TY - JOUR
AU - Liu,, Jingyue
AB - Abstract Scanning transmission electron microscopy (STEM) techniques can provide imaging, diffraction and spectroscopic information, either simultaneously or in a serial manner, of the specimen with an atomic or a sub-nanometer spatial resolution. High-resolution STEM imaging, when combined with nanodiffraction, atomic resolution electron energy-loss spectroscopy and nanometer resolution X-ray energy dispersive spectroscopy techniques, is critical to the fundamental studies of importance to nanoscience and nanotechnology. The availability of sub-nanometer or sub-angstrom electron probes in a STEM instrument, due to the use of a field emission gun and aberration correctors, ensures the greatest capabilities for studies of sizes, shapes, defects, crystal and surface structures, and compositions and electronic states of nanometer-size regions of thin films, nanoparticles and nanoparticle systems. The various imaging, diffraction and spectroscopy modes available in a dedicated STEM or a field emission TEM/STEM instrument are reviewed and the application of these techniques to the study of nanoparticles and nanostructured catalysts is used as an example to illustrate the critical role of the various STEM techniques in nanotechnology and nanoscience research. electron microscopy, STEM, Z-contrast microscopy, nanodiffraction, SEM, EELS, EDS, Auger, nanoparticle, supported catalyst, surface Introduction Advanced electron microscopy techniques, especially scanning transmission electron microscopy (STEM) techniques, are indispensable for characterizing interfaces and defects, nanodevices, nanoparticles and catalysts, and other nanosystems. The single most important feature of a STEM instrument is its versatility: atomic resolution images, diffraction patterns from nanometer regions and nanometer-scale spectroscopy data can be obtained either simultaneously or sequentially from the same region of the specimen. The availability of the various imaging, diffraction, and spectroscopy techniques within a single instrument makes STEM the most powerful microscope for characterizing the physicochemical nature of nanoscale systems. When an electron nanoprobe interacts with a specimen inside a STEM instrument, a variety of electron, electromagnetic and other signals can be generated. Figure 1 shows a schematic diagram illustrating the common signals that are used in a dedicated STEM instrument. All these signals can be used to form images or diffraction patterns of the specimen or can be analyzed to provide spectroscopic information. For example, by collecting high-angle scattered electrons with an annular detector, high-angle annular dark-field (HAADF) images (also called Z-contrast images) can be formed to provide information about structural variations across the sample on an atomic level. Electron energy-loss spectroscopy (EELS), which is based on the energy analysis of the inelastically scattered electrons, can provide information on the electronic structure, oxidation states, and chemical composition on an atomic or sub-nanometer scale. X-ray energy dispersive spectroscopy (XEDS) can give quantitative data describing changes of elemental composition associated with inhomogeneous structures of the sample. The combination of XEDS and EELS with HAADF imaging technique can provide detailed information on the composition, chemistry, and electronic and crystal structure of nanoscale systems with atomic resolution and sensitivity. By collecting or analyzing secondary electron (SE) and Auger electron (AE) signals emitted from the specimen surface, we can extract information about the surface topography or surface composition of the specimen. By positioning an electron nanoprobe at the area of interest, coherent electron nanodiffraction (CEND) patterns from individual nanocomponents can be acquired to provide multitudinous information on the nanostructure of the specimen. The powerful combination of high-resolution imaging with nanospectroscopy and nanodiffraction techniques has proved invaluable in solving a plethora of materials problems, including challenging industrial problems. Fig. 1 Open in new tabDownload slide Schematic diagram illustrates the various signals generated inside a scanning transmission electron microscope that can be used to form high-resolution images, nanodiffraction patterns or spectra of the region-of-interest. X-ray energy dispersive spectroscopy (XEDS); Auger electron spectroscopy (AES) and scanning Auger microscopy (SAM); secondary electron spectroscopy (SES) and secondary electron microscopy (SEM); annular dark-field (ADF) and high-angle annular dark-field (HAADF); coherent electron nano-diffraction (CEND); parallel electron energy-loss spectroscopy (PEELS); bright-field (BF) and dark-field (DF). Fig. 1 Open in new tabDownload slide Schematic diagram illustrates the various signals generated inside a scanning transmission electron microscope that can be used to form high-resolution images, nanodiffraction patterns or spectra of the region-of-interest. X-ray energy dispersive spectroscopy (XEDS); Auger electron spectroscopy (AES) and scanning Auger microscopy (SAM); secondary electron spectroscopy (SES) and secondary electron microscopy (SEM); annular dark-field (ADF) and high-angle annular dark-field (HAADF); coherent electron nano-diffraction (CEND); parallel electron energy-loss spectroscopy (PEELS); bright-field (BF) and dark-field (DF). Professor John M. Cowley dedicated >30 years of his research effort to exploring, developing and establishing various imaging, diffraction and spectroscopic techniques that can be practiced in a dedicated STEM instrument. Using a heavily modified HB5 STEM instrument from VG Microscopes, Ltd of England (see Fig. 2), Professor Cowley investigated the various modes of high-resolution STEM imaging [1–41], developed optical systems for conveniently recording nanodiffraction patterns, and established the nanodiffraction technique as a viable alternative to investigate the structures of nanoscale systems including small particles, surfaces and biological systems [4–8,42–66]. Throughout his experimental research activities at Arizona State University, Professor Cowley steadfastly explored various avenues, including holography, atomic focuser and diffraction imaging in recent years to improve resolution in STEM. Not only did he develop theories for various imaging and diffraction techniques but he also applied these new methods to the structural study of surfaces and interfaces, small particles and supported catalysts, localized defects and disordering, carbon nanotubes and many other nanosystems. Fig. 2 Open in new tabDownload slide The heavily modified VG HB-5 scanning transmission electron microscope of which Professor John M. Cowley used at Arizona State University for all his experimental research work. The black box (indicated by the arrow) contained the unique optical system that transfers the light to the various photomultipliers (PMs) and the low-light sensitivity TV camera. The ADF images were formed by positioning a light-absorbing mask in the center of the optical system. In the late 1980s, we used to use a US coin of a penny, a dime or a quarter as the mask of the diffraction pattern and were able to independently vary the inner and outer collection angles of the ADF detector. Other configured STEM detectors were also tried by masking the various parts of the diffraction pattern displayed on the optical system inside the black box. Fig. 2 Open in new tabDownload slide The heavily modified VG HB-5 scanning transmission electron microscope of which Professor John M. Cowley used at Arizona State University for all his experimental research work. The black box (indicated by the arrow) contained the unique optical system that transfers the light to the various photomultipliers (PMs) and the low-light sensitivity TV camera. The ADF images were formed by positioning a light-absorbing mask in the center of the optical system. In the late 1980s, we used to use a US coin of a penny, a dime or a quarter as the mask of the diffraction pattern and were able to independently vary the inner and outer collection angles of the ADF detector. Other configured STEM detectors were also tried by masking the various parts of the diffraction pattern displayed on the optical system inside the black box. The incorporation of atomic resolution STEM techniques into the newer generation field-emission TEM instruments [67–69] and the revival of the experimental dedicated STEM instruments, especially the ones incorporating the aberration correctors or monochromators [70–77], clearly demonstrated the increasing acceptance and the power of STEM techniques that Professor Cowley had been championing for the last 30 years. The recent achievement of sub-angstrom resolution imaging [72,73,76] and atomic scale spectroscopy [74,75] in Cs-corrected STEM instruments will undoubtedly further enhance and expand the impact of STEM techniques on nanoscience research. Professor Cowley recently stated: ‘STEM is finally coming into age and will soon become mainstream’. There have been many reports, in the last decade or so, on atomic resolution HAADF imaging and EELS techniques as well as the application of these techniques to the study of interfaces and defects [69,78–81], quantum dots [82], and nanoparticles and supported catalysts [83–90]. Most of these studies, however, focused primarily on the HAADF and EELS techniques and capabilities. While the combination of these two atomic resolution techniques has proved to be extremely powerful for solving materials problems (especially interface structures), other imaging, diffraction and spectroscopic techniques readily available in a STEM instrument can provide complementary and unique information on the specimen of interest. In this paper, we review some recent developments of the various STEM techniques, which are pioneered by Professor John M. Cowley, with a focus on applying these techniques to the fundamental study of nanoparticles and nanoparticle systems. STEM imaging: shadow image, projection microscopy and electron Ronchigrams The simplest form of imaging in a STEM instrument is shadow imaging (also called point projection microscopy). Projection microscopy was proposed as early as 1939 by Morton and Ramberg [91]. In a projection microscope, the greatly magnified shadow of an object can be obtained by using the quasi-radial propagation of a point or small electron source with the object inside the beam path; it is essentially a lensless microscope based on the radial propagation of an electron beam from a point source. The magnification of the shadow image on the observation screen is determined by the ratio of the distance between the observation screen and the point source to the distance between the object and the point source. Magnifications of 106–107 can be easily achieved when the object-to-point source distances are in the range of ≤10 nm. The use of a small electron source originating from a field-emission gun in a STEM instrument guarantees the complete coherence of the convergent electron nanoprobe impinging onto the specimen. In contrast to conventional high-resolution TEM (HRTEM) imaging, the individual incident rays of different angles within the coherent convergent nanoprobe can interfere with each other; the diffraction of the coherent beam by the specimen and the interference among the incident and diffracted beams can result in complicated forms of shadow images. In fact, in his original paper on holography, Gabor [92] proposed that a very small source of electrons should be placed close to a thin object to form a highly magnified shadow image that could be regarded as a hologram; with the correction of the lens aberrations, the object could be reconstructed thereby resulting in much improved image resolution. To accomplish what Gabor proposed, a high-brightness, ultra-stable, small electron source is needed; the availability of a nanoprobe in a dedicated STEM or field emission TEM/STEM instrument makes this proposal more feasible now. Some early experimental results explored the practicality of this reconstruction process [14]. The principal difficulty of employing this in-line STEM holography, however, originates from finding suitable ways of separating the conjugated images [14,27]. With the incorporation of Cs-correctors [70] or the use of high-brightness nanotips [93], which provide near-point sources, and the availability of high dynamic-range CCD cameras and fast computers, Gabor's proposal should become more practical and greatly improved resolution should be achievable by reconstructing the objects from in-line holograms. When a small electron probe interacts with a thin specimen in a STEM instrument, the high-energy incident electrons are scattered. The amplitude distribution of the transmitted electrons at the far-field can be described by a wave function Ψ(K,X). The variable K is a 2-D vector in the reciprocal space with |K| = 2sin(𝛉/2)/λ (where 𝛉 is the scattering angle and λ is the wavelength of the incident electrons) and X designates the electron probe position on the specimen. When the electron probe is scanned across the specimen, variations of Ψ(K,X) carry information about the electron beam–specimen interactions. If the wave function Ψ(K,X) of the transmitted high-energy electrons can be determined, we can extract structural information about the specimen. It is, however, not possible to directly measure Ψ(K,X); instead, the intensity distribution of the transmitted electrons is observed on the detector plane, which is located at a large distance from the specimen, I(K,X) = |Ψ(K,X)|2. The wave function Ψ(K,X), to first-order approximation, can be expressed as: \[{\Psi}\left(\mathbf{\mathrm{K}},\ \mathbf{\mathrm{X}}\right)\ =\ Q\left(\mathbf{\mathrm{K}}\right){\ast}\left[T\left(\mathbf{\mathrm{K}}\right)\mathrm{exp}\left({-}\mathrm{i}2{\pi}\mathbf{\mathrm{K}}{\cdot}\mathbf{\mathrm{X}}\right)\right]\] (1) where Q(K) is the Fourier transform of the transmission function, q(x), of the specimen and the * symbol represents convolution. The transfer function of the microscope, T(K), is given by: \[T\left(\mathbf{\mathrm{K}}\right)\ =\ A\left(\mathbf{\mathrm{K}}\right)\mathrm{exp}\left\{{-}\mathrm{i}{\chi}\left(\mathbf{\mathrm{K}}\right)\right\}\] (2) where the aperture function, A(K), is given by: \[A\left(\mathbf{K}\right)\ =\ \left\{\begin{array}{ll}1&\mathrm{for}\ K\ {<}\ K_{0}\\0&\mathrm{for}\ K\ {>}\ K_{0}\end{array}\right.