Читать книгу Amorphous Nanomaterials - Lin Guo - Страница 27

One of the biggest challenges in increasing resolution of the electron microscopy image has always been the improvement of image blurring that is caused by lens aberration. Even today, 80 years after the invention of TEM, the point resolution is still limited by the spherical aberration of its objective lens. In addition, the solution to improve this was to insert a numerical correction for this spherical aberration that was based either on a series of object images taken under variable objective focus conditions [1] or on the application of holograhpic technologies [2]. To improve the resolution of microscope, the phase-shifting effects of spherical aberration and other imaging defects that originated from second-, third-order objective lens astigmatism and misalignment coma need to be removed. The introduction of the aberration corrector then allows the microscopes to reach their technique limit, which is determined either by envelope functions due to beam divergence (spatial coherence) or focal spread (temporal coherence) or by incoherent effects, such as mechanical vibrations or stray magnetic fields. With the achievement of an aberration correction up to fifth order, microscope information limits have been further improved.

When reviewing the history of improving the resolution of TEM, the earliest work reported by Scherzer [3] suggested that the two principal axial aberrations, chromatic and spherical, could be corrected by electrostatic or magnetic multipole elements; however, this was beyond the technology available at that time. The spatial resolution in a TEM was limited to about 2 Å at that time with an energy of 100–200 keV. Scherzer also pointed out three possible ways to correct the limiting aberration: (i) the use of non-round lenses, (ii) the use of lenses with charge on axes, and (iii) the use of a time-varying field. These ideas inspired many attempts to develop a practical aberration corrector for the TEM. In the late 1960s, Crewe et al. [4] introduced an alternative to the TEM imaging geometry, where they used a very small electron probe that was scanned in a raster over the research area, called STEM. Until then, STEM has become an important technique because the generated signals can be used to scan the area of interests. In particular, annular dark-field (ADF) imaging, using high-angle elastic scattering which occurs near individual atoms [5], has emerged as a new imaging technique. Compared with conventional TEM imaging, the advantages of ADF STEM were that (i) the spatial resolution was better than that of the TEM mode, (ii) it is sensitive to the atomic numbers of the imaged atoms, and (iii) it provided a positive definite transfer of specimen spatial frequencies, which allowed a direct interpretation of results with fewer ambiguities. Meanwhile, the STEM was also compatible with other analytical techniques, including EELS [6], which was possible to collect atomic information about atomic species, bonding environment, and local electronic structures [7, 8].

The path toward realizing the hardware suitable for direct aberration correction at TEM was long and arduous. To fulfill an atomic resolution in electron microscopy, in the 1990s, Rose [9] proposed a solution based on two electromagnetic hexapoles and four additional lenses. The correction was achieved as the primary aberrations of second order from the first hexapole which are compensated by the second hexapole element. This long hexapole arrangement generates a combined aberration that compensates the positive spherical aberration of the objective lens through suitable hexapole excitation. Moreover, in the early 2000s, several indirect aberration compensation methods had been developed for experimental determination of the axial aberration coefficient, which mainly relied on measurements taken from image wave functions [10] and were applicable to crystalline specimens as well as thin amorphous materials. Initially, a phase correlation function (PCF) [11] was used to determine the defocus difference between neighboring images at high accuracy. The absolute focus and astigmatism are subsequently measured from the restored image wave function of a reference image using a phase contrast index (PCI) function. This provided estimates of the coefficients of the wave aberration function, which can then recover both phase and modulus of the specimen exit wave function under either linear or nonlinear imaging, also enhancing resolution to complement direct aberration correction. After that, indirect and direct approaches have been combined [12], for a focal series data set, and the elimination of tilt-induced axial coma relaxes the requirement of using parallel illumination and enables the illumination to be converged onto the specimen area of interest, which reduced delocalization of image components in the electron optically corrected image. Also, localized compensation of higher order aberrations up to the fifth order was possible. Notably, for modern electron microscope, the Cs-TEM is equipped with double correctors. Taking JEM-2200FS as an example, when only HRTEM data are required, the upper corrector is switched off and the condenser system of the microscope can be used as normal. A small voltage is applied to one of the hexapoles in the upper corrector to compensate for a residue threefold astigmatism arising from the gun lens or any residue field from the first hexapole. When switching to STEM imaging, a small adjustment to transfer lenses in the upper corrector is possible which allows broad parallel illumination to be achieved satisfactorily with both upper hexapoles strongly excited. The correction of both pre- and post-field allows the use of a large condenser and objective aperture sizes [13].

