Search

This article summarizes core aspects of beam-sample interactions in research that aims at exploiting the ability to detect single atoms at atomic resolution by mid-voltage transmission electron microscopy. Investigating the atomic structure of catalytic Co3O4 nanocrystals underscores how indispensable it is to rigorously control electron dose rates and total doses to understand native material properties on this scale. We apply in-line holography with variable dose rates to achieve this goal. Genuine object structures can be maintained if dose rates below ~100 e/Å2s are used and the contrast required for detection of single atoms is generated by capturing large image series. Threshold doses for the detection of single atoms are estimated. An increase of electron dose rates and total doses to common values for high resolution imaging of solids stimulates object excitations that restructure surfaces, interfaces, and defects and cause grain reorientation or growth. We observe a variety of previously unknown atom configurations in surface proximity of the Co3O4 spinel structure. These are hidden behind broadened diffraction patterns in reciprocal space but become visible in real space by solving the phase problem. An exposure of the Co3O4 spinel structure to water vapor or other gases induces drastic structure alterations that can be captured in this manner.

Decomposition of the solid solution of an Al–3Mg–0.4Cu–0.12Si (wt.%) alloy has been studied by examining samples aged for periods of 1–16 h at a temperature of 180 °C. High resolution transmission electron microscopy (HRTEM) was the primary method of investigation. Within the 4–16 h aging interval, nanometer scale, homogeneously dispersed, coherent, lath-like particles were detected in the matrix. The orientation of the lath particles in the matrix is such that the edges are aligned along the 〈1 0 0〉 Al directions. From the Fourier transformation of the HRTEM images obtained on 〈1 0 0〉 Al zones, it was possible to confirm that two variants give rise to diffraction spots that were previously associated with the presence of S″-phase in “Cu lean” Al–Mg–Cu alloys. Two other variants give rise to streaked diffraction spots positioned around the (1 1 0) Al . The detected particles are termed GPBII in this study, instead of previously suggested S″-phase notation. The findings of this study are also compared with those of the frequently studied “Cu rich” Al–Cu–Mg alloying system.

Microstructure evolution of an age hardenable Al–3Mg–0.4Cu–0.12Si (wt%) alloy has been studied during artificial aging at 180 °C prior to the formation of the stable S-phase. The primary investigation method used in this study was high-resolution transmission electron microscopy (HRTEM), coupled with image processing and image simulation. After 1 h of aging, the presence of super-lattice reflections was detected in the Fourier spectra of the HRTEM images, suggesting an L1 0 type ordering of Mg and Cu atoms in the Al matrix. After 4 and 8 h of aging, coherent particles were observed in the microstructure. These particles give rise to diffraction spots that in previous literature have been considered to be characteristic of the S″-phase in the “Cu-lean” Al–Mg–Cu alloys. It is shown that these diffraction spots can be indexed in terms of a crystal structure that is closely related to the L1 0 ordering formed at the shorter aging times. The crystal structure is orthorhombic with lattice parameters a =1.2 nm, b =0.4 nm, c =0.4 nm and space group Cmmm. We propose to identify these coherent particles as GPB-II zones, and the ordering that precedes them as GPB zones.

This article shows that the presence of low‐temperature‐grown GaAs (LT‐GaAs) in LT‐GaAs/AlAs/GaAs:Si heterostructures increases the Al/Ga interdiffusion at the heterostructure interfaces. The interdiffusion enhancement is attributed to the presence of Ga vacancies (VGa) in the As‐rich LT‐GaAs, which diffuses from a supersaturation of VGa frozen‐in during sample growth. Chemical mapping, which distinguishes between the AlAs and GaAs lattices at an atomic scale, is used to measure the Al concentration gradient in adjacent GaAs:Si layers. A correlation is observed between the Al/Ga interdiffusion and the gate breakdown voltage in metal‐insulator field‐effect transistor structures containing LT‐GaAs.