\] (3) where K0 is the cut-off wave-vector determined by the aperture size of the probe-forming lens. The aberration function of the objective lens, χ(K), is approximated by (for a non-corrected objective lens): \[{\chi}\left(K\right)\ =\ {-}{\pi}{\Delta}{\lambda}K^{2}\ +\ 0.5{\pi}C_{\mathrm{s}}{\lambda}^{3}K^{4}\] (4) where Δ is the defocus value of the electron probe and Cs is the spherical aberration coefficient of the objective lens. In the phase object approximation [24], the specimen transmission function q(x) can be approximated as: \[q\left(\mathbf{\mathrm{x}}\right)\ =\ \mathrm{exp}\left({-}\mathrm{i}{\sigma}{\phi}\left(\mathbf{\mathrm{x}}\right)\right)\] (5) where σ = π/(λE0) is the interaction constant, E0 is the accelerating voltage and ϕ(x) is the projected specimen potential along the incident beam direction. The amplitude distribution of the coherent incident probe is represented by: \[P\left(\mathbf{\mathrm{R}}\right)\ =\ {\int}T\left(\mathbf{\mathrm{K}}\right)\mathrm{exp}\left({-}\mathrm{i}2{\pi}\mathbf{\mathrm{K}}{\cdot}\mathbf{\mathrm{R}}\right)\mathrm{d}\mathbf{\mathrm{K}}\] (6) The amplitude distribution of the incident probe, P(R), is determined by the Fourier transform of T(K), which is determined by the aperture function A(K) and the aberration function χ(K) of the objective lens. The probe size, therefore, depends on the spherical aberration coefficient of the objective lens, the wavelength of the incident electrons, the size of the objective aperture and the defocus value of the electron beam. The integral in eq. (6), unfortunately, cannot be done analytically and must be obtained numerically. In practice, the spherical aberration coefficient of the objective lens and the wavelength of the incident electrons are not variables during an experiment; the operator, however, can manipulate the size and shape of the coherent electron nanoprobe by varying the size of the objective aperture and the defocus value of the electron beam. If no objective aperture or a very large objective aperture is used, then the stationary incident probe can be very large depending on the defocus value of the electron beam. Images formed in this way in the back focal plane are similar to low magnification TEM images. Depending on the sign of the electron beam defocus, the image magnification can be positive or negative and the image contrast can be reversed. This imaging mode can be conveniently used for the initial survey of specimen features or for monitoring the specimen movement when specimen tilt is required. Note that out-of-focus shadow images are projection images observed in the back-focal diffraction plane with a stationary electron beam. When the third-order spherical aberration is dominant, as in the round lenses used in electron microscopes, the projection image of the specimen placed close to the position of the minimum diameter of the incident probe, as illustrated in Fig. 3a, is greatly distorted due to the aberrations of the objective lens. For paraxial rays (e.g. the ray #4 in Fig. 3a), the beam crossover is after the specimen so that the magnification of the central part of the projection image is high but negative (region 4 in Fig. 3b). For marginal rays (e.g. the ray #1 in Fig. 3a), the crossover is before the specimen so that for the outer part of the projection image the magnification is high but positive (region 1 in Fig. 3b). For a particular set of rays (e.g. the ray #2 in Fig. 3a), the beam crossover is right at the specimen level so that the magnification of the projection image of that specimen region becomes infinity (region 2 in Fig. 3b). Taking into account of the 3-D nature of the rays and the specimen, we can deduce that there is one radius of infinite tangential magnification (labeled as T in Fig. 3b) and another radius of infinite radial magnification (labeled as R in Fig. 3b). Fig. 3 Open in new tabDownload slide Schematic diagram (a) illustrates the effect of spherical aberration of the probe-forming lens on the crossover of the STEM probe. Shadow images of amorphous carbon film at under-focus (b), near-focus (c) and over-focus (d). The tangential (T) and radial (R) circles of infinite magnification are clearly discernible in (b). The circle in (c) indicates the optimum angular size of the objective aperture to be used for high-resolution imaging. Fig. 3 Open in new tabDownload slide Schematic diagram (a) illustrates the effect of spherical aberration of the probe-forming lens on the crossover of the STEM probe. Shadow images of amorphous carbon film at under-focus (b), near-focus (c) and over-focus (d). The tangential (T) and radial (R) circles of infinite magnification are clearly discernible in (b). The circle in (c) indicates the optimum angular size of the objective aperture to be used for high-resolution imaging. Note that in the shadow image of Fig. 3b the image magnification changes continuously along the radius from the optical axis. Furthermore, each point in the image can be described by a wave vector K and its intensity I(K,X) is determined by the wave function Ψ(K,X) given in eq. (1). The effect of the beam defocus on the final image can be more appreciated if we imagine that instead of changing the beam defocus we move the specimen along the optical axis (see Fig. 3a). For example, when the sample is positioned right below the paraxial crossover, shadow images of amorphous materials similar to Fig. 3d can be obtained. All the points in the shadow image have a negative magnification and the contrast of the features is reversed (e.g. heavy scatterers appear bright). At close to the Gaussian defocus, a position called fusiform focus (uniform focus) [94] exists; the position of this fusiform focus is between the paraxial focus and the marginal one. For a thin amorphous film, an almost featureless disc appears in the center of the shadow image when the specimen is at the fusiform focus position as shown in Fig. 3c. The angular size of the disc is determined by the lens aberrations (primarily third-order spherical aberrations for non-corrected lenses) and the wavelength of the incident electrons. The smaller the Cs value of the microscope, the larger the featureless disc. The presence of the almost featureless disc in shadow images of amorphous materials proves to be very useful for the practical operations of a STEM instrument. First, the center of the disc defines the coma-free optical axis of the electron beam so that it can be used as the reference for alignment of other components [6,7,12,13,68]. Second, the size of the disc defines the optimum aperture size that should be used to form the smallest electron nanoprobe. Electrons arriving at the specimen from larger incident angles (outside the circle in Fig. 3c) do not contribute to the central peak of the coherent electron nanoprobe; instead, they add to the oscillating tails, thereby broadening the electron nanoprobe. Accordingly, if one desires to have most of the electrons confined to the smallest central peak, then one should allow only the electrons with incident angles smaller than the one defined by the circle in Fig. 3c to enter the objective aperture. For high-resolution annular dark-field (ADF) imaging, however, an objective aperture with an angular size larger than the optimum angle is sometimes used in order to have a narrower central peak of the electron probe (at a larger under-focus value), thereby providing higher image resolution at the expense of the image contrast [25]. Another important use of the shadow image of amorphous materials is to correct the axial astigmatism in the STEM imaging mode. If the objective lens has astigmatism, the circular symmetry of the featureless disc or the circle of infinite magnification is distorted. The degree and direction of the distortion are determined by the lens astigmatism. Similar to the use of a tableaux of diffractograms for astigmatism correction in a HRTEM instrument, shadow images of amorphous materials can be effectively used to correct these aberrations. Figure 4 shows a set of shadow images of the same region of an amorphous carbon film demonstrating the use of shadow images to correct the astigmatism of the probe-forming system. Shadow images can also be used to monitor the instabilities of the microscope and the specimen; instabilities well below 0.1 nm can be easily discerned in near-focus shadow images. Fig. 4 Open in new tabDownload slide A set of shadow images of an amorphous carbon film illustrates the use of shadow images to correct the astigmatism of the probe-forming lens, to find the coma-free optical axis, and to determine the defocus value of the electron beam. Fig. 4 Open in new tabDownload slide A set of shadow images of an amorphous carbon film illustrates the use of shadow images to correct the astigmatism of the probe-forming lens, to find the coma-free optical axis, and to determine the defocus value of the electron beam. Correlation of shadow images, obtained from different probe positions, can provide the exact magnification of the selected points in the shadow images. The quantification of the local magnifications in a large portion of the shadow image provides an avenue to calculate the axial aberration coefficients and other parameters that control the performance of the probe-forming systems. The availability of high-sensitivity CCD detectors and fast desktop computers makes it now possible to quickly auto-tune Cs-correctors based on the shadow images [71]. The effective use of shadow imaging to properly align and tune the electron optical system clearly contributes significantly to achieve resolution improvement in STEM instruments by using Cs-correctors [73]. When the specimen is a thin, crystalline material and when the beam is aligned along a zone axis or in a direction for systematic diffraction, the shadow image of a set of parallel lattice planes is distorted by the lens aberrations to give a set of loops or serpentine fringes. Such fringes are also known as Ronchi fringes or Ronchigrams in honor of Ronchi who observed such fringes when a diffraction grating was placed near the focus of a large telescope mirror and who correlated the presence of such fringes to the lens aberrations [94]. Typical Ronchi fringes in electron shadow images of crystalline materials are shown in Fig. 5. Similar to the discussion of shadow images of amorphous materials, the distortion of the straight lattice planes, especially near focus, is caused by the lens aberrations. The distortion is smaller for large defocus values and for lenses that have smaller Cs values. The Ronchi fringes in the electron Ronchigrams should have the exact symmetry as that of the crystal in that particular orientation. Thus, one can use the electron Ronchigrams to correct astigmatism, to align the microscope and to center the objective aperture. Measurements of the dimensions of the fringe features in electron Ronchigrams of thin crystals can provide accurate values of the spherical aberration coefficient of the objective lens and the exact focus value of the electron probe [13]. The electron Ronchigrams obtained at the under-focus (Fig. 5a), near-focus (Fig. 5b) and over-focus (Fig. 5c) settings have very different forms, which provide a simple and convenient way to align and tune the electron optical system and to orient the crystalline specimen region of interest. Fig. 5 Open in new tabDownload slide Electron Ronchigrams of a GaAs crystal at under-focus (a), near-focus (b) and over-focus (c). Under-focus electron Ronchigram of a GaAs crystal oriented along the [011] zone axis shows 2-D lattice fringes (d). Electron Ronchigrams of (a)–(d) were obtained (recorded on a tape) on the VG HB-5 STEM shown in Fig. 2. For comparison, image (e) shows an electron Ronchigram of a silicon crystal obtained (recorded on a CCD camera) on the JEOL 2010F TEM/STEM. Fig. 5 Open in new tabDownload slide Electron Ronchigrams of a GaAs crystal at under-focus (a), near-focus (b) and over-focus (c). Under-focus electron Ronchigram of a GaAs crystal oriented along the [011] zone axis shows 2-D lattice fringes (d). Electron Ronchigrams of (a)–(d) were obtained (recorded on a tape) on the VG HB-5 STEM shown in Fig. 2. For comparison, image (e) shows an electron Ronchigram of a silicon crystal obtained (recorded on a CCD camera) on the JEOL 2010F TEM/STEM. Large under-focus electron Ronchigrams of zone axis crystals can provide a readily interpretable image of the crystal lattice. Figure 5d shows such an electron Ronchigram of a GaAs crystal oriented along the [011] zone axis. Fringes representing the crystal lattice spacings can be seen in all directions near the optical axis. At large angles, the fringes are distorted because of the spherical aberration of the objective lens. Their distortion, as a function of angle from the coma-free optical axis, is circularly symmetric and their intensity distribution can be affected by the beam alignment with the exact zone-axis of the crystal. With the use of the aberration correctors in a STEM instrument, the regions of interpretable image of the crystal lattices in the under-focus electron Ronchigrams should be significantly enlarged. Therefore, high-resolution information of local regions can be extracted from the point projection microscopy images. The resolution obtainable in this projection microscopy mode should be comparable to that of the scanned STEM images. STEM imaging: convergent beam electron diffraction and bright-field and dark-field high-resolution imaging When an objective aperture is used to limit the large-angle rays entering the objective lens, convergent beam electron diffraction (CBED) patterns are formed on the detector plane. If the specimen is a thin crystal oriented along a principal zone-axis, instead of shadow images or electron Ronchigrams as discussed above, a CBED pattern consisting of sets of convergent beam discs is obtained as schematically illustrated in Figs 6a (side view) and 6b (top view). Each diffraction disc subtends the same semi-angle α, which is determined by the angular size of the objective aperture, at the specimen. If α > 𝛉B (𝛉B is the Bragg diffraction angle of the diffracting planes), then the convergent beam diffraction discs overlap as shown in Fig. 6c. For thin, perfect crystals, the electron intensity within non-overlapping regions (e.g. the region indicated by numeral 1 in Fig. 6b) is independent of the probe position and the aberrations of the probe-forming lens [4,5]. The electron intensity within regions where discs do overlap depends on the probe position, the lens aberrations and the defocus values of the objective lens. The intensity modulations in regions of overlapping discs are caused by coherent interference of high-energy electrons that have different incidence-beam directions (different incident wave vector Ki) but that are scattered into the same direction (the same final wave vector Kf) by the crystal. The formation of interference fringes in the overlapping regions is purely caused by the coherent nature of the convergent electron nanoprobe. Fig. 6 Open in new tabDownload slide Schematic diagrams show the formation of convergent electron beam diffraction patterns with overlapping disks: (a) side view and (b) top view. The numerals in (b) indicates the number of overlapping diffraction discs in that region of the diffraction plane. Coherent convergent electron beam diffraction pattern from a GaAs crystal oriented along the [011] zone axis (c) shows the overlapping discs. Bright-field STEM image of a TiO2 nanoparticle (d) shows 2-D lattice fringes and the corresponding diffractogram is shown in the inset. Fig. 6 Open in new tabDownload slide Schematic diagrams show the formation of convergent electron beam diffraction patterns with overlapping disks: (a) side view and (b) top view. The numerals in (b) indicates the number of overlapping diffraction discs in that region of the diffraction plane. Coherent convergent electron beam diffraction pattern from a GaAs crystal oriented along the [011] zone axis (c) shows the overlapping discs. Bright-field STEM image of a TiO2 nanoparticle (d) shows 2-D lattice fringes and the corresponding diffractogram is shown in the inset. If a detector is used to collect the signal in the diffraction plane, then a STEM image is formed when the electron nanoprobe is scanned across the specimen. Depending on the detector configuration and positioning, various forms of STEM images can be generated; the image interpretation, the achievable resolution and the contrast mechanisms of the acquired STEM image are determined by the shape and the size of the detector. In STEM imaging, in addition to the probe size, detector function is the other most important variable. For a fixed probe position X at the sample, the intensity distribution of the transmitted electrons on the diffraction observation screen is given by