Nowadays, as the aberration-corrected probe is much sharper and much more intense, atomic-scale microanalysis can be realized. However, for small negative-Cs HRTEM imaging, it is applicable only to very thin samples with a thickness of typically a few nanometers. On the other hand, the Cs-corrected high-angle ADF STEM imaging, using incoherent thermal diffuse scattering (TDS) electrons as the main signal sources, caused the signal of light atomic columns easily drowned out when they are located adjacent to the heavy atomic columns. This is attributed to the large difference in scattering power for specific TDS in different elements. Then, the annular bright-field (ABF) STEM realized a visualization of both light and heavy atomic column. This is due to the reduced effect of spatial coherence, which is Cs-dependent. This means that much larger bright-filed (BF) collection angles are allowed without loss of resolution. Thus, the bright-filed scanning transmission electron microscopy (BF-STEM) with aberration-corrected technique of high signal quality can be recorded, with the atomic columns identified by dark spots, which are independent of probe forming lens defocus and sample thickness [14]. Also, ABF can collect electrons with small angles relative to the direction of the incident electron beam to resolve atomic columns (Z) with high contrast [15]. The BF-STEM image is somewhat similar to an HRTEM image when a circular BF detector adopting a small collection angle that is close to an on-axis point detector, making it sensitive to light atomic columns. Nevertheless, the concern for this ABF-STEM is that the signal of light atomic columns is still weak, even though it is a breakthrough toward visualizing light atomic columns compared to those in the dark field. In the case of TEM, image delocalization at surfaces and interfaces is greatly reduced because of the highly coherent electron source. For this corrected TEM imaging, an additional benefit is the visibility of low-Z elements such as oxygen, which are adjacent to much heavier metal atoms in semiconductors, with a slightly negative Cs value [16]. Moreover, the study about the impact of beam–specimen interactions still continues, especially in case that the sample is damaged by an intense electron beam while the atoms or point defects conveniently remain in place. This is resolved by using an aberration-corrected STEM operated at 60 keV, which allows atom-by-atom structural and chemical analysis, and identification of individual atoms of low-Z elements with a negligible electron–beam damage [17]. The operation of the microscope at lower electron energies offers opportunities to characterize some important classes of materials that are ultrathin, such as single graphene sheets.

Using Cs-TEM to characterize nanomaterials with atomic resolution was an achievement in the past three decades. A contribution was to understand high-temperature superconductivity by attempting to image oxygen in YBa₂Cu₃O₇. It solved one key problem: how the occupation of specific atomic sites with oxygen influences electronic properties [18, 19]. In 1992, oxygen was accessible by an imaging technique, as shown that it was visible at atomic resolution in the electron wave function at the exit plane of the specimen when reconstructed by computational techniques [20]. With the development of aberration-corrected technique, oxygen concentration measurements were carried out for the first time in studying lattice defects in BaTiO₃ in 2004 [21]. The trend of modern aberration-corrected instruments is to produce images with quality that can virtually put finger on individual atomic positions and even the individual lateral atomic shift. One example for precise characterization of individual atoms is shown in Figure 2.1, where a ferroelectric domain boundary in the microelectronic storage material PbZr_0.2Ti_0.8O₃ of the order of 40 pm of the oxygen, zirconium, and titanium atoms out of their symmetry positions was clearly revealed [22].

