47.Google Scholar
Powell, C. J. (1984). Inelastic scattering of electrons in solids. Electron beam interactions with solids for microscopy, microanalysis and micro-lithography, edited by D. F. Kyser, H. Niedrig, D. E. Newbury & R. Shimizu, pp. 19–31. Chicago: SEM, Inc.Google Scholar
Powell, C. J. (1989). Cross sections for inelastic electron scattering in solids. Ultramicroscopy, 28, 24–31.Google Scholar
Pulay, P., Mawhorter, R. J., Kohl, D. A. & Fink, M. (1983). Ab initio Hartree–Fock calculation of the elastic electron scattering cross section of sulphur hexafluoride. J. Chem. Phys. 79, 185–191.Google Scholar
Raether, H. (1965). Electron energy loss spectroscopy. Springer Tracts Mod. Phys. Vol. 38, pp. 85–170. Berlin: Springer.Google Scholar
Raether, H. (1980). Excitation of plasmons and interband transitions by electrons. Spring Tracts Mod. Phys. Vol. 88. Berlin: Springer.Google Scholar
Rao, C. N., Thomas, J. M., Williams, B. G. & Sparrow, T. G. (1984). Determination of the number of d-electron states in transition metal compounds. J. Phys. Chem. 88, 5769–5770.Google Scholar
Rask, J. H., Miner, B. A. & Buseck, P. (1987). Determination of manganese oxidation states in solids by EELS. Ultramicroscopy, 21, 321–326.Google Scholar
Reimer, L. & Rennekamp, R. (1989). Imaging and recording of multiple scattering effects by angular resolved electron energy loss spectroscopy. Ultramicroscopy, 28, 258–265.Google Scholar
Rez, D., Rez, P. & Grant, I. (1994). Dirac–Fock calculations of X-ray scattering factors and contributions to the mean inner potential for electron scattering. Acta Cryst. A50, 481–497.Google Scholar
Rez, P. (1989). Inner shell spectroscopy: an atomic view. Ultramicroscopy, 28, 16–23.Google Scholar
Ritchie, R. H. (1957). Plasmon losses by fast electrons in thin films. Phys. Rev. 106, 874–881.Google Scholar
Rose, H. & Plies, E. (1974). Entwurf eines fehlerarmen magnetischen Energie Analysators. Optik (Stuttgart), 40, 336–341.Google Scholar
Rose, H. & Spehr, R. (1980). On the theory of the Boersch effect. Optik (Stuttgart), 57, 339–364.Google Scholar
Ross, A. W. & Fink, M. (1986). Atomic scattering factor and spin polarization calculations. Phys. Rev. 85, 6810–6811.Google Scholar
Sasaki, H., Konaka, S., Iijima, T. & Kimura, M. (1982). Small-angle electron scattering and electron density in carbon dioxide. Int. J. Quantum Chem. 21, 475–485.Google Scholar
Saunders, M., Bird, D. M., Midgley, P. A. & Vincent, R. (1994). Structure factor refinement by zone-axis CBED pattern matching. Proceedings of 13th International Congress on Electron Microscopy, Paris, France, 17–22 July 1994, Vol. 1, pp. 847–848.Google Scholar
Saxton, W. O. (1978). Computer techniques for image processing in electron microscopy, pp. 9–19. New York: Academic Press.Google Scholar
Saxton, W. O. (1980a). Recovery of specimen information for strongly scattering objects. In Computer processing of electron microscopy images. Topics in current physics, Vol. 13, edited by P. W. Hawkes, p. 35. Berlin/Heidelberg/New York: Springer Verlag.Google Scholar
Saxton, W. O. (1980b). Correction of artifacts in linear and nonlinear high resolution electron micrographs. J. Microsc. Spectrosc. Electron, 5, 665–674.Google Scholar
Sayers, D. E., Stern, E. A. & Lytle, F. M. (1971). New technique for investigating noncrystalline structures: Fourier analysis of the extended X-ray absorption fine structure. Phys. Rev. Lett. 27, 1204–1207.Google Scholar
Schäfer, L. & Seip, H. M. (1967). Studies on the failure of the first Born approximation in electron diffraction. VI. Ruthenium tetraoxide. Acta Chem. Scand. 21, 737–744.Google Scholar