I(K,X) = |Ψ(K,X)|2. The observed image intensity, I(X), as a function of the beam position X, is given by: \[I\left(\mathbf{\mathrm{X}}\right)\ =\ {\int}D\left(\mathbf{\mathrm{K}}\right)\ {\vert}{\Psi}\left(\mathbf{\mathrm{K}},\ \mathbf{\mathrm{X}}\right){\vert}^{2}\mathrm{d}\mathbf{\mathrm{K}}\] (7) where D(K) is the transmission function of the detector. The detector function, D(K), plays the most important role in determining the nature of STEM imaging. For example, if D(K) ≡ 1 for all scattering angles, the STEM image is formed by collecting all the high-energy electrons penetrating through a thin specimen. If the backscattering and the electron absorption effects by the sample are negligible, then, the image intensity I(X) should not vary with the beam position X at all because of the conservation of the total number of the high-energy electrons. Therefore, no contrast will be observed in the STEM image, and no information about the specimen can be inferred. When the specimen becomes thicker, however, the absorption and backscattering of high-energy electrons become appreciable so that an absorption contrast should be observable. (The imaging theory of backscattered electrons may be applicable here.) On the other hand, If D(K) = δ(K) or δ(K − G) where G is a reciprocal lattice vector, then eq. (7) reduces to: I(X) = |Ψ(0, X)|2 or |Ψ(G, X)|2. This is the same form as for BF or DF TEM imaging with parallel illumination. If a very small detector is positioned at any point in the overlapping regions of the diffraction discs, lattice fringes should be obtained by scanning the electron probe across the specimen. Two-dimensional lattice fringes can be obtained by positioning the STEM detector at a point where three or more non-systematic diffraction discs overlap. These multiple-beam interference regions are labeled as numeral 3 (three-beam interference) and numeral 4 (four-beam interference) in Fig. 6b. High-resolution BF STEM images can, therefore, be interpreted exactly as those of HRTEM images. Figure 6d shows such a BF STEM lattice image of a titania nanoparticle; the inset in Fig. 6d is the corresponding diffractogram of the high-resolution BF STEM image. Nanoparticles of titania are used in many commercial applications, including industrial catalysts, paints, coatings and fillings in fibers. High-resolution dark-field STEM images can be easily obtained by moving the detector to a point outside the directly transmitted disc. For example, a 2-D DF STEM lattice image can be obtained by shifting the STEM detector to position D, which is labeled in Fig. 6b. DF STEM imaging technique is useful for identifying small particles in supported metal catalysts, defects in extended crystals and different phases in polycrystalline nanophase materials. Similar to the tilted dark-field imaging in HRTEM, DF STEM technique can provide higher image resolution under optimum conditions [4,5]. The contrast of high-resolution STEM images varies with the displacement of the STEM detector. The movement of the STEM detector corresponds to beam tilt in TEM. In STEM, however, the relative shift of the BF detector is easily accomplished by deflecting the whole diffraction pattern with the use of scanning coils. Unlike beam tilt in TEM, the movement of scanning coils does not perturb the optical alignment of the STEM microscope. Thus, the contrast of specific features of a sample can be conveniently enhanced or reduced by shifting the position of the STEM detector without the complication of misaligning or realigning the microscope. This method is useful for imaging highly inhomogeneous samples, especially for identifying small particles or for imaging inter-phase interfaces with enhanced chemical sensitivity. The disadvantage of both the BF and DF STEM imaging modes is that most of the transmitted electrons are not utilized. STEM imaging: large-angle bright-field and ADF imaging The phase contrast of BF STEM images rapidly decreases with the increase of the detector size. By applying the Principle of Reciprocity [1], we can deduce that the increase of the detector size in STEM is equivalent to the increase in the illumination convergence angle in TEM. The use of large convergence angles of illumination in TEM pushes the first crossover of the contrast transfer function to higher values and causes a rapid damping of high frequency oscillations. Interpretable image resolution can be improved at the expense of image contrast. If the STEM detector is increased to just coincide with the disc of the directly transmitted electrons, i.e. D(K) = A(K), imaging theory suggests that, with a phase object approximation, the image intensity can be approximated by the method of Cowley [24]: \[I_{\mathrm{BF}}\left(\mathbf{\mathrm{X}}\right)\ =\ 1{-}2\left[1{-}\mathrm{cos}\left({\sigma}{\phi}\left(\mathbf{\mathrm{X}}\right)\right)\right]{\ast}{\vert}t\left(\mathbf{\mathrm{X}}\right){\vert}^{2}\] (8) where σϕ(X) is the projected potential along the beam direction and σ is the interaction constant. In a weak phase object approximation cos(σϕ(X)) ∼ 1 − 0.5(σϕ(X))2, thus: \[I_{\mathrm{BF}}\left(\mathbf{\mathrm{X}}\right)\ =\ 1{-}\left({\sigma}{\phi}\left(\mathbf{\mathrm{X}}\right)^{2}{\ast}{\vert}t\left(\mathbf{\mathrm{X}}\right){\vert}^{2}\right)\] (9) This is a form of incoherent imaging: the phase contrast is washed out and the image resolution is determined by the probe current distribution inside the sample. For dynamical diffraction in crystalline materials, the above consideration is not valid and complicated calculations need to be considered. Note that eq. (9) suggests that with the increase of the detector size, the STEM image, within the weak phase object approximation, changes from coherent imaging to completely incoherent imaging. Furthermore, the image formed by a large angle BF (LABF) detector, as represented by eq. (9), is not a linear image anymore with respect to the specimen potential. The image resolution achievable in LABF images should be at least double that of the point-detector BF STEM image or, by the Principle of Reciprocity, that of the HRTEM image. In practice, the relative size of the STEM detector can be continuously varied by changing the strength of the post-specimen or projector lenses (see Fig. 7). If all the directly transmitted electrons and a large portion of the scattered electrons are collected to form the STEM image, then, the dominant phase contrast, usually visible in BF STEM images, is significantly suppressed [25,84]. The contrast of LABF images is predominantly due to absorption effect, weak diffraction effect, plus an electron channeling effect; it is less sensitive to the change of beam defocus, sample thickness, or the Fresnel effects at interfaces or surfaces [25]. For crystals with principal zone-axes aligned in the incident beam direction, the diffraction and phase contrast are significantly reduced in LABF images; but the image resolution is improved [25]. Fig. 7 Open in new tabDownload slide Schematic diagram illustrates the geometric arrangement of BF, ADF and HAADF detectors. The parameter 𝛉 represents the collection (semi-) angle of the BF detector; α1 and α2 are the inner and outer collection (semi-) angle of the ADF detector, respectively; β1 and β2 are the inner and outer collection (semi-) angle of the HAADF detector, respectively. These collection angles can be changed by varying the strength of the post-specimen projector lenses. Fig. 7 Open in new tabDownload slide Schematic diagram illustrates the geometric arrangement of BF, ADF and HAADF detectors. The parameter 𝛉 represents the collection (semi-) angle of the BF detector; α1 and α2 are the inner and outer collection (semi-) angle of the ADF detector, respectively; β1 and β2 are the inner and outer collection (semi-) angle of the HAADF detector, respectively. These collection angles can be changed by varying the strength of the post-specimen projector lenses. To understand the contrast characteristics and the resolution of LABF STEM images, we can rewrite eq. (7) as: \begin{eqnarray*}&&I\left(\mathbf{X}\right)\ =\ {\int}\left[D_{\mathrm{LABF}}\left(\mathbf{\mathrm{K}}\right)\ +\ D_{\mathrm{ADF}}\left(\mathbf{\mathrm{K}}\right)\right]{\vert}{\Psi}\left(\mathbf{\mathrm{K}},\mathbf{\mathrm{X}}\right){\vert}^{2}\ \mathrm{d}\mathbf{K}\\&&=\ I_{\mathrm{LABF}}\left(\mathbf{\mathrm{X}}\right)\ +\ I_{\mathrm{ADF}}\left(\mathbf{\mathrm{X}}\right)\end{eqnarray*} (10) The diffraction plane is divided into two complementary parts: a bright field detector and the corresponding ADF detector (see Fig. 7). If we assume that the specimen is thin enough so electron backscattering and absorption is negligible, then: \[I_{\mathrm{LABF}}\left(\mathbf{\mathrm{X}}\right)\ +\ I_{\mathrm{ADF}}\left(\mathbf{\mathrm{X}}\right)\ {\equiv}\ 1\ \mathrm{and}\ I_{\mathrm{LABF}}\left(\mathbf{\mathrm{X}}\right)\ =\ 1{-}I_{\mathrm{ADF}}\left(\mathbf{\mathrm{X}}\right)\] (11) Therefore, the LABF image is complementary to the corresponding ADF image obtained with an inner collection angle as large as that of the LABF detector. LABF images can be interpreted in the same way as the corresponding ADF images: improvement in image resolution, increased atomic number sensitivity and less dependence on sample thickness. For thin specimens, the contrast of the LABF image could be lower than that of the corresponding ADF image since most of the directly transmitted electrons do not carry specimen information. In practice, this large background signal, however, can be easily subtracted by adjusting the brightness and contrast controls of the STEM detector. By collecting electrons scattered outside the central beam in the diffraction pattern (see Fig. 7), an ADF image of the sample is formed. In fact, atomic resolution imaging was first achieved in STEM by using an ADF detector to collect all the electrons scattered by heavy atoms supported on an ultra-thin, light-element substrate [95]. ADF images of thin crystals of Ti2Nb10O29, giving an image resolution much better than that of the corresponding BF STEM images, were obtained by Professor Cowley in the early 1980s [11]. For various reasons, this powerful high-resolution imaging mode, however, was not pursued aggressively by the research group at Arizona State University until the late 1980s [20,84]. The ADF imaging mode, however, has its drawbacks. Because of the low collection angle of the ADF detector, strong dynamical diffraction effects from crystalline materials obscure its compositional sensitivity; this is especially severe if one wants to detect small metal particles in supported metal catalysts. The contrast of ADF images of supported metal catalysts critically depends on the size of the inner collection angle, α1 (see Fig. 7), of the ADF detector. For imaging metal particles supported on a thick substrate, the contrast of the metal particles can change from dark, to almost none, then to bright with the increase of the inner collection angle of the ADF detector [20]. This angular-filtering effect is more useful if an annular ring detector is used, which will be discussed later. STEM imaging: HAADF imaging or Z-contrast microscopy The diffraction effects in ADF images of crystalline materials can be greatly suppressed by increasing the inner collection angle of the ADF detector beyond the Bragg reflections so that only high-angle scattered electrons contribute to the collected signal (see Fig. 7) [96]. This imaging mode is called HAADF or Z-contrast microscopy. The inner collection angle, β1, is the most critical parameter in determining the nature of the HAADF images; the outer collection angle, β2, is generally made large enough to collect more high-angle scattered signals. In HAADF imaging, the diffraction and phase contrast is significantly suppressed and the compositional sensitivity is recovered; but, the signal strength is greatly reduced. The development of HAADF imaging technique has proved very successful for characterizing small particles and supported metal catalysts with sub-nanometer or atomic resolution and high compositional sensitivity [83–85,90]. Small metal or alloy nanoparticles in high surface-area supports can be easily revealed in HAADF images. Figure 8a, for example, shows a high-resolution HAADF image of a γ-alumina supported PdCu alloy catalyst, clearly revealing the size and spatial distribution of the PdCu nanoparticles. In this particular industrial catalyst, the alloy nanoparticles as small as 0.5 nm in diameter can be revealed in the HAADF images with high contrast and visibility, thus making it possible to reliably extract information on the size and spatial distribution of subnanometer nanoparticles in commercial catalysts. Both the size and spatial distributions of metal particles in supported metal catalysts are important parameters that determine the activity, selectivity and stability of commercial catalysts. This type of information cannot be obtained from any other characterization technique. Fig. 8 Open in new tabDownload slide HAADF image of a Pd-Cu/γ-alumina alloy nanocatalyst shows the size and spatial distribution of the Pd–Cu alloy nanoparticles as well as the pore structure and morphology of the γ-alumina support (a). HAADF image of a 5wt%Au/TiO2 catalyst precursor material shows the presence of Au clusters and nanoparticles (b) as well as individual Au atoms (c) anchored onto the titania surface. Fig. 8 Open in new tabDownload slide HAADF image of a Pd-Cu/γ-alumina alloy nanocatalyst shows the size and spatial distribution of the Pd–Cu alloy nanoparticles as well as the pore structure and morphology of the γ-alumina support (a). HAADF image of a 5wt%Au/TiO2 catalyst precursor material shows the presence of Au clusters and nanoparticles (b) as well as individual Au atoms (c) anchored onto the titania surface. Figures 8b and 8c show high-resolution HAADF images of a Au/TiO2 catalyst precursor material, clearly revealing clusters of Au-containing species anchored onto the titania surface. Individual Au atoms as well as Au clusters and nanoparticles are revealed in the same image; the revelation of the co-existence of both individual Au atoms/ions and clusters/patches of Au-containing species on the titania surface is valuable for understanding the interaction between the Au compounds and the surface sites or defects present on the TiO2 nanoparticle support. Oxygen vacancies on the titania surface are believed to be the anchor sites for the precursor molecules and may act as the nucleation centers during the catalyst reduction process [97]. This type of atomic scale insight into the catalyst precursor material and how it interacts with the support surface is critical to developing nanostructured industrial catalysts with desired performances. The high atomic-number sensitivity of HAADF images can be utilized to differentiate clusters or nanoparticles in mixed metal or alloy catalysts. For example, Fig. 9a shows a HAADF image of a model catalyst consisting of mixed Pt and Pd nanoparticles, clearly revealing the Pt nanoparticles with a much higher intensity. The quantitative interpretation of the image contrast, however, is not straightforward. Figure 9b shows intensity line scans across a Pd and a Pt particle of similar size. The ratio of the peak intensity of the Pt particle to that of the Pd particle is ∼1.8—far smaller than the ratio of (ZPt/ZPd)2 ∼ 2.9—if we assume that the same number of atoms along the beam direction for both the Pt and Pd nanoparticles is measured. The difference between the measured and the expected values could originate from the different shapes of the particles or could be due to intermixing of Pt and Pd in the particles. Even if the nanoparticles have exactly the same