The preparation of frozen amorphous thin films of solutions containing crystals, and imaging with cryo energy filtered TEM, opens for the first time the possibility to image the chemistry of solids interacting with liquids. The future possible areas of research include: dynamic of crystal growth processes, adsorption mechanisms, ion exchange, structure determination of solids in equilibrium with solutions, etc.

The sample preparation uses a controlled-environment vitrification system1. Vitrification of the liquid phase was achieved by plunging the grid into liquid ethane. The vitrified specimen was transferred into the microscope by an Oxford CT3500 cryo-holder. The specimen temperature in the microscope column can be kept at -−183 °C. The thickness of the vitrified films, including the crystals, should be about 250 nm or preferably less. The Philips CM120 BioTWIN Cryo, here used for cryo-TEM, has a focal length of 6 mm and a structural resolution of 0.4 nm.

The energy resolution which can be attained in electron energy loss spectroscopy (EELS) is determined by the energy spread of the electron source. The energy width of a high brightness electron gun (typically 0.4 to 0.8 eV) blurs the energy spectrum. A pre-specimen energy filter or monochromator must be used to reduce the energy width of the beam below 0.1 eV to allow detailed EELS analysis of the electronic band structures in materials. The monochromator can not only improve EELS, but it is also capable of improving the spatial resolution in low voltage SEM, which is limited by the chromatic blur of the objective lens. A new type of monochromator the Fringe Field Monochromator has been designed and experiments in an ultra high vacuum setup show the monochromatisation of a Schottky Field Emission Gun.

It would be very desirable to obtain EELS at 0. l-0.2eV resolution for bandstructure studies at interfaces and defects. [1,2] It has been possible to build spectrometers that perform in this energy range, but it has been difficult to utilize this resolution, because the energy width of the cold field emission source is about 0.3-0.4 eV for the currents needed to do practical microscopy. [3] Numerical methods have been used to sharpen the spectra, showing unambiguously that EELS spectra would benefit greatly if a higher resolution source were available. [4,5]

This situation has led one of us (H.W. Mook) to investigate electron monochromator optical designs which would be compatible with the high brightness and current required by the STEM. [6] This work produced a 4cm long monochromator that can be integrated into the field emission source. As shown in Fig. 1, the monochromator optics includes a field lens formed by deceleration of the electron beam from the 3 KeV extraction energy to about 350 eV inside the monochromator, a short Wien Filter velocity dispersive field, 150nm wide Si energy selecting slits, and an acceleration field lens at the monochromator exit.

Diamond-like carbon (DLC) has attracted considerable interest with a variety of potential applications [1]. A new class of amorphous films has been produced recently by incorporating metals within the carbon network. The electrical conductivity of these films varies a wide range, however nonlinear with the metal concentration. It is believed it is because metals may form clusters. Such nano-phase metal cluster encapsulated by an inert matrix can offer interesting applications, for example as nano-electrodes used in electrochemistry[2]. It is therefore fundamentally important to observe the metal cluster formation. By using Z-contrast imaging [3], the metal distribution can be directly determined on the atomic level. The local metal concentration and other electronic properties cafi also be analyzed by spatial resolved electron energy loss spectroscopy (EELS)

Good quality polycrystalline diamond is transparent and has an optical band-gap very close to that of single crystal diamond, 5.5 eV. Tetrahedral amorphous carbon (ta-C) is a form of amorphous carbon which can have up to 90 % sp3 type bonds. It is the form of amorphous carbon, which is closest to having a true diamond-like structure. The optical band-gap of ta-C is however very different from that of polycrystalline diamond, typically in the range 2-3 eV. This difference in the optical bandgap has been ascribed to the electronic states formed by the residual graphitic bonding. The π − π* states at the graphitic sites effectively form tail states on the valance and conduction band edges created by the diamond-like bonding. These tail states extend deep into the band-gap, which would exist if there were only (100 %) diamond-like bonds. The poor electronic properties obtained thus far for ta-C are now accepted as being associated with localised conduction in the π − π* tail states.