This powerful technique fulfilled the old dream of materials science: a direct link between atomic-level information and macroscopic properties. Specifically, the realization of atomic resolution by aberration-corrected TEM can greatly influence the future development of semiconductor devices because their continued miniaturization relies on critical components, including those 5–10-atom-thick gate oxides in transistors [23, 24], magnetoresistive read heads with a thickness of 1–2 nm [25, 26], and tunnel junctions in magnetic memories with comparable thickness [27, 28]. The success or failure of the semiconductors mentioned above is decided by the determination of the thickness and composition of these ultrathin layers. It also gains insights into the chemistry, interdiffusion, and electronic structures of interlayers. Notably, STEM has also proved very effective in measuring the changes in compositions, electronic structure, and bonding of interfaces of those semiconductors [29, 30]. Detection of single-dopant atoms by STEM is regarded as a powerful tool to understand materials for transistor scaling [31], to detect the spatial distribution of single vacancies [32], or to study their electronic fingerprints on the local densities of states [33]. Besides, the typical examples of using Cs-corrected BF-STEM imaging can be found with the resolve of hydrogen atomic columns in a crystalline sample [34]. Also, studies on obtaining atomic resolution BF-STEM images using a medium collection angle have been carried out, where the detection of both light and heavy atomic columns with a medium collection angle for a [001]-oriented SrTiO₃ single crystal was realized [35]. This middle-angle BF-STEM imaging is particularly robust to against variations in the probe-forming lens defocus and sample thickness, which laid a good foundation to analyze realistic materials. After 2010, the ultra-STEM represents a new trend in atomic resolution imaging, which used lower acceleration voltages, the so-called “gentle STEM” [36]. The operating voltage is only 60 keV, which is well below the knock-on damage voltage of graphene, making it easy to study the intrinsic defects. Even so, the edge atoms and defects are more easily to be knocked into metastable configurations because they are weakly bonded. However, this can be taken advantages to investigate atomic dynamics of nanostructure and even created nanostructures by electron beams (Figure 2.2) [37, 38]. The ultra-STEM is well matched with the recent increasing interests in two-dimensional (2D) materials, as its simultaneous efficient ADF and EELS imaging can achieve insights into vacancy and defect configurations [39]. Meanwhile, discoveries such as ordered arrays of oxygen vacancies with dramatic effects on nanosheet properties were presented by this ultra-STEM [40].

Figure 2.1 Transversal inversion polarization domain wall in ferroelectric PZT. Arrows give the direction of the spontaneous polarization, which can be directly inferred from the local atom displacements. The shifts of the oxygen atoms (blue circles) out of the Ti/Zr-ato row (red circles) can be seen directly, as well as the change of the Ti/Zr-to-Pb (yellow circles) separation. Source: Reproduced with permission from Jia et al. [22]. Copyright 2008, Nature Publishing Group.

The studies of aberration-corrected electron microscopy are more frequently reported in scientific literature. Indeed, people are now able to see the complexities of structure and chemistry at the atomic scale never before, enabling a better understanding of reaction and transformation pathways that fabricated desirable materials and making new devices with enhanced properties. The improvement in multiple corrector systems allows aberration control of both probe size and detector field of view and also makes it possible to give precise control over amplitude and phase of the incident and scattered electrons. When applying HRTEM to the very thin specimen under negative Cs imaging conditions, even the projected atomic structure of complex crystals can be revealed because of its strong suppression of image imaging conditions. However, conventional HRTEM completely fails in obtaining directly interpretable images [41]. More studies for analysis of imperfections of complex layer compounds, such as stacking faults and layer undulations, should be carried out. As for the STEM mode, its ability to record compositional and bonding information in ultrathin materials would open up the study of inhomogeneities, including those symmetry-breaking and spatial variations in superconductors and charge-ordered materials, and also the interdiffusion and dead layer in ferroelectrics at the sub-nanolevel. Truly, the aberration corrector on TEM has brought great progress in the way of materials science, creating materials with desirable structures and properties. The journey to fabricate new devices attached to the electron microscopy is exciting and rewarding.

Figure 2.2 Controllable nanofabrication of MoSe nanowire network from a MoSe₂ monolayer by electron beam nanofabrication. Source: Reproduced with permission from Lin et al. [37]. Copyright 2014, Nature Publishing Group.