size and shape, the effect of electron channeling can modify the intensity ratio between the two particles if they are not oriented along exactly the same direction with respect to the electron beam. The effect of electron channeling and poor signal-to-noise ratio makes it difficult to extract quantitative information on the composition of mixed metal or alloy nanoparticles, especially in supported metal/alloy catalysts. Fig. 9 Open in new tabDownload slide HAADF image of a model catalyst consisting of Pt and Pd nanoparticles on a carbon support (a) and an intensity linescan across a Pd and a Pt nanoparticle of similar size (b). Pure Pt nanoparticles can be easily differentiated from the Pd nanoparticles. Electron channeling effect complicates quantitative interpretation of the observed contrast. Fig. 9 Open in new tabDownload slide HAADF image of a model catalyst consisting of Pt and Pd nanoparticles on a carbon support (a) and an intensity linescan across a Pd and a Pt nanoparticle of similar size (b). Pure Pt nanoparticles can be easily differentiated from the Pd nanoparticles. Electron channeling effect complicates quantitative interpretation of the observed contrast. Intensity analysis of HAADF images of metal nanoparticles in supported catalysts has been reported [83]. In practical industrial catalysts, however, it is difficult to determine the shapes or shape distributions of the metal nanoparticles because of the various errors introduced during the analysis process (e.g. background subtraction, poor signal-to-noise ratio, etc.) and the effect of electron channeling, especially for larger particles, on the integrated intensity of individual nanoparticles cannot be predicted or avoided. It is still an extremely challenging task to extract statistically meaningful data on the size distribution of metal or alloy nanoparticles in industrial heterogeneous catalysts. Quantitative correlation of size distributions of metal nanoparticles in a supported metal catalyst to its performance, which is extremely important for catalyst optimization processes, is difficult, if not impossible, to obtain. Quantitative and statistically meaningful information on the shape distribution of metal nanoparticles in supported metal catalysts, which can be important in determining the catalyst's selectivity, has not yet been reported. When the nanoparticles become much smaller, for example, clusters of a few to about a hundred atoms, the effect of electron channeling may not be significant. In this particular case, the HAADF image intensity may be linearly dependent on the sample thickness or the total number of atoms that the electron probe encounters. Thus, it is possible to determine the 3-D shape of nanoclusters if one can use the integrated intensity of individual single atoms as an internal calibration. For example, Fig. 10a shows an atomic resolution HAADF image of a small Pt cluster. Individual Pt atoms (indicated by the white arrows) were also revealed in the image. Figure 10b shows an intensity profile across the center of the small cluster, revealing that the distance between the center atomic column and the nearest neighbors is ∼0.14 nm. Intensity analysis of the individual atomic columns showed that some atomic columns contain three Pt atoms along the electron beam direction while others contain only two Pt atoms. Image instabilities caused by both external and internal interferences and the sample movement made it difficult to acquire high-quality images with good signal-to-noise ratio. Environmental control, sample stability and signal strength are the most challenging issues if one wants to routinely obtain high-quality, atomic resolution HAADF images of small clusters or nanoparticles and to perform quantitative intensity analysis of the individual clusters or nanoparticles. Fig. 10 Open in new tabDownload slide Atomic resolution HAADF image of a small Pt cluster and individual Pt atoms in a model nanocatalyst (a) and the intensity line scan across the small cluster (b). Using the intensity of the individual Pt atoms (indicated by the white arrows) as an internal calibration, the atomic layers (indicated by the numerals) in the cluster can be deduced. Fig. 10 Open in new tabDownload slide Atomic resolution HAADF image of a small Pt cluster and individual Pt atoms in a model nanocatalyst (a) and the intensity line scan across the small cluster (b). Using the intensity of the individual Pt atoms (indicated by the white arrows) as an internal calibration, the atomic layers (indicated by the numerals) in the cluster can be deduced. The use of Cs-correctors has made it possible to significantly improve the resolution of HAADF images and to greatly increase the effective probe current [73]. Thus, the location of single dopant atoms or promoter species can now be observed with clarity, providing structural information on the relationship between single dopant atoms and the substrate [90]. Structural promoters of single atoms and detailed structures of small nanoparticles and quantum dots can now be investigated with a sub-angstrom resolution [82,90]. The wide availability of Cs-corrected STEM or TEM/STEM instruments will undoubtedly enhance our understanding of the structure and physicochemical properties of small particles and clusters. It is now possible to study the atomic structure and chemistry of metal or alloy nanoclusters in a dedicated STEM or field emission TEM/STEM instrument. The direct imaging of the surface arrangement of different atoms in bimetallic or multimetallic clusters can provide extremely valuable information for understanding the performance, especially the selectivity, of nanocluster catalysts and for synthesizing nanocatalysts with desired performances. The nature of the signals collected by the HAADF detector has been extensively investigated [18,98–112]. The contrast characteristics of incoherent HAADF imaging include: (i) high atomic-number sensitivity—approaching Z2, (ii) less dependence on beam defocus and sample thickness, (iii) absence of proximity effects at interfaces or surfaces and (iv) higher image resolution. For thin samples, the image intensity is linearly proportional to the sample thickness; the electron channeling effect in crystalline materials, however, significantly modifies this relationship. The imaging theory of both HRTEM and BF STEM is a coherent, linear imaging theory: phase contrast and dynamical diffraction effect dominate the image contrast. For ADF imaging, however, the detector always detects interferences among the scattered waves and the directly transmitted beam may not reach the detector; ADF imaging is thus a non-linear imaging technique. The degree of coherence in ADF images varies with the size of the inner collection angle. For a phase object, the inner collection angle modulates a coherence envelope given by the Airy function that is the Fourier transform of the corresponding LABF detector. Within this spatial envelope, the electrons scattered by the spatially separated scatterers can interfere with each other. The extent of the lateral envelope is inversely proportional to the size of the inner collection angle of the ADF detector. Therefore, increasing the inner collection angle decreases the spatial extent of the coherence envelope. When the coherence envelope becomes much narrower than the distance between the neighboring scatterers, these scatterers can be treated as independent scattering centers; the primary electrons are then scattered incoherently. The strength of the high-angle scattering, which gives the HAADF imaging signal, depends on several parameters including (i) large angle elastically scattered electrons, (ii) phonon scattered electrons and (iii) multiply scattered electrons. The imaging theory of HAADF microscopy follows that of incoherent imaging: the high-angle scattered electrons can be treated as being scattered by independent scattering centers. The lateral coherence of the scattered electrons is almost completely suppressed because of detector geometry (averaging effect) and thermal diffuse (phonon) scattering. The columnar coherence of the scattered electrons is significantly reduced because of phonon scattering although a small residue of the coherence still exists along the incident beam direction [103–113]. For zone-axis crystals, high-energy electrons may preferentially travel along paths of low-energy potentials in the sample. This phenomenon is called the electron channeling in crystalline materials [113]. The propagation of a coherent convergent electron probe inside a perfect crystal in the zone-axis channeling condition has been widely investigated [31,99,113,114]. Remarkable electron focusing effects can occur under channeling conditions. In fact, the focusing action of the potential field of a single heavy atom or rows of atoms extending through a thin crystal can be used to significantly improve the resolution limits of modern electron microscopes [31]. The imaging properties of atomic focusers have been investigated and resolutions of better than 0.05 nm should be achievable in a STEM instrument with a probe size of ∼0.4 nm [32,33,35]. By using carbon-nanoshells as the atomic focusers, ultra-high-resolution images have been obtained and individual tungsten atoms with a size of ∼0.06 nm have been observed in diffraction images by Professor Cowley [35,36,38,39]. The penetration of the incident electrons is different for probes focused onto atomic columns of different species. The channeling effect of a convergent probe is important in interpreting high-resolution HAADF images of crystals oriented in principal zone-axes. Phonon scattering, plus the channeling effect, forms the basis of atomic resolution HAADF imaging of crystalline materials. Because of the effect of electron channeling and dechanneling on the high-angle scattered electrons small perturbations of the potential field can be manifested in HAADF images. For example, it is possible to distinguish the location of individual atoms within, or on the surface of, a substrate. This technique has been effectively utilized to image the location of individual Sb dopant atoms within a silicon crystal [115]. In the incoherent imaging limit, the image contrast becomes a pure ‘number contrast’: the total number of high-angle scattered electrons determines the image intensity at that pixel. Thus, HAADF images can be viewed as the convolution of the intensity distribution of the incident probe with appropriate cross-sections for high-angle scattering processes. Since high-angle scattering processes are highly localized, the resolution of HAADF images is necessarily determined by the size of the incident coherent electron probe. For crystalline materials, the image resolution may depend on the channeling or atomic focusing conditions of the specimen. With coherent, convergent beam illumination, the intensity distribution of the incident probe, I0(X), rather than the vaguely defined probe size, is usually used to describe the performance of a STEM instrument. The form of I0(X) strongly depends on the size of the objective aperture, the spherical aberration coefficient of the objective lens, the beam defocus value, the energy of the incident electrons and the instability of the microscope. The ‘optimum’ probe sizes in a STEM instrument depend on the selected imaging and analytical modes. With the use of smaller objective apertures, which is usually used for nanodiffraction, the intensity distribution of the electron probe does not vary appreciably with the change of beam defocus. On the other hand, with the use of larger objective apertures, such as those used in high-resolution STEM imaging, the intensity distribution within the electron probe becomes increasingly sensitive to the change of beam defocus. For high-resolution imaging of zone-axis crystals, it is desirable to use an objective aperture larger than the optimum aperture and to work at an under-focus value slightly larger than the Scherzer focus in order to improve image resolution without introducing complications in interpreting the image [25]. For imaging small particles, however, the high-resolution imaging condition may not be desirable since the large probe tails complicate the measurement of the particle size. A top-hat or Gaussian probe may be more appropriate for determining the size distributions of isolated individual nanoparticles. To determine the shapes and the surface atomic arrangements of small nanoparticles supported on high-surface-area supports is still a difficult task. Significant improvement in image resolution and probe current by using aberration correctors have made it possible to study the exact shapes and surface atomic arrangements of small particles and quantum dots [82]. Electron beam-induced modifications of the specimen, however, may become an important issue for determining the true surface structure of small clusters and nanoparticles. The high atomic-number sensitivity, the incoherent imaging characteristics, the higher image resolution achievable and the intuitive relationship to the specimen make HAADF imaging the most powerful STEM imaging technique for characterizing interfaces and defects, nanoparticles and nanoparticle systems and other nanoscale systems. STEM imaging: thin annular detector and other configured detectors As shown in Fig. 7, the central beam of the diffraction pattern can be expanded, by the use of post-specimen or projector lenses, to overlap the inner edge of the ADF detector. A thin ring at the outer edge of the directly transmitted beam, plus a small portion of the scattered beams, can be collected to form an image of the specimen. A specially designed thin annular detector (TAD) with only ∼10% difference between the inner and outer collection angle can be used to form images that carry unique information about the specimen [29]. The TAD can be used to form bright-field (TADBF) or dark-field (TADDF) images, depending on the size of the inner-collection angle with respect to the convergence angle of the incidence probe. By applying the Principal of Reciprocity, this imaging mode is equivalent to hollow-cone illumination imaging in TEM. Detailed treatment of the imaging process suggests that the resolution of TADBF images can be significantly improved [28,29]. The TAD imaging modes take advantage of selecting the range of desired frequencies that give higher image resolution and excluding the lower frequencies that contribute to the background signal. The TAD may also be used to collect signals of small angle scattering to produce images of amorphous materials or light element particles. For example, carbon nanoparticles supported on amorphous silica can be detected with good contrast [29]. The image contrast due to strain fields near defects, interfaces and surfaces can be enhanced in TAD images. Magnetic domains or domain boundaries may also be revealed in TAD images with high spatial resolution. The combination of TAD with HAADF imaging technique can be very effective in examining both heavy-element and light-element nanoparticles with atomic-scale resolution. Other specially configured detectors can be constructed to increase image resolution, to enhance image contrast or to extract unique information about certain features of the specimen. For example, circular detectors splitting into halves or quadrants have been used to study magnetic fields or magnetic domain structures of thin films and small particles [116,117]. Complex configured detectors have also been explored for increasing image resolution or for enhancing image contrast. For example, using the optical-lens-transfer system shown in Fig. 2, the Cowley group at Arizona State University