Atomic-Resolution Mapping of Localized Phonon Modes at Grain Boundaries.

Phonon scattering at grain boundaries (GBs) is significant in controlling the nanoscale device thermal conductivity. However, GBs could also act as waveguides for selected modes. To measure localized GB phonon modes, milli-electron volt (meV) energy resolution is needed with subnanometer spatial resolution. Using monochromated electron energy loss spectroscopy (EELS) in the scanning transmission electron microscope (STEM) we have mapped the 60 meV optic mode across GBs in silicon at atomic resolution and compared it to calculated phonon densities of states (DOS). The intensity is strongly reduced at GBs characterized by the presence of 5- and 7-fold rings where bond angles differ from the bulk. The excellent agreement between theory and experiment strongly supports the existence of localized phonon modes and thus of GBs acting as waveguides.

Cellular pathways of calcium transport and concentration toward mineral formation in sea urchin larvae.

Sea urchin larvae have an endoskeleton consisting of two calcitic spicules. The primary mesenchyme cells (PMCs) are the cells that are responsible for spicule formation. PMCs endocytose sea water from the larval internal body cavity into a network of vacuoles and vesicles, where calcium ions are concentrated until they precipitate in the form of amorphous calcium carbonate (ACC). The mineral is subsequently transferred to the syncytium, where the spicule forms. Using cryo-soft X-ray microscopy we imaged intracellular calcium-containing particles in the PMCs and acquired Ca-L2,3 X-ray absorption near-edge spectra of these Ca-rich particles. Using the prepeak/main peak (L2'/ L2) intensity ratio, which reflects the atomic order in the first Ca coordination shell, we determined the state of the calcium ions in each particle. The concentration of Ca in each of the particles was also determined by the integrated area in the main Ca absorption peak. We observed about 700 Ca-rich particles with order parameters, L2'/ L2, ranging from solution to hydrated and anhydrous ACC, and with concentrations ranging between 1 and 15 M. We conclude that in each cell the calcium ions exist in a continuum of states. This implies that most, but not all, water is expelled from the particles. This cellular process of calcium concentration may represent a widespread pathway in mineralizing organisms.

Toward Compositional Contrast by Cryo-STEM.

Electron microscopy (EM) is the most versatile tool for the study of matter at scales ranging from subatomic to visible. The high vacuum environment and the charged irradiation require careful stabilization of many specimens of interest. Biological samples are particularly sensitive due to their composition of light elements suspended in an aqueous medium. Early investigators developed techniques of embedding and staining with heavy metal salts for contrast enhancement. Indeed, the Nobel Prize in 1974 recognized Claude, de Duve, and Palade for establishment of the field of cell biology, largely due to their developments in separation and preservation of cellular components for electron microscopy. A decade later, cryogenic fixation was introduced. Vitrification of the water avoids the need for dehydration and provides an ideal matrix in which the organic macromolecules are suspended; the specimen represents a native state, suddenly frozen in time at temperatures below -150 °C. The low temperature maintains a low vapor pressure for the electron microscope, and the amorphous nature of the medium avoids diffraction contrast from crystalline ice. Such samples are extremely delicate, however, and cryo-EM imaging is a race for information in the face of ongoing damage by electron irradiation. Through this journey, cryo-EM enhanced the resolution scale from membranes to molecules and most recently to atoms. Cryo-EM pioneers, Dubochet, Frank, and Henderson, were awarded the Nobel Prize in 2017 for high resolution structure determination of biological macromolecules.A relatively untapped feature of cryo-EM is its preservation of composition. Nothing is added and nothing removed. Analytical spectroscopies based on electron energy loss or X-ray emission can be applied, but the very small interaction cross sections conflict with the weak exposures required to preserve sample integrity. To what extent can we interpret quantitatively the pixel intensities in images themselves? Conventional cryo-transmission electron microscopy (TEM) is limited in this respect, due to the strong dependence of the contrast transfer on defocus and the absence of contrast at low spatial frequencies.Inspiration comes largely from a different modality for cryo-tomography, using soft X-rays. Contrast depends on the difference in atomic absorption between carbon and oxygen in a region of the spectrum between their core level ionization energies, the so-called water window. Three dimensional (3D) reconstruction provides a map of the local X-ray absorption coefficient. The quantitative contrast enables the visualization of organic materials without stain and measurement of their concentration quantitatively. We asked, what aspects of the quantitative contrast might be transferred to cryo-electron microscopy?Compositional contrast is accessible in scanning transmission EM (STEM) via incoherent elastic scattering, which is sensitive to the atomic number Z. STEM can be regarded as a high energy, low angle diffraction measurement performed pixel by pixel with a weakly convergent beam. When coherent diffraction effects are absent, that is, in amorphous materials, a dark field signal measures quantitatively the flux scattered from the specimen integrated over the detector area. Learning to interpret these signals will open a new dimension in cryo-EM. This Account describes our efforts so far to introduce STEM for cryo-EM and tomography of biological specimens. We conclude with some thoughts on further developments.