Schattschneider, P. (1983). A performance test of the recovery of single energy loss profiles via matrix analysis. Ultramicroscopy, 11, 321–322.Google Scholar
Schattschneider, P. (1989). The dielectric description of inelastic electron scattering. Ultramicroscopy, 28, 1–15.Google Scholar
Scheinfein, M. & Isaacson, M. S. (1984). Design and performance of second order aberration corrected spectrometers for use with the scanning transmission electron microscope. Scanning Electron Microsc. 4, 1681–1696.Google Scholar
Scheinfein, M. & Isaacson, M. S. (1986). Electronic and chemical analysis of fluoride interface structures at subnanometer spatial resolution. J. Vac. Sci. Technol. B4, 326–332.Google Scholar
Scherzer, O. (1949). The theoretical resolution limit of the electron microscope. J. Appl. Phys. 20, 20–29.Google Scholar
Schiske, P. (1975). Phase determination from a focal series and the corresponding diffraction pattern in electron microscopy for strongly scattering objects. J. Phys. D, 8, 1372–1386.Google Scholar
Schnatterly, S. E. (1979). Inelastic electron scattering spectroscopy. Solid State Phys. 14, 275–358.Google Scholar
Schomaker, V. & Glauber, R. (1952). The Born approximation in electron diffraction. Nature (London), 170, 290–291.Google Scholar
Schröder, B. & Geiger, J. (1972). Electron spectrometric study of amorphous germanium and silicon in the two phonon region. Phys. Rev. Lett. 28, 301–303.Google Scholar
Seip, H. M. (1965). Studies on the failure of the first Born approximation in electron diffraction. I. Uranium hexafluoride. Acta Chem. Scand. 19, 1955–1968.Google Scholar
Seip, H. M. & Seip, R. (1966). Studies on the failure of the first Born approximation in electron diffraction. IV. Molybdenum- and tungsten hexafluoride. Acta Chem. Scand. 20, 2698–2711.Google Scholar
Seip, H. M. & Stølevik, V. (1966a). Studies on the failure of the first Born approximation in electron diffraction. II. Osmium tetraoxide. Acta Chem. Scand. 20, 385–394.Google Scholar
Seip, H. M. & Stølevik, V. (1966b). Studies on the failure of the first Born approximation in electron diffraction. III. Tellurium hexafluoride. Acta Chem. Scand. 20, 1535–1545.Google Scholar
Self, P. G., O'Keefe, M. A., Buseck, P. R. & Spargo, A. E. C. (1983). Practical computation of amplitudes and phases in electron diffraction. Ultramicroscopy, 11, 35–52.Google Scholar
Sevely, J. (1985). Voltage dependence in electron energy loss spectroscopy. Inst. Phys. Conf. Ser. 78, 155–160.Google Scholar
Shibata, S., Hirota, F., Kakuta, N. & Muramatsu, T. (1980). Electron distribution in water by high-energy electron scattering. Int. J. Quantum Chem. 18, 281–285.Google Scholar
Shiles, E., Sazaki, T., Inokuti, M. & Smith, D. Y. (1980). Self consistency and sum-rule tests in the Kramers Kronig analysis of optical data: applications to aluminium. Phys. Rev. B, 22, 1612–1628.Google Scholar
Shindo, D., Hiraga, K., Oikawa, T. & Mori, N. (1990). Quantification of electron diffraction with imaging plate. J. Electron Microsc. 39, 449–453.Google Scholar
Shuman, H. (1980). Correction of the second order aberrations of uniform field magnetic sections. Ultramicroscopy, 5, 45–53.Google Scholar
Shuman, H. & Kruit, P. (1985). Quantitative data processing of parallel recorded electron energy-loss spectra with low signal to background. Rev. Sci. Instrum. 56, 231–239.Google Scholar
Shuman, H. & Somlyo, A. P. (1987). Electron energy loss analysis of near-trace-element concentrations of calcium. Ultramicroscopy, 21, 23–32.Google Scholar
