Search

Scanning Electron Microscopy (SEM) has been widely used to characterize semiconductor materials. Approximately one half of all SEMs are used by the electronic industry. The SEM can be used as a local current source that can be scanned over semiconductor device surfaces. This imaging mode is called Electron Beam Induced Current (EBIC).A depletion zone formed by a Schottky diode is visualized using this technique. This Shottky diode is fabricated by placing a metal contact on a semiconductor surfaces. The work function of the metal and the semiconductor produce an equilibrium contact potential. This potential barrier occurs across a region free of charge carriers and thus an electric field is present. This region is called a depletion zone.

In order to perform quantitative X-ray microanalysis many parameters, representing the various stages of Xray generation and transport through the specimen at the chosen experimental conditions, must be known for all of the elements that might be encountered. Although ideally quantification is done by reference to standards so that only the functional variation of these parameters is required, in current practice it is increasingly necessary to work in situations where standardization is impossible and consequently where absolute magnitudes must be known. The quality and quantity of data that is now available varies widely.Ionization cross sectionsAlthough the amount of experimental data is limited, particularly in the energy range between 1 and 20keV, a critical assessment has concluded that for K-shell excitations both the magnitude of the cross-section σ and its variation with energy are adequately well known for overvoltages greater than about 2. For L- and M-shell excitations, however, and in all cases when the operational overvoltage is less than 2, the situation is much less satisfactory.

Although the benefits of operating the scanning electron microscope at low beam energies have been evident since the earliest days of the instrument, the successful utilization of the SEM under these conditions has required the development of high brightness field emission electron source, advanced lenses, and clean vacuums. As these technologies became available the level at which imaging became regarded as “low energy” has fallen from 10keV, first to 5keV, and more recently to 1keV. At this energy state of the art instruments can now provide an excellent balance between resolution – which becomes worse with decreasing energy – and desirable goals such as the minimization of sample charging and the reduction of macroscopic radiation damage – which tend to become more challenging as the energy is increased.An interesting new opportunity is to perform imaging in the ultra-low energy region between leV and 500eV. Over this energy range significant changes in the details of electron-solid interactions take place offering the chance of novel contrast modes, and the rapid fall in the electron beam range leads to the condition where the penetration of the incident beam into the sample is effectively limited to 1 or 2 nanometers. The practical problem is that of achieving useful levels of resolution and acceptable signal to noise ratios in the image. At energies below IkeV chromatic aberration dominates the probe formation in conventional instruments even when using an FEG source. However, the use of optimized retarding field optics essentially maintains chromatic aberration independent of landing energy down to very low values. Figure (1) shows an example of the performance that can be achieved on a commercial instrument – an Hitachi S-4500 – modified to operate in this mode, in this case at 50eV landing energy. The resolution of the image is judged from edge sharpness and detail to be significantly better than 0.1µm and, from experimental observation, this performance is apparently limited by residual astigmatism caused by uncorrected sample charging rather than by fundamental aberrations in the probe forming optics. Comparable, if somewhat lower resolution, images have been achieved on this, and other FEG SEMs, at energies as low as leV.

The passage of a beam of electrons through the low pressure gas atmosphere in an environmental SEM results in a scattering - and hence a broadening - of the probe, a reduction in the current available for imaging and analysis, and in the production of secondary ions which can also impact the specimen and modify the production of secondary and backscattered electrons. In order to optimize and fully exploit the capabilities of low vacuum SEMs it is desirable to be able to accurately predict these effects. The first step in achieving this aim is to be able to quantitatively predict the behavior of the incident beam.At a gas pressure of 100Pa. the mean free path for 20keV electrons is of the order of 1cm. For a typical working distance of 1cm., therefore, the average electron is only scattered once.

High resolution scanning electron microscopy (SEM) and energy dispersive x-ray analysis (EDXA) were performed to investigate microparticles in blended cements and their hydration products containing sodium-rich chemical wastes. The physical appearance of powder particles and the morphological development at different hydration stages were characterized by using high resolution SEM Hitachi S-900 and by SEM S-800 with a EDX spectrometer. Microparticles were dispersed on the sample holder and glued by 1% palomino solution. Hydrated bulk samples were dehydrated by acetone and mounted on the holder by silver paste. Both fracture surfaces and flat cutting sections of hydrating samples were prepared and examined. Some specimens were coated with an 3 nm thick Au-Pd or Cr layer to provide good conducting surfaces. For high resolution SEM S-900 observations the accelerating voltage of electrons was 1-2 KeV to protect the electron charging. Microchemical analyses were carried out by S800/EDS equipped with a LINK detector of take-off angle =40°.