Localized Phonon Densities of States at Grain Boundaries in Silicon.

Since it is now possible to record vibrational spectra at nanometer scales in the electron microscope, it is of interest to explore whether extended defects in crystals such as dislocations or grain boundaries will result in measurable changes of the phonon densities of states (dos) that are reflected in the spectra. Phonon densities of states were calculated for a set of high angle grain boundaries in silicon. The boundaries are modeled by supercells with up to 160 atoms, and the vibrational densities of states were calculated by taking the Fourier transform of the velocity­velocity autocorrelation function from molecular dynamics simulations with larger supercells doubled in all three directions. In selected cases, the results were checked on the original supercells by comparison with the densities of states obtained by diagonalizing the dynamical matrix calculated using density functional theory. Near the core of the grain boundary, the height of the optic phonon peak in the dos at 60 meV was suppressed relative to features due to acoustic phonons that are largely unchanged relative to their bulk values. This can be attributed to the variation in the strength of bonds in grain boundary core regions where there is a range of bond lengths.

Protein secondary structure signatures from energy loss spectra recorded in the electron microscope.

Infrared spectroscopy is a powerful technique for characterising protein structure. It is now possible to record energy losses corresponding to the infrared region in the electron microscope and to avoid damage by positioning the probe in the region adjacent to the structure being studied. Spectra from bacteriorhodopsin, a protein that is predominately a α helix, and OmpF porin, a protein that is mainly ß sheet show significant differences over a spectral range from ∼0.1 to 0.25 eV (∼1000 to 1800 cm-1 ). Although the energy resolution equivalent to 60 cm-1 is inferior to Fourier Transform InfraRed Spectroscopy (FTIR) the spectra are very sensitive to molecular orientation. Polar bonds aligned parallel to the specimen grid make particularly strong contributions to the energy loss spectra. Ultra-high-resolution energy loss spectroscopy in the electron microscope can potentially add useful information to imaging and diffraction for determining the secondary structure misfolding believed to be responsible for dementia diseases such as Alzheimer's.

Proteins are long linear molecular chains that when folded into complex three-dimensional shapes enable them to perform their biological functions. Infrared spectroscopy is a powerful technique for characterising protein folds, especially the proportions of helices and sheets that are significant building blocks in the overall structure. Traditionally, it was only possible to record infrared spectra from large amounts of material. In this paper, we show that it is possible to record the equivalent of the infrared spectrum from regions much smaller than a cell using a high-performance spectrometer coupled to electron microscopy. One great advantage is that the spectroscopic measurements can be combined with the standard high-resolution imaging and other characterisation techniques available in the electron microscope. We believe expansion of this method will impact diseases such as Alzheimer's, which are believed to be the results of an incorrect folding process. Our technique, where we combine infrared spectroscopic measurements with electron microscopy, could be invaluable in characterising the critical early stages of protein misfolding and/or assembly. This information will be invaluable in disease prognosis and the search for potential therapies.

Vibrational spectroscopy in the electron microscope.