Smith, D. J., Bursill, L. A. & Wood, G. J. (1985). Non-anomalous high-resolution imaging of crystalline materials. Ultramicroscopy, 16, 19–32.Google Scholar
Smith, D. J., Saxton, W. O., O'Keefe, M. A., Wood, G. J. & Stobbs, W. M. (1983). The importance of beam alignment and crystal tilt in high resolution electron microscopy. Ultramicroscopy, 11, 263–282.Google Scholar
Spargo, A. E. C. (1994). Electron crystallography and crystal structure. Proceedings of 13th International Congress on Electron Microscopy, Paris, France 17–22 July 1994, Vol. 1, pp. 959–960.Google Scholar
Spence, J. C. H. (1979). Uniqueness and the inversion problem of incoherent multiple scattering. Ultramicroscopy, 4, 9–12.Google Scholar
Spence, J. C. H. (1981). The crystallographic information in localized characteristic-loss electron images and diffraction patterns. Ultramicroscopy, 7, 59–64.Google Scholar
Spence, J. C. H. (1988a). Inelastic electron scattering: Parts I and II. High resolution transmission electron microscopy and associated techniques, edited by P. R. Buseck, J. M. Cowley & L. Eyring, pp. 129–189. Oxford University Press.Google Scholar
Spence, J. C. H. (1988b). High resolution electron microscopy, 2nd ed. New York: Oxford University Press.Google Scholar
Spence, J. C. H. (1998). Direct inversion of dynamical electron diffraction patterns to structure factors. Acta Cryst. A54, 7–18.Google Scholar
Spence, J. C. H. & Lynch, J. (1982). STEM microanalysis and inelastic imaging in crystals. Ultramicroscopy, 9, 267–278.Google Scholar
Spence, J. C. H., O'Keefe, M. A. & Iijima, S. (1978). On the thickness periodicity of atomic-resolution images of dislocation cores. Philos. Mag. A38, 463–482.Google Scholar
Spence, J. C. H. & Tafto, J. (1983). ALCHEMI: a new technique for locating atoms in small crystals. J. Microsc. 130, 147–154.Google Scholar
Spence, J. C. H. & Zuo, J. M. (1992). Electron microdiffraction. New York: Plenum.Google Scholar
Steeds, J. W. (1984). Further development in the analysis of convergent beam electron diffraction (CBED) data. EMAG 1983. Inst. Phys. Conf. Ser. No. 69, pp. 31–36.Google Scholar
Stohr, J., Sette, F. & Johnson, A. L. (1984). Near edge X-ray absorption fine structure studies of chemisorbed hydrocarbons: bond lengths with a ruler. Phys. Rev. Lett. 53, 1684–1687.Google Scholar
Strauss, M. G., Naday, Y., Sherman, I. S. & Zaluzec, N. J. (1987). CCD base parallel detection system for EELS spectroscopy and imaging. Ultramicroscopy, 22, 117–124.Google Scholar
Sturm, K. (1982). Electron energy loss in simple metals and semiconductors. Adv. Phys. 31, 1–64.Google Scholar
Swyt, C. R. & Leapman, R. D. (1982). Plural scattering in electron energy loss analysis. Scanning Electron Microsc. 1, 73–82.Google Scholar
Taft, E. A. & Philipp, H. R. (1965). Optical properties of graphite. Phys. Rev. A, 138, 197–202.Google Scholar
Taftø, J. & Gjønnes, J. (1985). The intersecting Kikuchi line technique: critical voltage at any voltage. Ultramicroscopy, 17, 329–334.Google Scholar
Tafto, J. & Krivanek, O. L. (1982). Site specific valence determination by EELS. Phys. Rev. Lett. 48, 560–563.Google Scholar
Taftø, J. & Metzger, T. H. (1985). Large-angle convergent-beam electron diffraction; a simple technique for the study of modulated structures with application to V2D. J. Appl. Cryst. 18, 110–113.Google Scholar
Tafto, J. & Zhu, J. (1982). Electron energy-loss near edge structure (ELNES), a potential technique in the studies of local atomic arrangements. Ultramicroscopy, 9, 349–354.Google Scholar