The Transmission Electron Aberration-corrected Microscope (TEAM) project was initiated by the US Department of Energy as a collaborative effort to redesign the electron microscope around aberration-corrected optics [1], and is aimed at achieving 50 pm resolution. But the ability to resolve deep sub-Angstrom spacing entails a number of unresolved questions that can now be addressed. Among them is an ongoing debate about the physical meaning of resolution. Traditional strategies include the recording of Young’s fringes, the detection of image Fourier components from STEM images, the demonstration of a suitable peak separation in periodic lattices or signal width measurements from images of single atoms, to name a few. The drawback is that seemingly conflicting results are produced [e.g. 2]. Further, these methods define resolution through a selectable object, unlike light microscopy where resolution is instrument-defined. Two limitations of this approach are electron channeling [3, 4] and elastic scattering at single crystals [5]. The TEAM Project adopted a pragmatic view of information transfer below 50 pm: detecting Young’s fringes in TEM and (660) image Fourier components from gold (111) STEM images at 48 pm. Recently the TEAM 0.5 prototype microscope achieved this goal [1].

This paper addresses advances in electron microscopy that were accomplished over the past years with the incorporation of new electron optical components such as aberration correctors, monochromators or high brightness guns. Many of these developments are currently pursued within the DoE’s TEAM project. As a result electron microscopy has reached 50 pm resolution. In this paper it is shown how the resolution improvement has helped to boost signal to noise ratios enabling a detection of single atoms across the Periodic Table of Elements. The described achievements allow for investigations of single point defects in nanoelectronic devices even if printed on single sheets of carbon atoms (graphene). Further it is now possible to access depth information from single projections with a precision that has reached interatomic distances.

This contribution reviews the current status of sub Angstrom electron microscopy. High Resolution Transmission Electron Microscopy (HRTEM) and Scanning Transmission Electron Microscopy (STEM) are considered and compared for imaging applications. While both techniques provide comparable sub Angstrom resolution around 0.8A noise levels and chemical discrimination are dissimilar, which result in complementary characteristics for the detection of different elements. In particular, the ability to detect single atoms benefits greatly from the ongoing resolution enhancement and corrections of lens aberrations. As a result it is feasible aiming at single atom analyses in three dimensions.

Extended abstract of a paper presented at Microscopy and Microanalysis 2011 in Nashville, Tennessee, USA, August 7–August 11, 2011.

Transmission Electron Microscopy developed from an imaging tool into a quantitative electron beam characterization tool that locally accesses structure, chemistry, and bonding in materials with sub Angstrom resolution. Experiments utilize coherently and incoherently scattered electrons. In this contribution, the interface between gallium nitride and sapphire as well as thin silicon gate oxides are studied to understand underlying physical processes and the strength of the different microscopy techniques. An investigation of the GaN/sapphire interface benefits largely from the application of phase contrast microscopy that makes it possible to visualize dislocation core structures and single columns of oxygen and nitrogen at a closest spacing of 85 pm. In contrast, it is adequate to investigate Si/SiOxNy/poly-Si interfaces with incoherently scattered electrons and electron spectroscopy because amorphous and poly crystalline materials are involved. Here, it is demonstrated that the SiOxNy/poly-Si interface is rougher than the Si/SiOx interface, that desirable nitrogen diffusion gradients can be introduced into the gate oxide, and that a nitridation coupled with annealing increases its physical width while reducing the equivalent electrical oxide thickness to values approaching 1.2 nm. Therefore, an amorphous SiNxOy gate dielectric seems to be a suitable substitute for traditional gate oxides to further increase device speed by reducing dimensions in Si technology.

In as‐grown GaP we investigated two phosphorous antisite related electron spin resonance spectra with respect to ligand hyperfine (lhf) interactions: basing upon a linewidth analysis, we attribute a hyperfine doublet with unresolved lhf structure (PA3) in n‐GaP:S to a P‐antisite defect with cubic symmetry. The lhf interaction of the well known lhf resolved PP4 doublet (PA1) in semi‐insulating GaP:Cr was found to be temperature dependent. In plastically deformed GaP the two spectra showed contrary linewidth changes. In highly deformed n‐GaP:S both signals were detectable. A new spectrum of low symmetry (GA1) was observed which we attribute to a gallium antisite related defect.