Vibrational spectroscopies using infrared radiation, Raman scattering, neutrons, low-energy electrons and inelastic electron tunnelling are powerful techniques that can analyse bonding arrangements, identify chemical compounds and probe many other important properties of materials. The spatial resolution of these spectroscopies is typically one micrometre or more, although it can reach a few tens of nanometres or even a few ångströms when enhanced by the presence of a sharp metallic tip. If vibrational spectroscopy could be combined with the spatial resolution and flexibility of the transmission electron microscope, it would open up the study of vibrational modes in many different types of nanostructures. Unfortunately, the energy resolution of electron energy loss spectroscopy performed in the electron microscope has until now been too poor to allow such a combination. Recent developments that have improved the attainable energy resolution of electron energy loss spectroscopy in a scanning transmission electron microscope to around ten millielectronvolts now allow vibrational spectroscopy to be carried out in the electron microscope. Here we describe the innovations responsible for the progress, and present examples of applications in inorganic and organic materials, including the detection of hydrogen. We also demonstrate that the vibrational signal has both high- and low-spatial-resolution components, that the first component can be used to map vibrational features at nanometre-level resolution, and that the second component can be used for analysis carried out with the beam positioned just outside the sample--that is, for 'aloof' spectroscopy that largely avoids radiation damage.

Exploring the theoretical basis and limitations of cryo-STEM tomography for thick biological specimens.

Scanning transmission electron microscope (STEM) imaging has recently been applied to the cryo-tomography of thick biological specimens. As previously shown for plastic sections, STEM has a number of advantages for cryo-imaging compared to conventional wide-field TEM imaging. STEM is insensitive to phase coherence and is therefore suitable for much thicker specimens than TEM. Imaging in focus, with a long depth of field, also circumvents the complications of an oscillatory contrast transfer function and missing information at low spatial frequencies. Moreover the image signal represents a quantitative measurement of the electron scattering pixel by pixel, so that absolute intensities can be interpreted in terms of material properties in the specimen. Resolution, however, is undoubtedly compromised for thick samples, especially in the regime of multiple elastic scattering. In this work we address the specific issues that arise in cryo-tomography of thick biological specimens. We formulate an imaging model based on a Boltzmann transport equation, complemented by Monte Carlo simulations. Using these theoretical tools, we identify conditions for image acquisition that will be compatible with the basic presumption of tomographic reconstruction, i.e., that for a given composition the imaging signal varies monotonically with thickness. For optimal resolution, contrast, and signal strength, we propose to generalize the on-axis bright field detector to collect at angles well beyond the illumination cone. Our results justify the generation of 3D images for micron thicknesses and beyond.

Phosphorus detection in vitrified bacteria by cryo-STEM annular dark-field analysis.

Bacterial cells often contain dense granules. Among these, polyphosphate bodies (PPBs) store inorganic phosphate for a variety of essential functions. Identification of PPBs has until now been accomplished by analytical methods that required drying or chemically fixing the cells. These methods entail large electron doses that are incompatible with low-dose imaging of cryogenic specimens. We show here that Scanning Transmission Electron Microscopy (STEM) of fully hydrated, intact, vitrified bacteria provides a simple means for mapping of phosphorus-containing dense granules based on quantitative sensitivity of the electron scattering to atomic number. A coarse resolution of the scattering angles distinguishes phosphorus from the abundant lighter atoms: carbon, nitrogen and oxygen. The theoretical basis is similar to Z contrast of materials science. EDX provides a positive identification of phosphorus, but importantly, the method need not involve a more severe electron dose than that required for imaging. The approach should prove useful in general for mapping of heavy elements in cryopreserved specimens when the element identity is known from the biological context.

Calculation of the infrared spectra of proteins.

The CHARMM22 force field with associated partial charges is used to calculate the infrared spectra of a number of small proteins and some larger biothreat proteins. The calculated high-frequency region, from about 2,500 to 3,500 cm(-1), is dominated by stretching modes of hydrogen bonded to other atoms, and is very similar in all proteins. There is a peak at 3,430 cm(-1) whose intensity is predicted by these calculations to be a direct measure of arginine content. The calculated low-frequency THz region, up to 300 cm(-1), is also very similar in all the proteins and just reflects the vibrational density of states in agreement with experimental results. Calculations show that the intermediate-frequency region between 500 and 1,200 cm(-1) shows the greatest difference between individual proteins and is also the least affected by water absorption. However, to match experimental measurements in the amide region, it was necessary to reduce the hydrogen partial charges.