Electron holography gives to the electron microscope the ability the visualize, and measure, electric and magnetic fields at high spatial resolution . The principle of the method is shown in figure (1). Some fraction of a coherent electron beam, typically produced from a cold field emission electron gun, is incident on the specimen while the rest passes through vacuum, or through some other region of the sample, to form a reference wave. An electrostatic bi-prism converges these two components to form an interferogram or ‘hologram’ at the viewing screen. The hologram consists of a fringe pattern, with a spacing determined by the potential on the bi-prism, modulated by the intensity variations of the transmitted bright-field signal. Variations in the phase of the transmitted signal are stored as shifts in the fringe pattern. The hologram can then be reconstructed, either optically or by using a computer and appropriate software, to yield two images, one - the 'Amplitude Image' - containing normal intensity contrast (bright field) information, and a second- the 'Phase Image'- which displays variations in the relative phase of the signal as changes in grey level. The phase of the transmitted beam will be shifted by an amount proportional to the mean inner potential of the specimen multiplied by the thickness, and also by the potential that the electron encounters as it travels through the sample.

Techniques to remove plural scattering from electron energy loss spectra (EELS) are important in bot hmicroanalysis and other quantitative applications of electron spectroscopy. The techniques used are either based on convolution, or Fourier transform deconvolution, methods, in which either the elastic scattering angular correction or both elastic and inelastic angular corrections are not included. In this work we propose a new method based on both angular and energy loss three-dimension Poisson statistics which includes elastic and inelastic mixed angular scattering correction in order to obtain more accurate quantitative analysis for EELS.The electron scattering distribution determined by angular and energy loss three-dimension Poissonstatistics is given by:where ITis the total incident electron intensity; t is the sample thickness; λi, λeand λTare inelastic , elastic and total scattering mean free paths; Si(θ) and Se(θ) are normalized single inelastic and elastic angular scattering distributions respectively, F(E) is the single scattering normalized energy loss distribution.

In analytical electron microscopy, it is often important to know the local thickness of a sample. The conventional method used for measuring specimen thickness by EELS is:where t is the specimen thickness, λiis the total inelastic mean free path, ITis the total intensity in an EEL spectrum, and I0is the zero loss peak intensity. This is rigorouslycorrect only if the electrons are collected over all scattering angles and all energy losses. However, in most experiments only a fraction of the scattered electrons are collected due to a limited collection semi-angle. To overcome this problem we present a method based on three-dimension Poisson statistics, which takes into account both the inelastic and elastic mixed angular correction.The three-dimension Poisson formula is given by:where I is the unscattered electron intensity; t is the sample thickness; λiand λeare the inelastic and elastic scattering mean free paths; Si(θ) and Se(θ) are normalized single inelastic and elastic angular scattering distributions respectively ; F(E) is the single scattering normalized energy loss distribution; D(E,θ) is the plural scattering distribution

In Physics, Chemistry, Materials Science, Biology and Medicine, it is very important to have accurate information about the stopping power of various media for electrons, that is the average energy loss per unit pathlength due to inelastic Coulomb collisions with atomic electrons of the specimen along their trajectories. Techniques such as photoemission spectroscopy, Auger electron spectroscopy, and electron energy loss spectroscopy have been used in the measurements of electron-solid interaction. In this paper we present a comprehensive technique which combines experimental and theoretical work to determine the electron stopping power for various materials by electron energy loss spectroscopy (EELS ). As an example, we measured stopping power for Si, C, and their compound SiC. The method, results and discussion are described briefly as below.The stopping power calculation is based on the modified Bethe formula at low energy:where Neffand Ieffare the effective values of the mean ionization potential, and the number of electrons participating in the process respectively. Neffand Ieffcan be obtained from the sum rule relations as we discussed before3 using the energy loss function Im(−1/ε).