Extinction distances for two beam conditions are readily available through many sources, but under more complex zone axis conditions extinction distances are not well known. The use of thickness fringes as an approximate thickness estimate under zone axis conditions, such as in high resolution Annular Dark Field imaging, can be a useful tool. We have measured thickness vs. fringe number under zone axis conditions using low loss Electron Energy Loss Spectroscopy (EELS). Work was done on the Cornell UHV HB501 STEM.

Several orientations of silicon were glued together and thinned in cross section by the wedge technique. Additional thinning was done by low angle ion milling. Surface oxide was removed with an HF etch before loading into the microscope.Each piece of silicon was tilted to the zone axis condition and low loss EELS spectra was taken at selected points. The thickness fringes were numbered with the following convention: the center of the first black fringes close to the edge is 1 and the center of the second black fringe is 2, etc.

The microstructure and mechanical properties of alloy systems are known to be influenced by the size and distribution of sub-micron dispersoids through boundary pinning and Orowan strengthening. Explaining and accurately predicting the behaviour of these systems relies in part on the measurement of the volume fraction of the second phase precipitates. The transmission electron microscope (TEM) provides the necessary resolving power to measure the dispersoids of interest, and when coupled to Electron Energy Loss Spectroscopy (EELS) based filtering systems, the thickness of the observed volumes can be determined using the log-ratio technique.

A series of nominally AA5754 alloys with increasing Manganese content (0%, 0.2%, 0.5%, 0.8%) was used to evaluate the method. The data was collected using a Philips FEGSTEM equipped with a Gatan Imaging Filter (GIF). A series of images were collected from each of a number of contiguous fields in each specimen. These included an unfiltered image, an image collected from the zero-loss region of the transmitted spectrum, and a filtered image generating sufficient precipitate contrast to allow subsequent spatial measurement. Using a filtered image, the contribution of diffuse scattered electrons is minimized, thus enabling the analysis of thicker regions.

The first MSA Special Symposium honors Professor Archie Howie CBE FRS for his pioneering contributions to the physics and applications of electron microscopy, initially in TEM and more recently also for SEM; and his leadership, outstanding teaching and wide ranging collaborations in these and related fields.

Professor Howie’s important contributions have included fundamental studies of diffraction contrast effects in the TEM which have been extensively leveraged in numerous applications and collaborations, by Howie, his coworkers and many others. They have had a major impact in column approximations, an early discovery of the weak beam effect in many beam computer simulations of coherency strain contrast and subsequent advances in dislocation weak beam imaging, and the first identification of 311 planar defects in irradiated Ge. Inelastic scattering and particularly valence losses, electron channeling, optical potential and x-ray production, amorphous materials were all major contributions, as were the series of projects centered on novel STEM methods.

Introduction

There is considerable interest in CVD diamond layers for a wide variety of applications. Growth of high quality and phase pure material is now achieved in a number of laboratories but it remains to optimize the growth process for particular applications so that physical properties approaching or even exceeding those of the best natural diamond can be achieved. We have found that the combined use of a wide range of different microscopies is effective in the process of refinement.

Diamond quality and phase purity

Raman spectroscopy is widely used to investigate the phase purity of diamond layers. Infra-red Raman is particularly sensitive to the presence of traces of non-diamond carbon in diamond while ultra-violet Raman reveals the presence of traces of diamond carbon in layers that largely are not sp bonded. For thick rather perfect layers confocal micro-Raman mapping is a convenient method for revealing the locations of non-diamond regions.

The work function ϕ, the bandgap Eg, the threshold energy level Et, for the inelastic scattering of excited electrons and the threshold energy transfer Ed for the onset of structural ionisation damage are clearly of major significance in various actively developing forms of hot carrier imaging. Two exciting examples here are the ability to image small and dynamic local changes in work function by PEEM and as well as mapping the variations in electronic structure between p and n-type regions of a semiconductor by SE imaging in the SEM. More recently still there have been indications that the SE signal in the ESEM might be sensitive to local changes in bandgap and suggestions that it might be even possible to image spatial variations in pH. It is increasingly clear that if these attractive opportunities are to be efficiently explored and developed, systematic and preferably quantitative observations are needed. Such work requires specimens whose atomic and electronic structure is either fully known beforehand or can be deduced from other signals available in a situation where the physical processes in the microscope are well understood.