Is localized infrared spectroscopy now possible in the electron microscope?

The recently developed in-column monochromators make it possible to record energy-c spectra with resolutions better than 30 meV from nanometer-sized regions. It should therefore in principle be possible to detect localized vibrational excitations. The scattering geometry in the electron microscope means that bond stretching in the specimen plane or longitudinal optic phonons dominate the scattering. Most promising for initial studies are vibrations with energies between 300 and 400 meV from hydrogen bonded to other atoms. Estimates of the scattering cross-sections on the basis of a simple model show that they are about the same as inner shell scattering cross-sections. Cross-sections also increase with charge transfer between the atoms, and theory incorporating realistic charge distributions shows that signal/noise is the only limitation to high-resolution imaging. Given the magnitude of the scattering cross-sections, minimizing the tail of the zero-loss peak is just as important as achieving a small-width at half-maximum. Improvements in both resolution and controlling the zero-loss tail will be necessary before it is practical to detect optic phonons in solids between 40 and 60 meV.

Highly automated electron energy-loss spectroscopy elemental quantification.

A model-based fitting algorithm for electron energy-loss spectroscopy spectra is introduced, along with an intuitive user-interface. As with Verbeeck & Van Aert, the measured spectrum, rather than the single scattering distribution, is fit over a wide range. An approximation is developed that allows for accurate modeling while maintaining linearity in the parameters that represent elemental composition. Also, a method is given for generating a model for the low-loss background that incorporates plural scattering. Operation of the user-interface is described to demonstrate the ease of use that allows even nonexpert users to quickly obtain elemental analysis results.

Does the order of elastic and inelastic scattering affect an image or is there a top bottom effect from inelastic scattering?

Especially for light elements inelastic scattering is more probable than the elastic scattering that conveys the structural information. The question arises as to whether an image using inelastically scattered electrons is different depending on whether the elastic or inelastic scattering happens first, is there a top-bottom effect. We show that since inelastic scattering is concentrated in a narrow range of angles, much less than typical Bragg angles in light element materials, the inelastic and elastic processes are separable and, to a very good approximation, there is no top-bottom effect. For weakly scattering thin biological specimens that are phase objects the separation is exact and there can be no top-bottom effect.

Evaluation of the role of Au in improving catalytic activity of Ni nanoparticles for the formation of one-dimensional carbon nanostructures.

In situ dynamic imaging, using an environmental transmission electron microscope, was employed to evaluate the catalytic activity of Au/SiO(2), Ni/SiO(2), and Au-Ni/SiO(2) nanoparticles for the formation of one-dimensional (1-D) carbon nanostructures such as carbon nanofibers (CNFs) and nanotubes (CNTs). While pure-Au thin-film samples were inactive for carbon deposition at 520 °C in 0.4 Pa of C(2)H(2), multiwalled CNTs formed from Ni thin films samples under these conditions. The number of nanoparticles active for CNF and CNT formation increased for thin films containing 0.1 mol fraction and 0.2 mol fraction of Au but decreased as the overall Au content in thin films was increased above 0.5 mol fraction. Multiwalled CNTs formed with a root growth mechanism for pure Ni samples, while with the addition of 0.1 mol fraction or 0.2 mol fraction of Au, CNFs were formed via a tip growth mechanism at 520 °C. Single-walled CNTs formed at temperatures above 600 °C in samples doped with less than 0.2 mol fraction of Au. Ex situ analysis via high-resolution scanning transmission electron microscopy (STEM) and energy-dispersive X-ray spectroscopy (EDS) revealed that catalytically active particles exhibit a heterogeneous distribution of Au and Ni, where only a small fraction of the overall Au content was found in the portion of each particle actively involved in the nucleation of graphitic layers. Instead, the majority of the Au was found to be segregated to an inactive capping structure at one the end of the particles. Using density-functional theory calculations, we show that the activation energy for bulk diffusion of carbon in Ni reduces from ≈1.62 eV for pure Ni to 0.07 eV with the addition of small amounts (≈0.06 mol fraction) of Au. This suggests that the enhancement of C diffusion through the bulk of the particles may be responsible for improving the number of particles active for nucleating the 1-D carbon nanostructures and thereby the yield.