Takayanagi, K. (1984). Surface structure imaging by electron microscopy. J. Microsc. 136, 287–298.Google Scholar
Tanaka, M. & Tsuda, K. (1990). Determination of positional parameters by convergent-beam electron diffraction. Proceedings of 12th International Congress on Electron Microscopy 1990, Vol. 2, pp. 518–519.Google Scholar
Tavard, C. & Bonham, R. A. (1969). Quantum-mechanical formulas for the inelastic scattering of fast electrons and their Compton line shape. Nonrelativistic approximation. J. Chem. Phys. 50, 1736–1747.Google Scholar
Tavard, C., Rouault, M. & Roux, M. (1965). Diffraction des rayons X et des électrons par les molécules. III. Une méthode de detémination des densités électroniques moléculaires. J. Chim. Phys. 62, 1410–1417.Google Scholar
Tavard, C. & Roux, M. (1965). Calcul des intensités de diffraction de rayons X et de électrons par les molécules. C. R. Acad. Sci. 260, 4933–4936.Google Scholar
Teo, B. K. & Joy, D. C. (1981). EXAFS spectroscopy techniques and applications. New York: Plenum.Google Scholar
Terasaki, O., Watanabe, D. & Gjønnes, J. (1979). Determination of crystal structure factor of Si by the intersecting-Kikuchi-line method. Acta Cryst. A35, 895–900.Google Scholar
Thakkar, A. J. & Smith, V. H. Jr (1978). Form factors and total scattering intensities for the helium-like ions from explicitly correlated wavefunctions. J. Phys. B, 11, 3803–3820.Google Scholar
Thole, B. T., van der Laan, G., Fuggle, J. C., Sawatzky, G. A., Karnatak, R. C. & Esteva, J.-M. (1985). 3d X-ray absorption lines and the multiplets of the lanthanides. Phys. Rev. B, 32, 5107–5118.Google Scholar
Tonomura, A. (1992). Electron-holographic interference microscopy. Adv. Phys. 41, 59–103.Google Scholar
Tossell, J. A., Vaughan, D. J. & Johnson, K. H. (1974). The electronic structure of rutile, wustite and hematite from molecular orbital calculations. Am. Mineral. 59, 319–334.Google Scholar
Treacy, M. M. J. & Rice, S. B. (1989). Catalyst particle sizes from Rutherford scattered intensities. J. Microsc. 156, 211–234.Google Scholar
Trebbia, P. (1988). Unbiased method for signal estimation in EELS. Concentration measurements and detection limits in quantitative analysis: methods and programs. Ultramicroscopy, 24, 399–408.Google Scholar
Unwin, P. N. T. & Henderson, R. (1975). Molecular structure determination by electron microscopy of unstained crystalline specimens. J. Mol. Biol. 94, 425–432.Google Scholar
Vainshtein, B. K. (1964). Structure analysis by electron diffraction. Oxford: Pergamon Press.Google Scholar
Van Dyck, D. (1980). Fast computational procedures for the simulation of structure images in complex or disordered crystals. J. Microsc. 119, 141.Google Scholar
Van Dyck, D., Op de Beeck, M. & Coene, W. M. J. (1994). Information in electron microscopy. Microsc. Soc. Am. Bull. 24, 427–437.Google Scholar
Venghaus, H. (1975). Redetermination of the dielectric function of graphite. Phys. Status Solidi B, 71, 609–614.Google Scholar
Vincent, R., Bird, D. M. & Steeds, J. W. (1984). Structure of AuGeAs determined by convergent beam electron diffraction. II. Refinement of structural parameters. Philos. Mag. A50, 765–786.Google Scholar
Vincent, R. & Midgley, P. A. (1994). Double conical beam-rocking system for measurement of integrated electron diffraction intensities. Ultramicroscopy, 53, 271–284.Google Scholar