The performance of many materials systems depends on our ability to control the distribution of atoms on a nanometre or sub-nanometre scale within those systems. This is as true for steels as it is for semiconductors. A key requirement for improving their performance is the ability to determine the distribution of the elements resulting from processing the material under a given set of conditions. Analytical electron microscopy (AEM) provides a range of powerful techniques with which to investigate this distribution. By combining information from different techniques, many of the ambiguities of interpretation of the data from an individual technique can be eliminated. The electron energy loss near edge structure (ELNES) present on an ionisation edge in the electron energy loss spectrum reflects the local structural and chemical environments in which the particular atomic species occurs. Thus it is a useful contribution to the information available. Since a similar local environment frequently results in a similar shape, ELNES is useful as a “fingerprint”.

The electronic structure of ceramics can be extracted quantitatively from the valence electron energy-loss spectroscopy (VEELS) of transitions between the valence and conduction bands. We obtained VEEL spectra of several ceramics (FIG. 1) with a VG HB501 dedicated STEM equipped with Gatan PEELS. Improved data acquisition and new methods of data analysis allowed us to treat the data fully quantitatively. The reliable and accurate removal of the zero loss peak was crucial because intensities at energy losses just above the band gap of the ceramic material have a large influence on the results. An asymmetric Pearson VII function was fitted into the zero loss peak up to an energy loss for which no transitions are expected (an energy smaller than the band gap of the ceramic) and then extrapolated to higher energies. This limits the analysis to non-metallic materials, exhibiting non-zero band gap energies. We are currently developing methods to perform the analysis on metallic materials, using ellipsometric data in the visible and extrapolate the energy-loss function to 0 eV and thereby remove the need for the no transition energy.

A series of recent experiments, using electron as well as photon probes [1-3], have attracted considerable interest on the importance of surface plasmon modes in nanostructured media such as nanospheres, nanotubes, nanowires, nanopores or nanoholes.

The basic principle for STEM operation had early been recognized to be well suited to the investigation of cases where non penetrating electrons propagate at a given impact parameter from an external surface (along "aloof7 trajectories, using a terminology proposed by Warmack et al. [4]). One of the most spectacular effect is that signals attributed to the excitation of-surface plasmon modes could be detected in vacuum at distances as large as a few tens of nanometers from the outside surface of a specimen, see for instance [5] for the planar geometry and [6], for the spherical one. Most interpretations accounting for the spatial dependence of EELS spectra in the "aloof geometries, rely on models using the bulk dielectric coefficients of the material to describe the induced charges and polarization responsible for the specimen electric field acting on the probe electron and consequently for the measured energy loss.

Even though two generations of electron microscopists have come to accept that the resolution of their instruments is limited by spherical aberration, three different aberration correctors showing that the aberration can be overcome have recently been built [1-3]. One of these correctors was developed by us specifically for forming small electron probes in a dedicated scanning transmission electron microscope (STEM) [3, 4]. It promises to revolutionize the way STEM instruments are built and the types of problems that they are applied to.

As was the case with the Berlin Wall, when a barrier that was once thought immovable finally crumbles, many of the consequences can be quite unexpected. For STEM, the removal of spherical aberration (Cs) as the main resolution limit is likely to lead to a new paradigm in which:

1) The resolution at a given operating voltage will improve by about 3x relative to today's best. When Cs can be adjusted arbitrarily in a STEM being used for microanalysis or dark field imaging, defocus and Cs are set to values that optimally oppose the effect of the 5th-order spherical aberration C5.

In all scattering experiments some measure is need of the strength of the scattering interaction. The scattering cross section, which has dimensions of area, is a quantity that can be defined for any scattering interactions, irrespective of the nature of the scatterer, or the particle or radiation being scattered. To define a scattering cross section, refer to the geometry of Figure 1. If I0 is the incident number of particles, Is the number of particles scattered through an angle θ with an energy loss ΔE, N is the number of scatterers/ unit volume and t is the thickness of the specimen (or length of the scattering region) then where σ(θ ,ΔE) is the scattering cross section. The product N t represents the number of scatterers per unit area as seen by the incident beam

In electron microscopy all scattering arises from the Coulomb interaction between the incident electron and the electrons or nuclei or the atoms in the specimen.