Calculations of the Evolution of the Ca L23 Fine Structure in Amorphous Calcium Carbonate.

Amorphous calcium carbonate (ACC) has been found in many different organisms. Biogenic ACC is frequently a precursor in the formation of calcite and aragonite. The process of structural transformation is therefore of great interest in the study of crystallization pathways in biomineralization. Changes in the prepeak/main peak (L2'/L2) intensity ratio of the Ca L23-edge X-ray absorption spectroscopy (XAS) of Ca-rich particles in skeleton-building cells of sea urchin larva revealed that ACC precipitates through a continuum of states rather than through abrupt phase transitions involving two distinct phases as formerly believed. Using an atomic multiplet code, we show that only a tetragonal or "umbrella-like" distortion of the Ca coordination polyhedron can give rise to the observed continuum of states. We also show on the basis of the structures obtained from previous molecular dynamics simulations of hydrated nanoparticles that the Ca L23-edge is not sensitive to atomic arrangements in the early stages of the transformation process.

Coherent and incoherent imaging of biological specimens with electrons and X-rays.

Since radiation damage is proportional to fluence, radiation damage limits the spatial resolution of biological structures determined by either X-ray or electron scattering. If only elastic scattering is used for structural information then electrons are superior as the ratio of elastic to inelastic scattering is higher than for X-rays. For soft X-rays in the water window below the O K edge photoabsorption contrast might be better than elastic scattering for distinguishing different biological materials. Phase contrast elastic scattering is most effective in the hard X-ray region up to about 10 keV. Radiation damage limits spatial resolution for most X-ray imaging to 10-20 nm. Local molar concentrations of Na,K and Ca ions can be determined at somewhat lower spatial resolutions using relevant absorption edges. At higher energies resolution res is only limited by the fluence available from the light source, since energy deposition is small.

Lattice resolution of vibrational modes in the electron microscope.

The combination of aberration correction and ultra high energy resolution with monochromators has made it possible to record images showing lattice resolution in phonon modes, both with a displaced collection aperture and more recently with an on -axis collection aperture. In practice the objective aperture has to include Bragg reflections that correspond to the observed lattice image spacings, and the specimen has to be sufficiently thick for adequate phonon scattered intensity. There has been controversy as to whether the images with the on axis detector are really a consequence of lattice resolution in a phonon mode or just a transfer of information from an image that was formed by elastically scattered electrons. We present results of calculations based on a theory that includes the possibility of dynamical electron diffraction for both incident and scattered electrons and the full phonon dispersion relation. We show that Umklapp scattering from the second Brillouin Zone back to the first Brillouin Zone is necessary for lattice resolution with the on axis detector and that it is therefore reasonable to attribute the lattice resolution to the phonon scattering.

The influence of surfaces and interfaces on high spatial resolution vibrational EELS from SiO2.

High-resolution monochromated electron energy-loss spectroscopy has the potential to map vibrational modes at nanometer resolution. Using the SiO2/Si interface as a test case, we observe an initial drop in the SiO2 vibrational signal when the electron probe is 200 nm from the Si due to long-range nature of the Coulomb interaction. However, the distance from the interface at which the SiO2 integrated signal intensity drops to half its maximum value is 5 nm. We show that nanometer resolution is possible when selecting the SiO2/Si interface signal which is at a different energy position than the bulk signal. Calculations also show that, at 60 kV, the signal in the SiO2 can be treated non-relativistically (no retardation) while the signal in the Si, not surprisingly, is dominated by relativistic effects. For typical transmission electron microscope specimen thicknesses, surface coupling effects must also be considered.