The characterization of polymer blend morphology by TEM often requires objectionable heavy metal staining to enhance contrast, enabling phase differentiation of materials with otherwise similar electron scattering properties. In some cases, the presence of a higher atomic number species (O, N, Cl, F, etc.) in the molecular framework of one phase provides sufficient contrast without staining. In the present example the only inherent contrast mechanism for a blend of polystyrene/polyethylene/ethylene-styrene copolymer (PS/PE/E-S) should be diffraction contrast due to the crystalline PE. Electron holography was investigated as an additional method of characterizing polymer blends.Electron holography was first described by Gabor in 1948 but did not become a practical tool until recently, due to the availability of field emission gun (FEG) TEMs equipped with an electron bi-prism. Electron holography is a wave front reconstruction process that involves hologram formation and image reconstruction. An interference pattern (hologram) between a reference electron beam (passing through vacuum) and a beam passing through the specimen contains both the amplitude and phase of the transmitted electron wave.

The effective use of a CCD camera to record TEM images requires the use of a scintillator to convert the energy of the incident electrons into light. In order to record the image with the highest resolution possible from the given CCD array we require that the broadening of the electron beam that occurs as it travels through the scintillator be less than, or comparable to, the diameter of the pixel. Beam spreading that is greater than the pixel size not only reduces the line-pair resolution, but also causes correlation between adjacent pixels which may result in the measured detector quantum efficiency (DQE) apparently exceeding unity. At the same time we need the electron to transfer the maximum possible amount of energy to the scintillator since the intensity of the signal produced by the CCD array depends directly on the energy deposited. Clearly these two requirements are mutually opposed since the conditions which minimize the beam spreading will also likely be the conditions under which the least energy is transferred.

Electron energy loss spectroscopy (EELS) gives an inportant insight into the variety of excitations a sample may undergo when irradiated by an electron beam. The focus of this work was to simulate electronic excitations within the energy range from a few to several hundred eV. Our recently developed modulation scheme, combines both convolution and deconvolution techniques, to provide quantitative information about elementary inelastic scattering processes without knowledge of sample parameters such as thickness or optical constants.In the low energy loss region of the spectrum the primary excitation mechanisms include interband transitions, and surface and bulk plasmons. In general these individual excitation events overlap in the spectrum. A FFT convolution procedure was developed where the basic inelastic processes may be represented by the dielectric theory . The dielectric function ε is used to describe both single excitations and collective excitations, whereHere ωp2=4πNe2/m is the bulk plasmon frequency, N is number of free electrons per unit volume, e and m are the charge and mass of the electron respectively and ω0is a constant which is finite for a bound state but zero for a free electron.

Sample thickness is an important parameter in quantitative electron microscopy,. Several methods for determining sample thickness have been described previously. A theoretical method presented here utilizes the low loss region of the electron energy loss spectrum to calculate sample thickness. The method avoids many of the limitations of other methods and is applicable to a wide variety of samples. The method is compared to film thicknesses determined experimentally as described below.Chromium films were deposited simultaneously onto collodion supported TEM grids and cured epoxy blanks using a Denton Hi Res 100 coater equipped with a Syton quartz crystal monitor. The cured epoxy blanks were covered with a layer of unpolymerized epoxy that was then cured. The chromium sandwich was cross sectioned using ultramicrotomy techniques employing a diamond knife. Sections were produced which were as thin as possible (

Holography is a two step imaging process. After a hologram has been recorded it must then be reconstructed in order to obtain the stored image information. Gabor originally envisaged the use of optical reconstruction methods, but these are difficult to set up and limited in their flexibility. The alternative is to use digital techniques to reconstruct and manipulate the hologram and its reconstructions. This paper describes a program designed for the reconstruction of off-axis electron holograms.An off-axis electron hologram is an interferogram between a reference electron wave and the signal wave which has passed through the specimen. The hologram consists of a pattern of fringes modulated in position and amplitude as a result of phase shifts in, and scattering of, the signal wave. To recover phase and amplitude images from the hologram it is necessary to perform three operational steps. Firstly, the digitized hologram is Fourier transformed producing an array of complex numbers, one for each pixel in the original hologram.