The main purpose of the article by A.C. Diebold and coworkers (2003) is to propose a robust method for determination of gate oxide thickness. O'Keefe objects to a statement in this paper that “Lattice images do NOT depict the projected atom columns; instead, they are interference patterns of the directly transmitted beam with diffracted beams.”

We describe how general lattice images may be used to measure the variation of the potential in crystalline solids in any projection, with no knowledge of the imaging conditions. This approach is applicable to structurally perfect samples, in which interfacial topography, or changes in composition are of interest. We present the first atomic-level topographic map of a Si/SiO2 interface in plan-view, and the first microscopic compositional map of a Si/GeSi/Si quantum well in cross-section.

A quantitative analysis is given for three distinguishable effects contributing to inhomogeneities of the band gap of plastically deformed and doped silicon: (a) capture of charge from the chemical dopants to deep-point-like defects, (b) capture of the charge by deep defects in the core of dislocations, and (c) capture of the charge into states split off from the band edges that are caused by the presence of dislocations in the crystals. In crystals deformed below 700 \ifmmode^\circ\else\textdegree\fi{}C process (a) dominates and it gives rise to local shifts of the band edges. Contributions from processes (a) and (b) can be distinguished by only analyzing the ESR spectra of the deformed samples, but they do not lead to an accumulation of delocalized charge along the dislocations, which, in the case of process (c), is shown to be evident from the appearance of electric-dipole spin resonances. Observation of the last process requires special deformation procedures that produce straight dislocation segments.

Nowadays, High Resolution Electron Microscopes are capable to resolve structures on a scale below 100 pm. They can be equipped for Electron Holography in order to detect electric / magnetic fields and for chemical analyses (Electron Energy Loss Spectroscopy & Energy Dispersed X-ray's) that can be performed with a lateral resolution of 0.5 to 1 nm [1]. With the aid of computer sciences it became also possible to quantify local strain. We utilize Philips CM200 and CM300 field emission instruments with attached image filters, the JEOL Atomic Resolution Microscope and specialized software [2] to perform these tasks. On the other hand, a recent highlight in materials sciences is the development a GaN technology that is driven by a fast trial and error approach and aims to revolutionize lighting [3]. It was unavoidable that basic materials properties of the nano-structured thin films are barely understood because of the rapid progress [4].

It is reported that annealing and deformation experiments at 800°C and 650°C in a H 2 gas atmosphere change the plastic -, electrical - and paramagnetic properties of deformed silicon. Effusion experiments prove that hydrogen can be introduced by the used annealing/deformation procedures into the bulk of Si if dislocations are present. In particular, the annealing of Si at 800°C in a H 2 gas ambient leads to an activation of surface sources for the introduction of dislocations during the deformation and the stress exponent m for dislocation motion is reduced. In addition it is shown that the deformation induced ESR-defect spectrum can be reduced by exposure of the crystals to molecular hydrogen in a rather similar but more efficient way than it can be done by exposure of the crystals to an atomic hydrogen plasma at low temperature. From this similarity and from a H 2 induced reduction of the specific resistance we consider it most likely that passivation of deep defects was achieved by exposure of the crystals to the H 2 gas ambient but it cannot be excluded that the H 2 treatments just changed the defect formation rate. In order to observe such effects, a critical temperature T c in the range of 650°C c ≤ 800°C needs to be overcome and this indicates the presence of an activation barrier against incorporation of hydrogen into Si from the H 2 gas ambient.