explored the possibility of forming images by excluding all the Bragg diffraction spots from thin crystals or by using only electrons that are scattered into certain regions of the diffraction plane. It is, however, difficult to perform these experiments using physical masks. The use of high dynamic-range CCD cameras and faster computers to record the diffraction patterns at each pixel has made it possible to perform various diffraction-imaging configurations. With configured STEM detectors, we can rewrite eq. (10) as: \[I_{i}\left(\mathbf{\mathrm{X}}\right)\ =\ {{\int}_{K_{i}}^{K_{i\ +\ 1}}}D_{i}\left(\mathbf{\mathrm{X}}\right){\vert}\ {\Psi}\left(\mathbf{\mathrm{K}},\mathbf{\mathrm{X}}\right){\vert}^{2}\ \mathrm{d}\mathbf{K}\] (12) \[{{\sum}_{i}}{{\int}_{K_{i}}^{K_{i\ +\ 1}}}D_{i}\left(\mathbf{\mathrm{K}}\right){\vert}\ {\Psi}\left(\mathbf{\mathrm{K}},\mathbf{\mathrm{X}}\right)\ {\vert}^{2}\ \mathrm{d}\mathbf{\mathrm{K}}\ {\equiv}\ 1\] (13) where the summation is over the whole diffraction plane. The resolution and contrast of the STEM images is then dependent on the configuration of the configured STEM detector. By selecting the frequency or the direction of the wave vector in the diffraction plane, a plethora of imaging modes can be used to extract complementary or unique information about the specimen. By digitally recording the whole diffraction pattern with energy (E) discrimination for each pixel (probe position X) on the sample, a 5-D function I(K, X, E) can be generated. All information about the specimen can be extracted by off-line processing of these digitally stored, energy-selected diffraction patterns. By selecting certain portion(s) of the scattered electrons as an input signal, various types of images can be formed to give information about the structure and the chemistry of the sample with atomic resolution. This process, however, needs a tremendous amount of computer work, fast image-acquisition systems, a large collection of data, and a high stability of the microscope and the specimen. Alternatively, reconstruction of the wave function in amplitude and phase can be accomplished by analyzing 4-D functions in the diffraction space. Initial attempts, using the technique of ptychography [118], on super-resolution STEM imaging has been successful [119,120]. An image resolution better than 0.14 nm has been achieved on an STEM with a nominal resolution of only 0.42 nm [120]. With the rapid advancement in the image acquisition systems, faster desk-top computers, specially designed microscope environments and the use of pulsed electron beams, we should be able to efficiently use the various signals available in a STEM instrument with minimum exposure of delicate specimens to the electron beam. STEM techniques will become more critical to the fundamental studies of nanoscale systems and will contribute significantly to the new era of nanotechnology and nanoscience research. STEM imaging: secondary and Auger electron microscopy and scanning reflection electron microscopy of surfaces In a STEM instrument, the specimen is usually placed inside the pole pieces of a highly excited objective lens. The emitted secondary electrons first experience a strong magnetic field before being collected by an SE detector. Owing to the effect of this magnetic field, an emitted SE spirals in a cyclotron orbit with a radius R that depends on the energy and the emission angle of the SE as well as the strength of the magnetic field. After spiraling out of the bores of the objective lens, secondary electrons are collected by an SE detector through a transverse electric field. Because of the effect of the magnetic field on the trajectory of the emitted secondary electrons, the SE collection efficiency in a STEM instrument is high. The collection efficiency of low-energy electrons can be further enhanced by the use of electron ‘parallelizers’ located inside the bores of the objective lens [121,122]. The energy distribution of the collected secondary electrons can be analyzed by an electron spectrometer. Secondary electron spectroscopy can be used to investigate the energy distribution of secondary electrons from different materials, to measure the work function of solid specimens, and to study the charging effects of non-conducting materials. Sub-nanometer surface details can be observed in high-resolution SE images [16,17]. This implies that the generation processes of secondary electrons are localized to within 1 nm or less. It was first pointed out [17] and later experimentally proved [123] that the generation of secondary electrons is directly related to large-angle inelastic scattering of the high-energy incident electrons. There exist large momentum transfer mechanisms during the inelastic scattering processes such as Umklapp (high-momentum, low-energy transfer processes) or phonon-assisted electron excitation processes. Inelastic scattering events involving these processes are highly localized. The resolution obtainable in SE images is currently limited to ∼0.5 nm. Small particles are often observed with a bright contrast in high-resolution SE images; Fig. 11 shows a set of high-resolution SE images of metal particles supported on various oxides, revealing the high-spatial resolution and good image contrast of small nanoparticles. The particle contrast in SE images can be parameterized by the ratio of the particle radius (R) to the average escape-depth (L) of the collected secondary electrons. If R/L < 1, the brightness of a particle increases with the size of the particle and the image intensity has a maximum at the center of the particle. If R/L > 1, the particle intensity slowly increases with the size of the particle and the highest image intensity is approximately at a distance d = (R − L) from the center of the particle. For very large particles, the particle contrast evolves into the edge-brightness contrast commonly observed in SE images. Although the resolution of SE images is comparable to the size of the incident probe, it is impossible to extract information about the shape of nanoparticles with sizes less than the escape depth of the collected secondary electrons. Therefore, we cannot extract information about detailed surface morphology of very small particles. We can obtain, however, useful information about the relative locations of nanoparticles with respect to the surface topography of the supports. Detailed discussions on the origin of small particle contrast and the resolution achievable in SE images have been reported previously [124]. Fig. 11 Open in new tabDownload slide High-resolution secondary electron images of Ag (a) and Fe (b) nanoparticles on MgO smoke crystals and Ag nanoparticles on α-alumina powders (c) and Pd nanoparticles on γ-alumina crystals (d). Small metal clusters and nanoparticles on oxide supports are clearly revealed with high resolution and bright contrast. Fig. 11 Open in new tabDownload slide High-resolution secondary electron images of Ag (a) and Fe (b) nanoparticles on MgO smoke crystals and Ag nanoparticles on α-alumina powders (c) and Pd nanoparticles on γ-alumina crystals (d). Small metal clusters and nanoparticles on oxide supports are clearly revealed with high resolution and bright contrast. The production of Auger electrons is essentially similar to that of low-energy secondary electrons; the initial excitation produced by the inelastic scattering of the incident electrons decays to generate a low-energy electron that can escape into the vacuum. In contrast to the diffusion of secondary electrons, Auger electrons must escape from the specimen surface without losing any energy in order to be registered as Auger peak signals. The reason that Auger electron spectroscopy is a surface-sensitive technique lies in the intense inelastic scattering that occurs for electrons in this energy range; only Auger electrons generated from the outmost atomic layers of a solid can survive to be ejected and registered as Auger electrons. Most of the emitted Auger electrons are produced within a very short distance from the sample surface, typically 0.3–3 nm. In a STEM instrument, Auger electrons, emitted from either the entrance or the exit surface of a specimen, can be collected and analyzed by a cylindrical mirror analyzer (CMA) or a concentric hemispherical analyzer (CHA) electron spectrometer. Because of the high energy and high brightness of the incident electrons, the employment of magnetic ‘parallizers’, and the use of thin specimens in a STEM instrument, high-quality Auger electron spectra can be acquired with extremely high peak-to-background ratios [125–127]. Figure 12a shows a high-energy resolution Auger electron spectroscopy (AES) spectrum of clean silver nanoparticles supported on a small MgO crystal; the silver MNN doublet is clearly resolved. Figure 12b shows the corresponding oxygen KLL Auger peak from the same specimen area. Surface compositional analysis of individual nanoparticles is essential for understanding the activity and selectivity of industrial bimetallic or multi-component catalysts used in a variety of chemical processes. The overall composition of these individual nanoparticles can usually be obtained by XEDS. It is, however, extremely difficult to extract information about preferential surface segregation or aggregation of individual components in nanoparticles of different sizes. Because of the high-surface sensitivity of Auger electrons, it is possible to determine qualitatively and, in some cases, quantitatively, the surface composition of nanoparticles consisting of multiple components. High-spatial resolution Auger electron spectra can provide information about the surface enrichment of specific elements and information about how this enrichment varies with the size of the nanoparticles. Fig. 12 Open in new tabDownload slide Auger electron spectra of (a) Ag MNN and (b) O KLL peaks of an Ag/MgO model catalyst. Auger maps of silver and oxygen are shown in (c) and (d), respectively. Auger electron spectra and scanning Auger microscopy images were obtained from the UHV STEM-MIDAS (Microscope for Imaging, Diffraction, and Analysis of Surfaces) housed at Arizona State University. Fig. 12 Open in new tabDownload slide Auger electron spectra of (a) Ag MNN and (b) O KLL peaks of an Ag/MgO model catalyst. Auger maps of silver and oxygen are shown in (c) and (d), respectively. Auger electron spectra and scanning Auger microscopy images were obtained from the UHV STEM-MIDAS (Microscope for Imaging, Diffraction, and Analysis of Surfaces) housed at Arizona State University. For electron transparent specimens, typically used in STEM instruments, an image resolution <1 nm can be achieved [125–127] in scanning Auger microscopy (SAM) images. Silver nanoparticles <1 nm in diameter and containing as few as 15 silver atoms have been detected [125]. Figures 12c and 12d show, respectively, Ag and O maps of an Ag/MgO model catalyst, clearly revealing the high-spatial resolution of Auger elemental maps, obtainable in dedicated STEM instruments. The resolution in SAM images depends on several sample- and instrument-related effects. The sample-related effects include: (i) surface topography, (ii) escape depth of the collected Auger electrons, (iii) contribution from backscattered electrons and (iv) localization of the Auger electron generation processes. The last factor sets the ultimate resolution limit that will be achievable in SAM images. Since the primary inelastic scattering processes involve excitation of inner-shell electrons, the generation of Auger electrons is highly localized. With thin specimens and high-energy incident electrons, the contribution from backscattered electrons should be negligible; it may, however, degrade the image resolution and affect the image contrast of bulk samples. The instrument-related effects include: (i) the intensity distribution of high-energy electron probes, (ii) the collection efficiency of the emitted Auger electrons and (iii) the instability of the STEM microscopes. At present, the instrument-related factors set the limits of obtainable resolution to ∼1 nm in Auger peak images of thin specimens. The minimum detectable mass in high-spatial resolution SAM images is <3 × 10−21 g [125]. Scanning reflection electron microscopy (SREM) is another technique that can be used to examine surface details of bulk crystals. In SREM, the electron nanoprobe is scanned over the surface with a grazing angle and a selected part of the resulting diffraction pattern [a convergent-beam high-energy electron diffraction (CBRHEED) pattern] is used to form an image of the specimen surface [23]. Surface steps, dislocations and other types of defects on the surfaces of bulk crystals can be imaged with high contrast and resolution. CBRHEED, scanning reflection EELS and XEDS techniques can be used to provide information on the surface structure, composition and even electronic states of bulk crystals [23]. If a HAADF detector is used, then the resulting high-resolution surface image depends strongly on the atomic number of the elements present on the specimen surface and the phase contrast and dynamical diffraction effects are greatly suppressed [23]. Similar to transmission HAADF imaging, high-resolution detail of atom rows along which the electrons can be channeled and atomic scale information on the surface defects should be obtainable. STEM diffraction: CEND The advantage of STEM is that the electron beam can be stopped at any point of interest on a sample and diffraction or spectroscopy can be performed at that point with an atomic or nanometer scale resolution. As discussed earlier, CBED patterns can be formed in the observation screen and the size of the diffraction discs is determined by the convergence angle of the incident probe (see Fig. 6). These CBED patterns are, however, different from those obtained in a TEM. First, the sizes of the electron probes are usually ≤1 nm in diameter, much smaller than those obtainable in TEM; thus, the diffraction patterns obtained in STEM are usually called micro- or nano-diffraction patterns. Second, the use of a field emission gun warrants the coherent nature of a convergent nanoprobe: the illuminating aperture is filled with completely coherent radiation and the final probe entering the specimen can be treated as perfectly coherent. In contrast, the illuminating aperture in conventional TEM is considered incoherently filled and the illumination is treated as completely incoherent. CEND is the only technique that gives full diffraction information about individual nanoparticles. Diffraction patterns from the various parts of a nanoparticle can be obtained to provide information about the structure as well as the morphology of the nanoparticle. For a perfect, thin crystal (no thickness variation, no defects, no bending), there are no differences in the diffraction patterns that are obtained with either a coherent or an incoherent electron beam provided the diffraction discs do not overlap (α < 𝛉B in Fig. 6a). This is a consequence of the Bragg law: for each incident direction, only scattering through mutiples of the Bragg angle is allowed; thus, electrons with different incident beam directions cannot interfere with each other although the incident electron probe is completely coherent. If the crystal is thicker, the intensity distribution within the diffraction discs may become non-uniform, with sets of lines, bands or complicated shapes. This is mostly due to dynamical diffraction effects giving a variation of the incident and diffracted beam intensities as a function of the incidence angle. If we ignore the fine-details, CEND patterns of perfect crystals can be treated the same way as those generated by an