The Bremsstrahlung spectrum from a bulk specimen has been successfully studied with high accuracy by Monte Carlo techniques. However, such procedures take considerable computing time since any change of experimental parameters requires a fresh computation. This paper describes an alternative approach which allows x-ray signals for a given structure to be computed within a fraction of a second. The method is illustrated here by an application to Bremsstrahlung. The procedure used is that of a diffusion matrix in which the strength of individual diffusion sources is derived by an initial Monte Carlo computation.The grid of a diffusion matrix Φ(yi,zj,Ek) which is established by Monte Carlo simulation contains the intensity of Bremsstrahlung x-ray coordinated at yi,zj with the photon energy of Ek. In the simplest case the matrix can be computed so that any photon crossing all the cells is accumulated. The results are stored in a 20×40×30 matrix for -RjiR in steps of R/20 where R is the Bethe electron range, and 0k0 in steps of E0/30. The accuracy of the matrix depends on the quality of the Monte Carlo simulation used to generate it.

The ionization of atoms by electrons is a process of great importance in physics, because it is the primary mechanism for energy loss of electron in matter and for X-ray microanalysis. Powell has critically compared various theoretical, semi- or empirical formulations with experiments and it is clear that although calculations and measurements have been carried out previously for K and to a lesser extent for L shells, very little data, both experimental and theoretical, exist for M shells.We have calculated K ,L (L1 and L23 ) and M( M1 ,M23 M45 )shell ionization cross section covering the entire periodic table and spanning the energy range from the critical ionization energy for a particular element up to 100 keV using Hartree-Slater central self- consistent field model. The calculation algorithm is essentially that described in Ref.3 but both exchange and correlation energy effects have been included in the computations to ensure that the computed cross-sections are valid at low overvoltage ratios, and relativistic corrections have also been included for accuracy at high incident energies.

Electron microscope use an electron beam to obtain various kind of information about specimen. The electron beam is focussed by electrostatic and magnetic fields and electron detectors employ electrostatic fields to attract or deflect electrons. In many cases the demand to calculate the electron trajectories in a fast and visual way is very strong. One of the most important questions is the problem of the secondary electron (SE) trajectories inside the SEM chamber and the effect of sample charging on detector yield. This is especially important in the low voltage SEM when investigating an uncoated, non-conductive specimen. A relatively large number of calculated trajectories gives a possibility to optimize SE detector placement as well as detector bias.The main problem is solving the Laplace equation in a 3-D space. In the 3-D space composed of cubic cells of dimension Δ3, the Laplace equation takes the following form

A recent paper by Ogura displays backscattered electron (BSE) images from a bulk, layered structure of AlAs/GaAs with repeat spacings of the order of 10nm. Typically BSE images are thought of as being inherently limited in resolution to a level which is of the order of a fraction of the Bethe range of the incident electrons, which for the beam energy and specimen used in these experiments would be of the order of 5 μm. The purpose of this paper is, therefore, to examine the conditions required for the nanometer level resolution achieved in this case.Backscattered electrons can be grouped into two classes, the BSI which are backscattered by a single high angle event soon after their entry into the specimen, and the BSII which are produced as the cumulative result of plural scattering. There is no sharp division between these groups but the two classes can be separated experimentally by energy filtering, since the BSI will have lost little energy in their short transit through the sample.

The scanning electron microscope (SEM) builds up an image by sampling contiguous sub-volumes near the surface of the specimen. A fine electron beam selectively excites each sub-volume and then the intensity of some resulting signal is measured and then plotted as a corresponding intensity in an image. The spatial resolution of such an image is limited by at least three factors. Two of these determine the size of the interaction volume: the size of the electron probe and the extent to which detectable signal is excited from locations remote from the beam impact area. A third limitation emerges from the fact that the probing beam is composed of a number of discrete particles and therefore that the accuracy with which any detectable signal can be measured is limited by Poisson statistics applied to this number (or to the number of events actually detected if this is smaller). As in all imaging techniques, the limiting signal contrast required to recognize a morphological structure is constrained by this statistical consideration. The only way to overcome this limit is to increase either the contrast of the measured signal or the number of beam/specimen interactions detected. Unfortunately, these interactions deposit ionizing radiation that may damage the very structure under investigation. As a result, any practical consideration of the high resolution performance of the SEM must consider not only the size of the interaction volume but also the contrast available from the signal producing the image and the radiation sensitivity of the specimen.