Lattice images obtained by Transmission Electron Microscopy (TEM) are routinely used to infer the subsurface microstructure of crystalline materials. In principle, a lattice image is a map of the sample (Coulomb) potential, projected along a zone axis (see, e.g., [1]). In practice, it is difficult to extract quantitative information from lattice images. This stems from two primary reasons. First, electrons are multiply scattered during their passage through crystalline samples of realistic thickness (>10Å). This results in a complex, highly nonlinear relationship between the sample potential and the characteristics of the lattice image. This relationship changes rapidly with the sample thickness, and thus from point to point over the sample. Second, electromagnetic lenses have severe aberrations. The image details thus depend sensitively on the (contrast) transfer function of the microscope, and hence the lens defocus. It is not possible to establish a general relationship between the sample potential and the image features.

We made use of a periodic adiabatic technique to resolve distributions of spin-lattice relaxation times T1, investigating Si:P and deformed silicon. In the case of SO a continuous distribution of T1 values can be evaluated by this method. An even more interesting application is given by separating discrete, slightly different relaxation processes shown for the deformation-induced resonance SiY.

Hydrogen effusion spectra of oxidized Si samples are investigated and characteristic activation energies associated with the rupture of hydrogen bonds are estimated. Dislocation‐enhanced solubility of hydrogen is found in deformed crystals and macroscopic diffusion depth (mm) can be realized. Hydrogenation is performed by exposure of the crystals to H2 gas at 800 °C. Beside the desorption of hydrogen bound close to the Si surface, the rupture of hydrogen‐hydrogen bonds of molecules stored in deformed crystals is observed. The storage of the H2 molecules requires the presence of dislocations or deformation‐induced point defects, and the H:H binding energy ranges from 2.7 to 3.8 eV. Most likely this energy variation comes from different local strain around dislocations. Atomic hydrogen is found to dominate the effusion kinetics.

The velocity of 60°- and screw-dislocations in silicon is measured at 422°C (0.41Tm) in a wide range of stresses (50 MPa ≦ τ ≦ 300 MPa). The double etching technique is used taking into account the individual depth of bending of the 60° dislocations below the surface. Dislocations of four different glide systems are investigated exhibiting clearly different stress exponents m = = d ln v/d ln τ. This exponent appears to be composed of two parts; one part depends on the climb force acting on the dislocation and is interpreted as a consequence of the change of the Peierls potential by core deformation. The second part depends on temperature and dislocation type. Die Geschwindigkeit von 60°- und Schraubenversetzungen in Silizium wird bei 422°C (0,41Tm) in einem grosen Spannungsbereich (50 MPa ≦ τ ≦ 300 MPa) gemessen. Die Doppelatztechnik wird angewandt, wobei die Tiefe des Umknickens der 60°-Versetzungen unter der Oberflache in Rechnung gestellt wird. Versetzungen aus vier Gleitsystemen werden untersucht, wobei sich verschiedene Spannungsexponenten m = d ln v/d ln τ ergeben. Dieser Spannungsexponent ist offenbar aus zwei Anteilen zusammengesetzt: der eine hangt von der auf die Versetzung wirkenden Kletterkraft ab und berucksichtigt die Abhangigkeit des Peierlspotentials von der Verzerrung des Versetzungskerns. Der zweite Anteil hangt von der Temperatur und dem Versetzungstyp ab.

The velocity v of dislocation half-loops introduced into swirl-free floating-zone grown undoped silicon has been measured at 420°C in the resolved shear stress range 30 In the second part the papers review EPR spectroscopy of plastically deformed silicon and collects new results on the activity of dislocations in this material as trapping / recombination centers (decay of photo-EPR, photoluminescence, EBIC microscopy and photoplastic effect).

High‐pressure/low‐temperature plastic deformation of silicon leads to the appearance of new electron spin resonance active centers. One of them could be identified to be the pentavacancy Si‐P1, which also can be produced by irradiation. Depending on the deformation axis the defect occurs in several homologous orientations. Spin‐lattice relaxation time measurements performed on both plastically deformed and neutron‐irradiated silicon show the existence of an Orbach‐type relaxation with an activation energy of about 23 meV corresponding to the thermal excitation energy of the resonant electron into its high‐temperature state.