Voigt-Martin, I. G., Yan, D. H., Gilmore, C. J., Shankland, K. & Bricogne, G. (1994). The use of maximum entropy and likelihood ranking to determine the crystal structure of 4-[4′-(N-dimethylamino)benzylidene]pyrazolidine-3,5-dione at 1.4 Å resolution from electron diffraction and high-resolution electron microscopy image data. Ultramicroscopy, 56, 271–288.Google Scholar
Von Festenberg, C. (1968). Retardierungseffekte im Energieverlustspektrum von GaP. Z. Phys. 214, 464.Google Scholar
Voss, R., Lehmpfuhl, G. & Smith, D. J. (1980). Influence of doping on the crystal potential of silicon investigated by the convergent beam electron diffraction technique. Z. Naturforsch. Teil A, 35, 973–984.Google Scholar
Vvedensky, D. D., Saldin, D. K. & Pendry, J. B. (1985). Azimuthal and polar angle dependence in XANES of low symmetry adsorption sites. Surf. Sci. 162, 909–912.Google Scholar
Wade, R. H. & Frank, J. (1977). Electron microscope transfer functions for partially coherent axial illumination. Optik (Stuttgart), 49, 81–92.Google Scholar
Wang, J., Esquivel, R. O., Smith, V. H. Jr & Bunge, C. (1995). Accurate elastic and inelastic scattering factors from He to Ne using correlated wavefunctions. Phys. Rev. A, 51, 3812–3818.Google Scholar
Wang, J., Sagar, R. P., Schmider, H. & Smith, V. H. Jr (1993). X-ray elastic and inelastic scattering factors for neutral atoms Z = 2–92. At. Data Nucl. Data Tables, 53, 233–269.Google Scholar
Wang, J., Tripathi, A. N. & Smith, V. H. Jr (1994). Chemical binding and electron correlation effects in X-ray and high energy electron scattering. J. Chem. Phys. 101, 4842–4854.Google Scholar
Warren, B. E. (1941). X-ray diffraction in random layer lattices. Phys. Rev. 59, 693–698.Google Scholar
Watanabe, D., Uyeda, R. & Fukuhara, A. (1969). Determination of the atomic form factor by high-voltage electron diffraction. Acta Cryst. A25, 138–140.Google Scholar
Wehenkel, C. (1975). Mise au point d'une nouvelle méthode d'analyse quantitative des spectres de pertes d'énergie d'électrons rapides diffusés dans la direction du faisceau incident: application à l'étude des métaux nobles. J. Phys. (Paris), 36, 199–207.Google Scholar
Weickenmeier, A. & Kohl, H. (1991). Computation of absorptive form factors for high-energy electron diffraction. Acta Cryst. A47, 590–597.Google Scholar
Weng, X. D. & Rez, P. (1989). Multiple scattering approach to oxygen K near edge structures in EELS spectroscopy of alkaline earths. Phys. Rev. B, 39, 7405–7412.Google Scholar
Weng, X. D., Rez, P. & Ma, H. (1989). Carbon K-shell near-edge structure: multiple scattering and band theory calculations. Phys. Rev. B. 40, 4175–4178.Google Scholar
Weng, X. D., Rez, P. & Sankey, O. F. (1989). Pseudo-atomic orbital band theory applied to EELS near edge structures. Phys. Rev. B, 40, 5694–5704.Google Scholar
Wien, W. (1897). Vert. Deutschen Phys. Res. 16, 165.Google Scholar
Williams, B. G. & Bourdillon, A. J. (1982). Localised Compton scattering using EELS. J. Phys. C, 15, 6881–6890.Google Scholar
Williams, B. G., Sparrow, T. G. & Egerton, R. F. (1984). Electron Compton scattering from solids. Proc. R. Soc. London Ser. A, 393, 409–422.Google Scholar
Williams, D. B. & Carter, C. B. (1996). Transmission electron microscopy. New York: Plenum Press.Google Scholar
Wilson, A. J. C. (1949). Diffraction by random layers: ideal line profiles and determination of structure amplitudes from observed line profiles. Acta Cryst. 2, 245–251.Google Scholar
Wilson, A. J. C. (1962). X-ray optics, 2nd ed. London: Methuen.Google Scholar
Wilson, A. R., Spargo, A. E. C. & Smith, D. J. (1982). The characterisation of instrumental parameters in the high resolution electron microscope. Optik (Stuttgart), 61, 63–78.Google Scholar