With the increasing importance of Low Voltage Scanning Electron Microscopy, the problem of describing the influence exerted by Secondary electron detectors on the path of primary electrons as well as its effects on the trajectories followed by secondary electrons become increasingly important. In situations where uncoated, insulating specimens are studied in an SEM, the additional problem of sample charging also has to be considered. Characterizing these interactions can be very difficult by conventional programming methods. The large number of points and the interdependence of the potentials at all of the points make the “bookkeeping” very difficult to manage. Spreadsheet programs with macroinstruction languages, however, can make these calculations much easier to perform. Not only can spreadsheets be used to calculate the potential field within a microscope column, macro programming can be used to calculate trajectories throughout that field. For the computations described in this paper, Microsoft Excel for the Macintosh was the spreadsheet chosen because of its effective use of the graphics capabilities of the Macintosh.

The spatial resolution of the scanning electron microscope is limited by at least three factors: the diameter of the electron probe, the size and shape of the beam/specimen interaction volume with the solid for the mode of imaging employed and the Poisson statistics of the detected signal. Any practical consideration of the high-resolution performance of the SEM must therefore also involve a knowledge of the contrast available from the signal producing the image and the radiation sensitivity of the specimen. With state-of-the-art electron optics, resolutions of the order of 1 nm are now possible. The optimum conditions for achieving such performance with the minimum radiation damage to the specimen correspond to beam energies in the range 1-3 keV. Progress beyond this level may be restricted by the delocalization of SE production and ultimate limits to electron-optical performance.

Introduction The Hitachi HF-2000 is a 200keV cold field emitter TEM, designed to offer both high resolution imaging performance and a nanometer probe-forming mode for microanalytical operation. The purpose of this paper is to report some initial microanalytical results obtained from the HF-2000 installed at the University of Tennessee.General Electron-Optical Parameters The cold field emission gun, which runs at about 5x10-9pA pressure, uses an electrostatic lens configuration and a six stage accelerator. The entire electron gun system is under the control of a dedicated microprocessor which permits all of the operational parameters of the gun to be set, monitored, and adjusted through a keyboard and display. The tip emission current can be adjusted up to 40μA, and the gun voltage ratio (that is the ratio between the voltage on the second anode and the tip extraction voltage) can be varied from 4.5 to 7.5, permitting a significant degree of flexibility in optimizing the optics of the emitter.

Electron stopping power is one of the most important fundamental parameters describing the interaction of an electron with a solid. The magnitude of the stopping power and its variation with energy determines the range of the electron, the rate of production of secondary electron, electron-hole pairs, and the exact form of the ϕ(ρz) curve. In its simplest form the stopping power, i.e the energy dE transferred to the sample of density ρ, atomic number Z and atomic weight A, as the electron of energy E travels a distance ds, is:(1)where the mean ionization potential I represents all of the possible inelastic scattering events available to the electron and is, to a zero-order approximation, about 11.6*Z eV. It has been well established that equation (1) is an accurate prediction of the stopping power at energies E of the order of tens of keV, but at the lower energies (i.e

Radiation damage to organic specimens is the major limiting factor in high resolution electron microscopy studies of biological systems. Electron beam irradiation compromises resolution by altering chemical microstructure, resulting in local mass loss and volume shrinkage in a specimen. All significant mass loss is thought to occur prior to a total incident dose of 50 electrons/ square angstrom If this is the case it is hard to reconcile the observation that images must be recorded at doses of less than 100 el/Å in order to avoid excessive mass loss and shrinkage while microanalytical (EDS and EELS) studies of the same tissue are routinely carried out at doses of 104 - 105el/Å2. Also, since most workers typically use either low dose (for imaging) or high dose (for microapalysis) there are apparently no studies in the literature which attempt to follow the process of radiation damage between these two extremes.We have chosen to investigate mass loss in polymer embedding resins such as are routinely used for TEM imaging as well as for X ray microanalytical applications.

A spectrum is a linear compilation of the output data, i.e signal intensity as a function of wavelength or energy, from an X-ray or electron energy loss spectrometer. In the earliest forms of the instrumentation the spectra existed only as traces on an oscilloscope or as the output plot of a chart recorder, and the operations that could be performed on them were therefore restricted to the identification of major features, and to manipulations that could be performed manually or graphically, such as estimating a background or integrating the area beneath a peak. In current instrumentation the spectrum is stored digitally in the memory of a computer. Since the data forming the spectral set can now be examined at will and transformed mathematically in any desired manner the spectrum can be processed to yield additional types of information. This paper will examine how spectral processing must be applied to obtain data from the X-ray and electron energy loss spectra generated from thin foils in the analytical electron microscope.