incoherent electron beam with a nanometer-size probe. For crystals containing defects (edges and bending, stacking faults and dislocations, thickness variations, etc.) elastic scattering from these defects can coherently interfere with each other or with the Bragg-diffracted electrons. Diffraction patterns, characteristic of the unique nature of the defects, can be observed. For thicker or strongly scattering samples, any discontinuity in the sample can have some observable effect on the CEND patterns. For example, when a small electron beam scans across a straight edge of a MgO cube aligned along the [001] zone-axis, first the central transmitted disc shows strong streaking towards the crystal; then diffraction spots appear at non-Bragg positions (Fig. 13). Fine structures in these CEND patterns change rapidly with the movement of the probe position. The streaking of the central transmitted spot is attributable to the influence of the crystal inner potential. The interference among waves arriving from different incident beam directions gives rise to perturbations of the Bragg diffraction spots. When only part of the incident probe is positioned inside the MgO crystal, electrons with different incident directions interact with different parts of the crystal. The scattered electrons interfere with each other to give a diffraction pattern characteristic of the beam position and that part of the specimen. Simulations using dynamical electron diffraction theory show that the intensity distributions in CEND patterns are sensitive to the surface or the defect structures of the specimen [128]. A surface channeling effect may also be responsible for the fine features observed in CEND patterns from straight edges of small crystals; in this case, the diffraction pattern can be treated as the combination of transmission and reflection high-energy electron diffraction. Fig. 13 Open in new tabDownload slide CEND patterns from a MgO cube oriented along the [001] zone axis. From left to right: the electron beam was moved toward the MgO crystal. The spot streaking and splitting are unique characteristics of coherent electron nanodiffraction resulting from the discontinuities at the specimen surface. Fig. 13 Open in new tabDownload slide CEND patterns from a MgO cube oriented along the [001] zone axis. From left to right: the electron beam was moved toward the MgO crystal. The spot streaking and splitting are unique characteristics of coherent electron nanodiffraction resulting from the discontinuities at the specimen surface. When the incident probe is positioned near the edge of a crystal, CEND discs may show annular rings or splitting of diffracted spots [47], which are clearly shown in Fig. 13. Internal discontinuities, such as fault planes, out-of-phase boundaries and thin layer precipitates, may give characteristic structures in their corresponding CEND patterns. For example, CEND patterns from antiphase domain boundaries in ordered alloys show spot splitting of superlattice reflections [49]. It is, however, impossible to make accurate measurements of lattice parameters in CEND patterns because of the large sizes of the diffraction spots. An error of 5% or higher than that is common in determining lattice constants of small particles by the CEND technique, and much larger errors can frequently occur because of coherent interference effects. It is important to correlate the characteristic features of CEND patterns to particle properties, such as the structure of the particle, the nature of defects within the particle, or the shape and size of the particle. A frequently observed characteristic feature is the splitting of diffraction spots along certain crystallographic directions of a small particle. Figure 14 shows a set of CEND patterns that were obtained from different positions on a small Au cuboctahedral particle, demonstrating the various features of CEND patterns from nanoparticles. The spot splitting in non-overlapping CEND patterns is attributable to the coherent nature of electrons diffracting from an abrupt discontinuity of the scattering potential at particle edges. It is also observed that the spot splitting is related to the geometric forms of the diffracting particles; some splitting occurs in a well-defined crystallographic direction. Depending on the probe position relative to the center of the particle, annular rings may be observed (see diffraction pattern 3 in Fig. 14). CEND patterns are sensitive to edges, thickness variations and facets; the 3-D information of nanoparticles is reflected in their CEND patterns. Analysis of CEND patterns at each pixel element of a nanoparticle should provide detailed information about the 3-D structure of small clusters and nanoparticles. Fig. 14 Open in new tabDownload slide CEND patterns obtained at different positions of a small Au cuboctahedral nanoparticle. These nanodiffraction patterns contain information on the 3-D shape of the Au nanoparticle (see text for detailed discussions). Fig. 14 Open in new tabDownload slide CEND patterns obtained at different positions of a small Au cuboctahedral nanoparticle. These nanodiffraction patterns contain information on the 3-D shape of the Au nanoparticle (see text for detailed discussions). Dynamical simulations reveal that for a particle that has facets smaller than the size of the incident probe, the incident electrons may interact with several facets of the small particle [128]. The thickness of the particle may vary rapidly even within a region of only ∼1 nm in diameter. The electron probe effectively interacts with the ‘particle morphology’ under illumination. The intensity variations of the splitting spots are related to the probe positions with respect to the particle facets and are related to the length of the facets along the incident beam direction. The direction of spot streaking or splitting is directly related to specific edges or facets of a small particle (Fig. 14). Furthermore, the intensity profiles across the splitting spots vary with the types of particle wedges. In principle, it is possible to deduce the 3-D structure of nanoparticles by quantitatively analyzing the intensity distributions of their CEND patterns. Before this technique can be effectively and reliably utilized to extract the rich information coded in CEND patterns of small particles, many experimental difficulties, such as particle stability, contamination and accurate control of beam defocus, have to be overcome. When metal atoms aggregate from the vapor phase or in a liquid, they usually form a crystal, having shapes of regular pentagonal bi-prisms or icosahedra. Their internal structure is a complex arrangement of 5 or 20 twinned components. Large metal particles of cuboctahedron, decahedron, icosahedron and other multiple-twinned structures can be examined in HRTEM images [129,130]. For particles with sizes <2 nm in diameter, however, it is difficult to unambiguously determine their shape by imaging techniques. CEND technique can provide information about the shape of clean, metallic nanoparticles. For example, a large portion of clean silver nanoparticles with sizes <3 nm in diameter was observed to give unique CEND patterns exhibiting 5-fold-symmetry. Figure 15a shows such a CEND pattern and Fig. 15c shows a simulated CEND pattern of a small icosahedron with the incident beam direction along the 5-fold symmetry axis. The simulated pattern closely matches the experimental one. Figures 15b and 15d show, respectively, experimental and simulated CEND patterns of a small icosahedral Ag nanoparticle oriented along its 3-fold axis. Although the general features are similar between the experimental and the simulated CEND patterns, quantitative comparison has not yet been performed. These small particles are not stable under intense electron beam irradiation and their structure fluctuates rapidly during observation. Detailed quantitative analyses of digitally recorded CEND patterns will provide information on the shape of, as well as the defective structure in, small nanoparticles. Simulations of CEND patterns of various shapes of nanoparticles should provide insight into the nature of CEND from nanoparticles or other nanosystems. Fig. 15 Open in new tabDownload slide CEND patterns obtained from small Ag icosahedral nanoparticles along the 5-fold symmetry axis (a) and the 3-fold symmetry axis (b); the corresponding simulated patterns are shown in (c) and (d), respectively. Fig. 15 Open in new tabDownload slide CEND patterns obtained from small Ag icosahedral nanoparticles along the 5-fold symmetry axis (a) and the 3-fold symmetry axis (b); the corresponding simulated patterns are shown in (c) and (d), respectively. CEND technique has been applied to the study of defects [49,50], supported catalysts [55,56,89], structure of carbon nanotubes [59,60], and biological systems [63–65]. The recent application of using coherent beams of small diameter for deriving the structure of double-walled carbon nanotubes with high resolution and contrast clearly demonstrated the potential of employing coherent nanoprobes to extract structural information of periodic or non-periodic objects [131]. STEM nanospectroscopy: XEDS and EELS XEDS is now routinely used, in TEM, SEM or STEM instruments, to identify unknown phases or to obtain information on the spatial distribution of certain phases of interest. In a modern FEG TEM/STEM instrument, XEDS can be conveniently used to analyze the features revealed in HAADF images by stopping the incident probe at any point of interest. With the recent development of image and spectrum acquisition systems, both qualitative and quantitative information on the composition of individual nanocomponents can be obtained. The availability of faster computers for automation and online data analysis make it possible to analyze extremely complex nanoscale systems and to quickly diagnose their basic composition. One of the most useful techniques for understanding the behavior of bimetallic nanoparticles and for guiding the development of industrial bimetallic catalysts is the composition-size plot method developed by the Lyman's group [132,133]. The composition-size plots can provide the compositional profiles of the individual bimetallic nanoparticles or clusters; they reveal whether the compositions of individual nanoparticles vary with their sizes or with their relative locations with respect to the catalyst support. When ultramicrotomed samples are used, this method can quantitatively map out how the compositional profiles vary within the supports, the treatment conditions, or the preparation procedures. The composition-size plot method can also be applied to studying the compositional evolution of individual bimetallic nanoparticles during the catalytic reactions. Figure 16 shows examples of how the composition-size plots are used to provide information on the nature of bimetallic catalysts and how the information can be used to develop new synthesis strategies in order to obtain particular structures that give desired performances. Figure 16a is a composition-size plot obtained from a 2wt%Pd1wt%Cu/γ-Al2O3 bimetallic catalyst. Each data point represents the composition of that individual bimetallic nanoparticle. The nanoparticles selected for analysis are located at different regions of the catalyst powders. The plot shows that the composition of the individual Pd–Cu nanoparticles does not vary much with the size of the particles; but they do change significantly with the location of the individual particles. Further analysis of the corresponding HAADF images show that the compositional variations of the individual Pd–Cu nanoparticles revealed in Fig. 16a are related to the macro- and nano-structure of the γ-Al2O3 aggregates/powders. Therefore, the metal precursor deposition processes have to be modified in order to obtain bimetallic nanoparticles with a uniform composition throughout the catalyst powders. Fig. 16 Open in new tabDownload slide Composition-size plots of (a) 2wt%Pd1wt%Cu/γ-Al2O3 and (b) 5wt%Pd1wt%Ni/TiO2 bimetallic catalysts. These composition-size plots of bimetallic nanoparticles provide critical information on the synthesis–structure–performance relationships of nanostructured bimetallic or multiphase heterogeneous catalysts. Fig. 16 Open in new tabDownload slide Composition-size plots of (a) 2wt%Pd1wt%Cu/γ-Al2O3 and (b) 5wt%Pd1wt%Ni/TiO2 bimetallic catalysts. These composition-size plots of bimetallic nanoparticles provide critical information on the synthesis–structure–performance relationships of nanostructured bimetallic or multiphase heterogeneous catalysts. Figure 16b shows another example of studying bimetallic catalysts. The composition-size plot clearly shows that the composition of the individual Pd–Ni particles in a 5wt%Pd1wt%Ni/TiO2 bimetallic catalyst changes significantly with the sizes of the individual nanoparticles and also varies with their relative locations with respect to the substrate structure: smaller particles contain more Pd and larger particles contain more Ni. This observation can be explained if Pd preferentially segregates to the particle surface or if Pd-rich particles do not sinter as much during the catalyst preparation processes. By changing the synthesis procedures, the composition-size profile can be modified. Comparison of composition-size plots to the catalyst's performance can provide important information on the synthesis–structure–performance relationships. In practical applications to developing industrial catalysts, hundreds of data points in each composition-size plot are usually needed to provide statistically meaningful data of the catalyst of interest. In developing industrial catalysts, in order to optimize the synthesis protocols to make a potential commercial catalyst tens or even hundreds of catalysts will have to be tested and analyzed; this is time consuming and extremely expensive. Automated analyses and faster data acquisition systems are highly desired for wide applications of the composition-size plot method for solving challenging nanoscale materials problems. A consequence of using small electron probes to achieve high-spatial resolution is that the X-ray signal originates from a much smaller volume; thus, a weaker signal is collected and longer acquisition times are usually needed to obtain statistically meaningful data points. Specimen-drift correction, either manually or automatically, is usually required for obtaining statistically meaningful X-ray signals when very small particles are analyzed. Nevertheless, XEDS can detect the presence of just a few atoms if the analyzed volume is small enough. With the recent development of Cs-correctors for dedicated STEM instruments, the total probe current can be significantly increased, thereby providing higher counts of the collected X-ray signals. Future development that focuses on significantly improving the X-ray collection efficiency and reduction of beam-induced modifications (e.g. by using pulsed electron beam) can have a profound impact on the fundamental understanding of bimetallic nanoparticles or quantum dots. Elemental maps can provide valuable information on the 2-D elemental distributions; they are especially useful for characterizing multiphase materials. Electron–specimen interaction processes and the effective generation volume of the X-rays determine the ultimate resolution of X-ray mapping of nanophase materials. In practice, however, the extremely low counts of the collected X-ray signal from small nanoparticles and specimen drifting limit the achievable resolution in X-ray elemental maps. A statistically meaningful elemental map requires longer acquisition time that in turn requires the use of