Wong, T. C. & Bartell, L. S. (1973). Three atom scattering in gas-phase electron diffraction. II. A general treatment. J. Chem. Phys. 58, 5654–5660.Google Scholar
Xie, S.-D., Fink, M. & Kohl, D. A. (1984). Basis set dependence of ab initio SCF elastic, Born, electron scattering cross sections for C2H4. J. Chem. Phys. 81, 1940–1942.Google Scholar
Yagi, K. (1993). RHEED and REM. In Electron diffraction techniques, Vol. 2, edited by J. M. Cowley. IUCr/Oxford University Press.Google Scholar
Yagi, K. & Cowley, J. M. (1978). Electron microscopy study of ordering of potassium ions in cubic KSbO3. Acta Cryst. A34, 625–634.Google Scholar
Yates, A. C. (1970). Relativistic effects in high-energy inelastic electron–atom collisions. Chem. Phys. Lett. 6, 49–53.Google Scholar
Yates, A. C. (1971). A program for calculating relativistic elastic electron–atom collision data. Comput. Phys. Commun. 2, 175–179.Google Scholar
Yates, A. C. & Bonham, R. A. (1969). Use of relativistic electron scattering factors in electron diffraction analysis. J. Chem. Phys. 50, 1056–1058.Google Scholar
Zaanen, J., Sawatzky, G. A., Fink, J., Speier, W. & Fuggle, J. C. (1985). L23 absorption spectra of the lighter 3d transition metals. Phys. Rev. B, 32, 4905–4913.Google Scholar
Zakharov, N. D., Pasemann, M. & Rozhanski, V. N. (1982). Observations of point defects in silicon by means of dark-field lattice plane imaging. Phys. Status Solidi A, 71, 275–281.Google Scholar
Zanchi, G., Perez, J. P. & Sevely, J. (1975). Adaptation of magnetic filtering device on a one megavolt electron microscope. Optik (Stuttgart), 43, 495–501.Google Scholar
Zanchi, G., Sevely, J. & Jouffrey, B. (1977). Second order image aberration of a one megavolt magnetic filter. Optik (Stuttgart), 48, 173–192.Google Scholar
Zhukhlistov, A. P., Avilov, A. S., Ferraris, G., Zvyagin, B. B. & Plotnikov, V. P. (1997). Statistical distribution of hydrogen over three positions in brucite Mg(OH)2 structure from electron diffractometry data. Crystallogr. Rep. 42, 774–777.Google Scholar
Zimmermann, S. (1976). The dielectric function of InSb determined by electron energy losses. J. Phys. C, 9, 2643–2649.Google Scholar
Zou, X. D., Sukharev, Y. & Hovmöller, S. (1993). ELD – a computer program for extracting intensities from electron diffraction patterns. Ultramicroscopy, 49, 147–158.Google Scholar
Zuo, J. M., Høier, R. & Spence, J. C. H. (1989). Three-beam and many-beam theory in electron diffraction and its use for structure-factor phase determination in non-centrosymmetrical crystal structures. Acta Cryst. A45, 839–851.Google Scholar
Zuo, J. M., Spence, J. C. H., Downs, J. & Mayer, J. (1993). Measurement of individual structure-factor phases with tenth-degree accuracy: the 00.2 reflection in BeO studied with electron and X-ray diffraction. Acta Cryst. A49, 422–429.Google Scholar
Zuo, J. M., Spence, J. C. H. & O'Keefe, M. (1988). Bonding in GaAs. Phys. Rev. Lett. 61, 353–356.Google Scholar
Zvyagin, B. B. (1967). Electron diffraction analysis of clay mineral structures. New York: Plenum.Google Scholar
Zvyagin, B. B., Vrublevskaya, Z. V., Zhukhlistov, A. P., Sidorenko, O. V., Soboleva, S. V. & Fedotov, A. F. (1979). High-voltage electron diffraction in the study of layered minerals. Moscow: Nauka.Google Scholar
Zvyagin, B. B., Zhukhlistov, A. P. & Plotnikov, V. P. (1996). The development of electron diffractometry of minerals. Structural studies of crystals (For the 75th birthday of B. K. Vainshtein), pp. 225–234. Moscow: Nauka Physmathgis.Google Scholar