In the past quarter century low voltage SEM (LVSEM) has been developed for use in a wide range of applications, and it has become the main area of innovation in SEM, and now perhaps more widely in electron microscopy. The intellectual challenge (1,2) has been transformed into a vital tool of modern technology in the semiconductor, polymer and chemical industries, and is widely used in the materials and biological sciences (3, 4). It has been claimed by this author and others that LVSEM is the rational and preferred state of SEM, but unfortunately at the time of the first SEMs only higher voltage systems could be built with the technologies available half a century ago (1). The primary advantage of the SEM remains the use of real bulk specimens with a minimum of potentially invasive preparation. A wide area in the range of square mms to sq inches can contain aperiodic, or otherwise infrequent, events in or on the surface, or in a simple snapped cross-section, and always using relatively stable bulk samples.

Single shell carbon nanotubes are members of the Fullerene family of carbon molecules. Typically, single shell carbon nanotubes measure about 0.7 — 3 nm in diameter and are usually several microns in length. Structurally, they can be thought of as narrow graphite sheets that have been bent around the long axis and joined at opposite edges to form long seamless hollow shells of carbon. Typically, a hemispheroidal cap that contains exactly six 5-rings terminates each end, as shown in Figure 1.

Graphite is known to have an in-plane elastic modulus of ∼1 TPa, one of the highest values known. Consequently, it is expected that single shell nanotubes should be very stiff — a fact that makes them potentially useful in fiber reinforced composites. However, because of their small size, it is impractical to measure their stiffness directly by conventional mechanical means. Recently, we demonstrated that thermal vibrations in freestanding multiwalled nanotubes could be used to estimate their stiffness [1].

Electron Microscopy of surfaces has always been a topic for a few stubborn souls. In addition to the standard headaches of electron microscopes (maintenance, breakdown) one has to handle the additional complications of maintaining a clean enough vacuum to make the results mean anything. (Without vacuum control the structures observed may be due to the contaminants, and hence meaningless.) To complicate things further, in addition to conventional ex-situ sample preparation one has to clean and equilibriate the surface while still retaining thin enough regions for useful study. The final step is to work out which techniques provide surface information and learn how to exploit them. It is appropriate to look at some of the key elements that are needed.

Sample Handling: The key to success with surfaces is not really the microscopy, but instead the sample treatment. The first crucial element is control over the sputter cleaning process.(While simple heating would be less damaging, if one wants to look at something other than silicon sputter cleaning is a must.)

Widespread application of metal oxides in catalysis, gas sensing, and as substrates for thin film growth has stimulated a strong interest in the atomic and electronic surface structure of these materials. The electronic structure of many metal oxide surfaces is characterised, and complicated from the theoretical modelling point of view, by the presence of strong on-site Coulomb repulsion (strong correlations) between valence electrons localised on the metal ions. Additionally, experimental studies have to deal with the difficulties associated with the electrically insulating nature of many of these oxides. We have overcome these problems through the development of novel experimental and theoretical techniques capable of providing structural and electronic information about strongly correlated insulating metal oxide surfaces.

The surfaces of NiO and CoO were investigated through elevated temperature scanning tunnelling microscopy (STM) thereby overcoming problems of low electrical conductivity of the samples.We show atomically resolved elevated temperature STM images of (001) cobalt and nickel monoxide surfaces obtained under similar conditions which show an order of magnitude difference in the atomic corrugation heights.

When I (JMG) began my Ph.D. studies under the supervision of Archie Howie in 1975, he was still licking his wounds over one of the very few apparently incorrect publications[l] of his career. I am proud that our recent work has shown that, far from being wrong, Archie and Lee Rudee were actually 25 years before their time. It was in the early 1970’s that electron microscopy instrumentation reached the performance level that was adequate to resolve the interplanar spacings in elemental materials, such as the semiconductors silicon and germanium. It immediately became of interest to examine the structure of amorphous silicon and germanium, in the expectation that direct imaging could resolve the old controversy between micro-crystallite and random network structural models. Porai-Koschits was the first to suggest that amorphous materials comprised extremely small crystalline domains (microcrystals), in contrast Zachariasen[2] proposed that a continuous random network was a better description of glassy materials.