automatic drift correction or ultra-stable microscopes and samples. Higher-resolution maps, however, are needed to determine the degree of surface segregations, which may be accomplished by using Cs-correctors to reduce the probe size but still having enough beam current. Automatic specimen-drift correction may also have to be used to reduce specimen-drifting effect. The conditions for optimum X-ray mapping include: (i) high beam current within a small probe size, (ii) high X-ray collection efficiency and (iii) long acquisition times per pixel if automatic specimen-drift correction techniques are used. Figure 17 shows XEDS spectra and the elemental maps of a 5wt%Pd1wt%Ni/TiO2 catalyst, clearly revealing that the individual nanoparticles dispersed onto the TiO2 powders contain both Pd and Ni. The elemental distribution within each individual Pd–Ni bimetallic nanoparticles, especially within the small nanoparticles, however, is not revealed in these elemental maps. Information on the surface segregation of individual Pd–Ni bimetallic nanoparticles, which is most important for understanding the performance of bimetallic nanocatalysts, cannot be extracted from Fig. 17. Fig. 17 Open in new tabDownload slide XEDS spectra from a Pd–Ni alloy nanoparticle (darker line) and from the TiO2 support (lighter line) of a 5wt%Pd3wt%Ni/TiO2 bimetallic catalyst. Elemental maps of Ni, Pd and Ti show the distribution of Pd and Ni across the TiO2 support. Fig. 17 Open in new tabDownload slide XEDS spectra from a Pd–Ni alloy nanoparticle (darker line) and from the TiO2 support (lighter line) of a 5wt%Pd3wt%Ni/TiO2 bimetallic catalyst. Elemental maps of Ni, Pd and Ti show the distribution of Pd and Ni across the TiO2 support. Electron energy-loss signals carry detailed information on the composition, chemistry and electronic structure of nanoparticles with atomic resolution and sensitivity. The combination of atomic resolution HAADF imaging with EELS has already proved extremely valuable for extracting atomic-scale information on the composition and electronic structure of various materials systems [74,75,78–80,86–89, 134,135]. The combination of EELS with HAADF in a STEM instrument significantly extends the usefulness of STEM in solving challenging nanoscale or atomic scale materials problems. It is important to understand the effect of the electronic structure of the interfacial regions between the metal nanoparticles and the support on the catalytic performances of supported catalysts. Since catalytic reactions usually involve bonding and electron transfer processes, the electronic properties of the metal–support interfacial regions can play a critical role in determining surface adsorption and electron transfer processes. Furthermore, the interfacial regions may also act as active sites during the catalytic reactions since these regions may have a structure that represents neither the metal nanoparticles nor the support. Knowledge of the atomic and electronic structure of the interfaces can help us better understand the performance of heterogeneous catalysts. Recently, atomic resolution EELS and HAADF techniques have been applied to the study of nanophase Sn(Sb)O2 catalysts [86], reduction behavior of metal nanoparticles in alumina-supported Pd catalysts [87] and the alloying behavior of supported Pd–Cu bimetallic catalysts [88]. These preliminary investigations already showed that atomic resolution EELS, together with HAADF imaging technique, can provide valuable information on the fundamental understanding of the electronic structure as well as surface composition of the individual nanoparticles and their interactions with the support. Figure 18a shows a HAADF image and Fig. 18b shows the corresponding EELS spectra that were obtained from the different regions labeled in 18a, respectively. The sample is a Pd–Ni/TiO2 bimetallic catalyst. The EELS spectra suggest that the bimetallic particle can be described by the ‘grape’ model: a thin skin of pure Pd layer encapsulates a Pd–Ni alloy core. This type of structure may have profound effect on the adsorption and catalytic properties of bimetallic catalysts. The knowledge of preferential surface segregation of individual bimetallic nanoparticles is critical to designing supported bimetallic catalysts and to an understanding of their performance. Knowing how the preferential surface segregation of individual bimetallic nanoparticles depends on the particle size, composition and the catalyst preparation methods, one can make significant progress in tuning the properties of supported bimetallic catalysts to achieve high selectivity and activity. The concept of binding-energy engineering in developing heterogeneous catalysts refers to tuning the molecular adsorption and dissociation behavior of nanoparticles by controlling their sizes, atomic structure, shape, surface and bulk composition, and interface structures. Fig. 18 Open in new tabDownload slide HAADF image of a 5wt%Pd1wt%Ni/TiO2 bimetallic catalyst shows a Pd–Ni alloy nanoparticle in contact with the TiO2 support (a). (b) EELS spectra, obtained from the corresponding points of the Pd–Ni alloy nanoparticle shown in (a), show that the Pd is preferentially segregated to the surface of the alloy nanoparticle. The spectra also reveal the presence of TiOx species across the surface of the Pd–Ni alloy nanoparticle. Fig. 18 Open in new tabDownload slide HAADF image of a 5wt%Pd1wt%Ni/TiO2 bimetallic catalyst shows a Pd–Ni alloy nanoparticle in contact with the TiO2 support (a). (b) EELS spectra, obtained from the corresponding points of the Pd–Ni alloy nanoparticle shown in (a), show that the Pd is preferentially segregated to the surface of the alloy nanoparticle. The spectra also reveal the presence of TiOx species across the surface of the Pd–Ni alloy nanoparticle. Figure 18 also shows that strong metal support interactions occurred during the catalyst preparation processes; TiOx species had migrated to the surfaces of the Pd–Ni alloy nanoparticle. Detailed analyses of the EELS spectra revealed that the TiOx species covered the whole surface of this particular Pd–Ni alloy particle. The coverage of TiOx species on metal or alloy nanoparticles depend on the precursor species, catalyst treatment, properties of the particle surface, and the interfacial properties between the metal/alloy particles and the support. The metal–support interaction profoundly affects the adsorption behavior of the metal or alloy nanoparticles. EELS technique can be used to analyze individual atomic columns or single atoms located either inside or on the surface of a substrate [74]. With the use of Cs-correctors to form sub-angstrom probes, it is expected that atomic resolution EELS can be performed on a wide variety of materials to provide chemical and electronic structure of the elements of interest. Further incorporation of monochromators in the field emission TEM/STEM instruments [77] will significantly enhance the power of atomic resolution EELS in studying nanoparticles and nanoparticle systems; and information on the fine electronic structure of nanoparticles and nanoclusters may provide a better understanding of the selectivity of heterogeneous catalysts. Concluding remarks In this paper, we discussed the recent development of STEM techniques and illustrated their applications by using nanoparticles and nanostructured catalysts as examples. The various imaging, spectroscopy and diffraction techniques can now be realized in both the dedicated STEM and the modern FEG TEM/STEM instruments and can be successfully applied to the study of nanoparticles or other nanoscale systems. The combination of the various STEM techniques significantly expands the usefulness of electron microscopy in solving critical problems in nanoscience and nanotechnology. The use of Cs-correctors and monochromators in the next-generation electron microscopes will undoubtedly make STEM techniques indispensable for understanding the fundamental properties of materials at a nanometer or subnanometer scale. The realization of sub-angstrom electron probes in STEM instruments makes it possible for us to explore the nature of nanoparticles and other nanoscale systems with unprecedented imaging and spectroscopy tools. Advanced STEM techniques will undoubtedly make significant contribution to the recent explosive research activities in nanoscience and nanotechnology. The intrinsic nature of high-spatial resolution techniques, however, poses the most significant challenge for the electron microscopy community: how to obtain statistically meaningful data with high-throughput and automation. With the rapid development of novel microscope designs, super fast computers, significantly improved signal acquisition systems and innovative image analysis algorithms, it is expected that statistically meaningful data should be routinely obtainable with user-friendly, high-throughput and automated high-resolution electron microscopes. Dedicated to the memory of the late Professor John M. Cowley. The author is deeply indebted to Professor John M. Cowley for his continuous mentoring and coaching for over 20 years, especially for his tireless training of the author in mastering the various STEM techniques and for his encouragement in applying these powerful STEM techniques to solving challenging industrial problems. References 1 Cowley J M ( 1969 ) Image contrast in transmission scanning electron microscopy. Appl. Phys. Lett. 15 : 58 –60. Google Scholar 2 Cowley J M ( 1976 ) The extension of scanning transmission electron microscopy by use of diffraction information. Ultramicroscopy 1 : 255 –262. Google Scholar 3 Cowley J M ( 1976 ) Scanning transmission electron microscopy of thin specimens. Ultramicroscopy 2 : 3 –16. Google Scholar 4 Cowley J M and Spence J C H ( 1978 ) Innovative imaging and microdiffraction in STEM. Ultramicroscopy 3 : 433 –438. Google Scholar 5 Spence J C H and Cowley J M ( 1978 ) Lattice imaging in STEM. Optik 50 : 129 –142. Google Scholar 6 Cowley J M and Au A V (1978) Image signals and detector configurations for STEM. Scan. Electron Microsc . 1 : 53 –60. Google Scholar 7 Cowley J M ( 1979 ) High resolution studies of crystals using STEM. Chemica Scripta 14 : 33 –38. Google Scholar 8 Cowley J M ( 1981 ) Coherent interference effects in STEM and CBED. Ultramicroscopy 7 : 19 –26. Google Scholar 9 Cowley J M ( 1982 ) Microdiffraction, STEM imaging and ELS at crystal surfaces. Ultramicroscopy 9 : 231 –236. Google Scholar 10 Butler J H and Cowley J M ( 1983 ) Phase contrast imaging using a scanning transmission electron microscope. Ultramicroscopy 12 : 39 –50. Google Scholar 11 Cowley J M ( 1984 ) Scanning transmission electron microscopy and microdiffraction techniques. Bull. Mater. Sci. 6 : 477 –490. Google Scholar 12 Cowley J M ( 1986 ) Electron diffraction phenomena observed with a high resolution STEM instrument. J. Electron Microsc. Tech. 3 : 25 –44. Google Scholar 13 Lin J A and Cowley J M ( 1986 ) Calibration of operating parameters for an HB-5 STEM instrument. Ultramicroscopy 19 : 31 –42. Google Scholar 14 Lin J A and Cowley J M ( 1986 ) Reconstruction from in-line holograms by digital processing. Ultramicroscopy 19 : 179 –190. Google Scholar 15 Liu J and Cowley J M ( 1987 ) High-resolution scanning electron microscopy of surface reactions. Ultramicroscopy 23 : 463 –472. Google Scholar 16 Liu J and Cowley J M ( 1988 ) High resolution secondary electron imaging in a scanning transmission electron microscopy instrument. Scanning Microsc. 2 : 65 –81. Google Scholar 17 Liu J and Cowley J M ( 1988 ) Contrast and resolution of secondary electron images in a scanning transmission electron microscope. Scanning Microsc. 2 : 1957 –1970. Google Scholar 18 Wang Z L and Cowley J M ( 1989 ) Simulating high-angle annular dark-field STEM images including inelastic thermal diffuse scattering. Ultramicroscopy 31 : 437 –454. Google Scholar 19 Cowley J M ( 1990 ) High-resolution side-band holography with a STEM instrument. Ultramicroscopy 34 : 293 –297. Google Scholar 20 Liu J and Cowley J M ( 1991 ) Imaging with high-angle scattered electrons and secondary electrons in STEM. Ultramicroscopy 37 : 50 –71. Google Scholar 21 Liu J ( 1990 ) Atomic resolution imaging in the STEM. In: Proceedings 12th International Congress on Electron Microscopy, Vol. 1, eds Peache, L D and Williams D B, pp. 33–34 (San Francisco Press, San Francisco). Google Scholar 22 Cowley J M ( 1992 ) Twenty forms of electron holography. Ultramicroscopy 41 : 335 –348. Google Scholar 23 Liu J and Cowley J M ( 1993 ) Scanning reflection electron microscopy and associated techniques for surface studies. Ultramicroscopy 48 : 381 –416. Google Scholar 24 Cowley J M ( 1993 ) Configured detectors for STEM imaging of thin specimens. Ultramicroscopy 49 : 4 –13. Google Scholar 25 Liu J and Cowley J M ( 1993 ) High-resolution scanning transmission electron microscopy. Ultramicroscopy 52 : 335 –346. Google Scholar 26 Cowley J M and Liu J ( 1993 ) Contrast and resolution in REM, SEM and SAM. Surf. Sci. 298 : 456 –467. Google Scholar 27 Cowley J M and Gribelyuk M A ( 1994 ) High resolution coherent imaging and holography in STEM. Bull. Microsc. Soc. Am. 24 : 438 –450. Google Scholar 28 Cowley J M, Hansen M S, and Wang S Y ( 1995 ) Imaging modes with an annular detector in STEM. Ultramicroscopy 58 : 18 –24. Google Scholar 29 Cowley J M, Merkulov V I, and Lannin J S ( 1996 ) Imaging of light-atom nanocrystals with a thin annular detector in STEM. Ultramicroscopy 65 : 61 –70. Google Scholar 30 Mankos M, Cowley J M, and Scheinfein M R ( 1996 ) Quantitative micromagnetics at high spatial resolution using far-out-focus STEM electron holography. Phys. Status Solidi A 154 : 469 –504. Google Scholar 31 Cowley J M, Spence J C H, and Smirnov V V ( 1997 ) The enhancement of electron microscope resolution by use of atomic focusers. Ultramicroscopy 68 : 135 –148. Google Scholar 32 Cowley J M, Dunin-Borkowski R E, and Hayward M ( 1998 ) The contrast of images formed by atomic focusers. Ultramicroscopy 72 : 223 –232. Google Scholar 33 Sanchez M and Cowley J M ( 1998 ) The imaging properties of atomic focusers. Ultramicroscopy 72 : 213 –222. Google Scholar 34 Cowley J M, Ooi N, and Dunin-Borkowski R E ( 1999 ) Moire patterns in electron microscopy with atomic focuser crystals. Acta Crystallogr. A 55 : 533 –542. Google Scholar 35 Cowley J M ( 2000 ) Electron holography with atomic focusers. Phys. Rev. Lett. 84 : 3618 –3621. Google Scholar 36 Cowley J M ( 2000 ) Atomic-focuser imaging in electron nanodiffraction from carbon nanoshells. Ultramicroscopy 81 : 47 –55. Google Scholar 37 Cowley J M ( 2001 ) STEM imaging with a thin annular detector. J. Electron Microsc. 50 : 147 –155. Google Scholar 38 Cowley J M and Winterton J ( 2001 ) Ultra-high-resolution electron microscopy of carbon nanotube walls. Phys. Rev. Lett. 87 : 016101 –016104. Google Scholar 39 Smirnov V V and Cowley J M ( 2002 ) In-line electron holography with an atomic focuser source. Phys. Rev. B 65 : 064109 . Google Scholar 40 Cowley J M ( 2003 ) Ultra-high resolution with off-axis STEM holography. Ultramicroscopy 96 : 163 –166. Google Scholar 41 Cowley J M ( 2004 ) Off-axis STEM or TEM holography combined with four-dimensional diffraction imaging. Microsc. Microanal. 10 : 9 –15. Google Scholar 42 Cowley J M and Jap B K ( 1976 ) The use of diffraction information to augment STEM imaging. Scan. Electron Microsc. 9 : 377 –384. Google Scholar 43 Cowley J M ( 1979 ) Adjustment of a STEM instrument by use of shadow images. Ultramicroscopy 4 : 413 –418. Google Scholar 44 Cowley J M ( 1981 ) Electron microdiffraction studies of the potential field at crystal surfaces. Ultramicroscopy 7 : 181 –188. Google Scholar 45 Cowley J M ( 1981 ) Coherent convergent beam electron microdiffraction. Sov. Phys. Crystallogr. 26 : 549 –553. Google Scholar 46 Cowley J M and Roy R A (1981) Microdiffraction from gold microcrystals. Scanning Electron Microsc. 1 : 143 –152. Google Scholar 47 Cowley J M and Spence J C H ( 1981 ) Convergent beam electron microdiffraction from small crystals. Ultramicroscopy 6 : 359 –366. Google Scholar 48 Cowley J M ( 1982 ) Surface energies and surface structure of small crystals studied by use of a STEM instrument. Surf. Sci. 114 : 587 –606. Google Scholar 49 Zhu J and Cowley J M ( 1983 ) Microdiffraction from antiphase domain boundaries in copper-gold (Cu3Au). Acta Crystall. A-Crys. A 38 : 718 –724. Google Scholar 50 Zhu J and Cowley J M ( 1983 ) Microdiffraction from stacking faults and twin boundaries in fcc. crystals. J. Appl. Crystallogr. 16 : 171 –175. Google Scholar 51 Monosmith W B and Cowley J M ( 1983 ) Electron microdiffraction from very small gold particles. Ultramicroscopy 12 : 177 –183. Google Scholar 52 Cowley J M ( 1983 ) The STEM approach to the imaging of surfaces and small particles. J. Microsc. 129 : 253 –261. Google Scholar 53 Cowley J M, Osman M A, and Humble P ( 1984 ) Nanodiffraction from platelet defects in diamond. Ultramicroscopy 15 : 311 –318. Google Scholar 54 Cowley J M ( 1984 ) Microdiffraction and STEM of interfaces. Ultramicroscopy 14 : 27 –35. Google Scholar 55 Cowley J M and Plano R J ( 1987 ) Microdiffraction study of gold–ruthenium catalyst particles. J. Catal. 108 : 199 –207. Google Scholar 56 Pan M, Cowley J M, and Chan I Y ( 1987 ) The structure of platinum particles on γ-alumina support. J. Appl. Crystallogr. 20 : 300 –305. Google Scholar 57 Cowley J M and Ou H J ( 1989 ) Observation of microdiffraction patterns with a dedicated STEM instrument. J. Electron Microsc. Tech. 11 : 143 –154. Google Scholar 58 Cowley J M ( 1996 ) Electron nanodiffraction: progress and prospects. J. Electron Microsc. 45 : 3 –10. Google Scholar 59 Cowley J M and Packard S D ( 1996 ) Coherent nanodiffraction from phase objects: carbon nanotubes. Ultramicroscopy 63 : 39 –47. Google Scholar 60 Cowley J M, Nikolaev P, Thess A, and Smalley R E ( 1997 ) Electron nano-diffraction study of carbon single-walled nanotube ropes. Chem. Phys. Lett. 265 : 379 –384. Google Scholar 61 Cowley J M ( 1997 ) Applications of STEM instruments for surface studies. Surf. Rev. Lett. 4 : 567 –575. Google Scholar 62 Cowley J M ( 1999 ) Electron nanodiffraction. Microsc. Res. Tech. 46 : 75 –97. Google Scholar 63 Cowley J M, Janney D E, Gerkin R C, and Buseck P R ( 2000 ) The structure of ferritin cores determined by electron nanodiffraction. J. Struct. Biol. 131 : 210 –216. Google Scholar 64 Cowley J M ( 2001 ) Electron nanodiffraction methods for measuring medium-range order. Ultramicroscopy 90 : 197 –206. Google Scholar 65 Quintana C, Cowley J M, and Marhic C ( 2004 ) Electron nanodiffraction and high-resolution electron microscopy studies of the structure and composition of physiological and pathological ferritin. J. Struct. Biol. 147 : 166 –178. Google Scholar 66 Cowley J M ( 2004 ) Applications of electron nanodiffraction. Micron 35 : 345 –360. Google Scholar 67 James E M, Browning N D, Nicholls A W, Kawasaki M, Xin Y, and Summer S ( 1998 ) Demonstration of atomic resolution Z-contrast imaging in a 200 kV field emission TEM/STEM. J. Electron Microsc. 47 : 561 –574. Google Scholar 68 James E M and Browning N D ( 1999 ) Practical aspects of atomic resolution imaging and analysis in STEM. Ultramicroscopy 78 : 125 –139. Google Scholar 69 Browning N D, Arslan I, Ito Y, James E D, Klie R F, Moeck P, Topuria T, and Xin Y ( 2001 ) Application of atomic scale STEM techniques to the study of interfaces and defects in materials. J. Electron Microsc. 50 : 205 –218. Google Scholar 70 Krivanek O L, Dellby N, and Lupini A R ( 1999 ) Towards sub-A electron beams. Ultramicroscopy 78 : 1 –11. Google Scholar 71 Dellby N, Krivanek O L, Nellist P D, Batson P E, and Lupini A R ( 2001 ) Progress in aberration-corrected scanning transmission electron microscopy. J. Electron Microsc. 50 : 177 –185. Google Scholar 72 Batson P E, Dellby N, and Krivanek O L ( 2002 ) Sub-angstrom resolution using aberration corrected electron optics. Nature 418 : 617 –620. Google Scholar 73 Krivanek O L, Nellist P D, Delby N, Murfitt M F, and Szilagyi Z ( 2003 ) Towards sub-0.5 angstrom electron beams. Ultramicroscopy 96 : 229 –237. Google Scholar 74 Allen L J, Findlay S D, Lupini A R, Oxley M P, and Pennycook S J ( 2003 ) Atomic-resolution electron energy loss spectroscopy imaging in aberration corrected scanning transmission electron microscopy. Phys. Rev. Lett. 91 : 105503/1–105503/4. Google Scholar 75 Varela M, Findlay S D, Lupini A R, Christen H M, Borisevich A Y, Dellby N, Krivanek O L, Nellist P D, Oxley M P, Allen L J, and Pennycook S J ( 2004 ) Spectroscopic imaging of single atoms within a bulk solid. Phys. Rev. Lett. 92 : 095502/1–095502/4. Google Scholar 76 Nellist P D, Chisholm M F, Delby N, Krivanek O L, Murfitt M F, Szilagyi Z S, Lupini A R, Borisevich A, Sides W H Jr, and Pennycook S J ( 2004 ) Direct sub-angstrom imaging of a crystal lattice. Science 305 : 1741 . Google Scholar 77 Ziegler A, Idrobo J C, Cinibulk M K, Kisielowski C, Browning N D, and Ritchie R O ( 2004 ) Interface structure and atomic bonding characteristics in silicon nitride ceramics. Science 306 : 1768 –1770. Google Scholar 78 Batson P E ( 1993 ) Simultaneous STEM imaging and electron-energy-loss spectroscopy with atomic column sensitivity. Nature 366 : 727 –728. Google Scholar 79 Silcox J ( 1998 ) Core-loss EELS. Curr. Opin. Solid State Mater. Sci . 3 : 336 –342. Google Scholar 80 Muller D A and Mills M J ( 1999 ) Electron microscopy: probing the atomic structure and chemistry of grain boundaries, interfaces, and defects. Mater. Sci. Eng. A 260 : 12 –28. Google Scholar 81 Pennycook S J ( 2002 ) Structure determination through Z-contrast microscopy. Adv. Imag. Electron Phys. 123 : 173 –206. Google Scholar 82 McBride J R, Kippeny T C, Pennycook S J, and Rosenthal S J ( 2004 ) Aberration-corrected Z-contrast scanning transmission electron microscopy of CdSe nanocrystals. Nano Lett. 4 : 1279 –1283. Google Scholar 83 Treacy M M J and Rice S B ( 1989 ) Catalyst particle sizes from Rutherford scattered intensities. J. Microsc . 156 : 211 –234. Google Scholar 84 Liu J and Cowley J M ( 1990 ) High-angle ADF and high-resolution SE imaging of supported catalyst clusters. Ultramicroscopy 34 : 119 –128. Google Scholar 85 Nellist P D and Pennycook S J ( 1996 ) Direct imaging of the atomic configuration of ultradispersed catalysts. Science 274 : 413 –415. Google Scholar 86 Sun K, Liu J, and Browning N D ( 2002 ) Correlated atomic resolution microscopy and spectroscopy studies of Sn(Sb)O2 nanophase catalysts. J. Catal . 205 : 266 –277. Google Scholar 87 Sun K, Liu J, Nag N K, and Browning N D ( 2002 ) Studying the metal–support interaction in Pd/γ-Al2O3 catalysts by atomic-resolution electron energy-loss spectroscopy. Catal. Lett. 84, 193–199. Google Scholar 88 Sun K, Liu J, Nag N K, and Browning N D ( 2002 ) Atomic scale characterization of supported Pd-Cu/γ-Al2O3 bimetallic catalysts. J. Phys. Chem. B 106 : 12239 –12246. Google Scholar 89 Liu J ( 2004 ) Advanced electron microscopy characterization of nanostructured heterogeneous catalysts. Microsc. Microanal. 10 : 55 –76. Google Scholar 90 Wang S, Borisevich A Y, Rashkeev S N, Glazoff M V, Sohlberg K, Pennycook S J, and Pantelides S T ( 2004 ) Dopants adsorbed as single atoms prevent degradation of catalysts. Nat. Mater. 3 : 143 –146. Google Scholar 91 Morton G A and Ramberg E G ( 1939 ) Point projector electron microscope. Phys. Rev. 56 : 705 . Google Scholar 92 Gabor D ( 1949 ) Microscopy by reconstructed wave fronts. Proc. R. Soc. Lond. A 197 : 454 –486. Google Scholar 93 Binh V T, Garcia N, and Purcell S T ( 1996 ) Electron field emission from atom-sources: fabrication, properties, and applications of nanotips. Adv. Imag. Electron Phys. 95 : 63 –153. Google Scholar 94 Ronchi V ( 1964 ) Forty years of history of a grating interferometer. Appl. Optics 3 : 437 –451. Google Scholar 95 Crew A V, Wall J, and Langmore J ( 1970 ) Visibility of a single atom. Science 168 : 1338 –1340. Google Scholar 96 Howie A ( 1979 ) Image contrast and localized signal selection techniques. J. Microsc. 117 : 11 –23. Google Scholar 97 Wahlström E, Lopez N, Schaub R, Thostrup P, Rønnau A, Africh C, Lægsgaard E, Nørskov J K, and Besenbacher F ( 2003 ) Bonding of gold nanoclusters to oxygen vacancies on rutile TiO2(110). Phys. Rev. Lett. 90 : 026101/1–026101/4. Google Scholar 98 Ade G ( 1977 ) On the incoherent imaging in the scanning transmission electron microscope. Optik 49 : 113 –116. Google Scholar 99 Russell F L, Kirkland E J, and Silcox J ( 1988 ) Visibility of single heavy atoms on thin crystalline silicon in simulated annular dark-field STEM images. Acta Crystallogr. A 44 : 912 –927. Google Scholar 100 Pennycook S J and Jesson D E ( 1990 ) High-resolution incoherent imaging of crystals. Phys. Rev. Lett. 64 : 938 –941. Google Scholar 101 Loane R F, Xu P, and Silcox J ( 1991 ) Thermal vibrations in convergent-beam electron diffraction. Acta Crystallogr. A 47 : 267 –278. Google Scholar 102 Loane R F, Xu P, and Silcox J ( 1992 ) Incoherent imaging of zone axis crystals with ADF STEM. Ultramicroscopy 40 : 121 –138. Google Scholar 103 Treacy M M J and Gibson J M ( 1993 ) Coherence and multiple-scattering in Z-contrast images. Ultramicroscopy 52 : 31 –53. Google Scholar 104 Treacy M M J and Gibson J M ( 1995 ) Atomic contrast transfer in annular-dark-field images. J. Microsc. 180 : 2 –11. Google Scholar 105 Hartel P, Rose H, and Dinges C ( 1996 ) Conditions and reasons for incoherent imaging in STEM. Ultramicroscopy 63 : 93 –114. Google Scholar 106 Amali A and Rez P ( 1997 ) Theory of lattice resolution in high-angle annular dark-field images. Microsc. Microanal. 3 : 28 –46. Google Scholar 107 Nellist P D and Pennycook S J ( 1999 ) Incoherent imaging using dynamically scattered coherent electrons. Ultramicroscopy 78 : 111 –124. Google Scholar 108 Nellist P D and Pennycook S J ( 2000 ) The principles and interpretation of annular dark-field Z-contrast imaging. Adv. Imag. Elect. Phys. 113 : 147 –203. Google Scholar 109 Rafferty B, Nellist P D, and Pennycook S J ( 2001 ) On the origin of transverse incoherence in Z-contrast STEM. J Electron Microsc. 50 : 227 –233. Google Scholar 110 Allen L J, Findlay S D, Oxley M P, and Rossouw C J ( 2003 ) Lattice resolution contrast from a focused coherent electron probe: part I. Ultramicroscopy 96 : 47 –63. Google Scholar 111 Findlay S D, Allen L J, Oxley M P, and Rossouw C J ( 2003 ) Lattice resolution contrast from a focused coherent electron probe: part II. Ultramicroscopy 96 : 65 –81. Google Scholar 112 Peng Y P, Nellist P D, and Pennycook S J ( 2004 ) HAADF-STEM imaging with sub-angstrom probes: a full Bloch wave analysis. J. Electron Microsc. 53 : 257 –266. Google Scholar 113 Fertig J and Rose H ( 1981 ) Resolution and contrast of crystalline objects in high-resolution scanning transmission electron microscopy. Optik 59 : 407–429. Google Scholar 114 Pennycook S J and Jesson D E ( 1991 ) High-resolution Z-contrast imaging of crystals. Ultramicroscopy 37 : 14 –38. Google Scholar 115 Voyles P M, Muller D A, and Kirkland E J ( 2004 ) Depth-dependent imaging of individual dopant atoms in silicon. Microsc. Microanal. 10 : 291 –300. Google Scholar 116 Chapman J M, Batson P E, Waddell E M, and Ferrier R P ( 1978 ) The direct determination of magnetic domain wall profiles by differential phase contrast electron microscopy. Ultramicroscopy 3 : 203 —214. Google Scholar 117 Mankos M, Cowley J M, and Scheinfein M R ( 1996 ) Quantitative micromagnetics at high spatial resolution using far-out-focus STEM electron holography. Phys. Stat. Solidi A 154 : 469 –504. Google Scholar 118 Hoppe W ( 1982 ) Trace structure analysis, ptychography, phase tomography. Ultramicroscopy 10 : 187 –198. Google Scholar 119 Rodenburg J M and Bates R H T ( 1992 ) The theory of super-resolution electron microscopy via Wigner-distribution deconvolution. Phil. Trans. R. Soc. Lond. A 339 : 521 –553. Google Scholar 120 Nellist P D, McCallum B C, and Rodenburg J M ( 1995 ) Resolution beyond the “information limit” in transmission electron microscopy. Nature 374 : 630 –632. Google Scholar 121 Kruit P and Venables J A ( 1988 ) High-spatial-resolution surface-sensitive electron spectroscopy using a magnetic parallelize. Ultramicroscopy 25 : 183 –193. Google Scholar 122 Hembree G G, Crozier P A, Drucker J S, Krishnamurthy M, Venables J A, and Cowley J M ( 1989 ) Biassed secondary electron imaging in a UHV-STEM. Ultramicroscopy 31 : 111 —115. Google Scholar 123 Drucker J, Scheinfein M R, Liu J, and Weiss J K ( 1993 ) Electron coincidence spectroscopy studies of secondary and Auger electron generation mechanisms. J. Appl. Phys. 74 : 7329 –7339. Google Scholar 124 Liu J ( 2000 ) Contrast of highly dispersed metal nanoparticles in high-resolution secondary electron and backscattered electron images of supported metal catalysts. Microsc. Microanal. 6 : 388 –399. Google Scholar 125 Liu J, Hembree G, Spinnler G, and Venables J ( 1992 ) High-resolution Auger electron spectroscopy and microscopy of a supported metal catalyst. Surf. Sci. 262 : L111 –L117. Google Scholar 126 Liu J, Hembree G, Spinnler G, and Venables J ( 1992 ) High resolution Auger electron spectroscopy and microscopy of small metal particles. Catal. Lett. 15 : 133 –143. Google Scholar 127 Liu J, Hembree G, Spinnler G, and Venables J ( 1993 ) Nanometer-resolution surface analysis with Auger electrons. Ultramicroscopy 52 : 369 –376. Google Scholar 128 Pan M, Cowley J M, and Barry J C ( 1989 ) Coherent electron microdiffraction from small metal particles. Ultramicroscopy 30 : 385 –394. Google Scholar 129 Tehuacanero S, Herrera R, Avalos M, and Jose-Yacaman M ( 1992 ) High resolution TEM studies of gold and palladium nano-particles, Acta. Metall. Mater. 40 : 1663 –1674. Google Scholar 130 Marks L D ( 1994 ) Experimental studies of small particle structures. Rep. Prog. Phys. 57 : 603 –649. Google Scholar 131 Zuo J M, Vartanyants I, Gao I M, Zhang R, and Nagahara L A ( 2003 ) Atomic resolution imaging of a single double-wall carbon nanotube from diffraction intensities. Science 300 : 1419 –1421. Google Scholar 132 Lyman C E, Lakis R E, and Stenger H G Jr ( 1995 ) X-ray emission spectrometry of phase separation in Pt–Rh nanoparticles for nitric oxide reduction. Ultramicroscopy 58 : 25 –34. Google Scholar 133 Lyman C E, Lakis R E, Stenger H G Jr, Totdal B, and Prestvik R ( 2000 ) Analysis of alloy nanoparticles. Mikrochim. Acta 132 : 301 –308. Google Scholar 134 Browning N D, Chisholm M F, Pennycook S J, Norton D P, and Lowndes D H ( 1993 ) Atomic-resolution chemical analysis using a scanning transmission electron microscope. Nature 366 : 143 –146. Google Scholar 135 Browning N D, Wallis D J, Nellist P D, and Pennycook S J ( 1997 ) EELS in the STEM: determination of materials properties on the atomic scale. Micron 28 : 333 –348. Google Scholar © The Author 2005. Published by Oxford University Press on behalf of Japanese Society of Microscopy. All rights reserved. For permissions, please email: journals.permissions@oxfordjournals.org
TI - Scanning transmission electron microscopy and its application to the study of nanoparticles and nanoparticle systems
JF - Journal of Electron Microscopy
DO - 10.1093/jmicro/dfi034
DA - 2005-06-01
UR - https://www.deepdyve.com/lp/oxford-university-press/scanning-transmission-electron-microscopy-and-its-application-to-the-tRIVAwnFG0
SP - 251
EP - 278
VL - 54
IS - 3
DP - DeepDyve
ER -