Electron energy loss spectrometry (EELS) is the study of the inelastic scattering events which an electron undergoes during its passage through a sample. When an analysis of the elemental constituents of the specimen is required the most important inelastic events are the ionizations of inner shells of the atom since these cause discontinuities ("edges") in the EEL spectrum at energies which are charactertistic of the element concerned. A simple electron spectrometer combined with an electron microscope makes it possible to use this information for sensitive elemental identification and localization for all elements above Lithium. The EELS can thus be regarded as an adjunct to the more conventional energy dispersive X-ray spectrometer (EDS), which is restricted to elements above Magnesium. This paper discusses the necessary parameters of a EEL spectrometer designed for materials microanalysis, its coupling to the microscope and the basic techniques of data collection and analysis.

Much of the success of the scanning electron microscope (SEM) can be attributed to the fact that the images that it produces appear similar to those seen by our eyes. Consequently the interpretation of SEM micrographs is perceived as a straightforward process. Although this casual approach works well enough for many qualitative purposes, when the SEM is to be used to make precise dimensional measurements of micron-sized features, such as resist strips or conductor pads, then more care must be paid to the details of image formation.Two types of electron signal, backscattered (BS) and secondary (SE), are usually available in an SEM. In many respects quantitative image interpretation is more straightforward for backscattered than for secondary electrons because Monte Carlo trajectory tracing techniques can be employed to compute the expected variation of signal with beam position.

In transmission electron spectroscopy the fundamental formula for elemental quantitation using inner-shell excitations gives the number n of atoms per cm2 contributing to the K-edge aswhere Ik and Io are the integrated intensities in the K-edge and zero-loss peak, respectively. Both of these integrals are measured for a spectrometer acceptance angle 2α and an energy interval ΔE. The parameter αk(α,ΔE) is the ionization cross-section for the same angular and energy parameters. The variation of αk with α and ΔE is a function of the generalized oscillator strength and little detailed information on this quantity is available. Therefore it is necessary to proceed empirically and the simplest assumption is thatwhere σk is a saturation (x-ray) cross-section and ηα,ηΔEcan be identified with efficiency factors.The accuracy of Equation (1) for K-edges from light elements (Li ≤ Z ≤ Al) is being tested by computer curve fitting and background stripping (see Fig. 1).

Personal computers (PCs) are a powerful resource in the EM Laboratory, both as a means of automating the monitoring and control of microscopes, and as a tool for quantifying the interpretation of data. Not only is a PC more versatile than a piece of dedicated data logging equipment, but it is also substantially cheaper. In this tutorial the practical principles of using a PC for these types of activities will be discussed.The PC can form the basis of a system to measure, display, record and store the many parameters which characterize the operational conditions of the EM. In this mode it is operating as a data logger. The necessary first step is to find a suitable source from which to measure each of the items of interest. It is usually possible to do this without having to make permanent corrections or modifications to the EM.

The resolution of a secondary electron image is determined by the spatial distribution of the secondaries leaving the surface. This distribution results from two components, SE1 which are the secondaries generated by the incident beam and which carry the high resolution information, and SE2 which are the secondaries produced by exiting electrons and which carry information mimicing that in the backscattered signal. The integrated intensities of these two components are comparable, but their spatial characteristies are quite different. In order to study the factors which limit resolution it is therefore necessary to model the events which give rise to the two groups of secondaries.

Although the "cold" field emission gun has been used successfully for both transmission and scanning electron microscopy it requires ultra-high vacuum which is not obtained easily when such a gun is interfaced to a conventional microscope system. Recently, the "thermal" field emission gun (TFEG) in which the emitting tip is held at around 1700°K has been proposed as an alternative electron source for such applications. Under this condition the tip is cleaned continuously, and surface asperities are smoothed, therefore stable operation is possible in a high vacuum. In this paper we report on the build-up characteristics, current stability and brightness of a TFEG which has been interfaced to a JEOL JEM 100B microscope equipped with a scanning attachment. The gun consists of a (111) tungsten emitter set on a rhenium filament, three anodes and a two stage magnetic alignment system. The gun chamber is ion pumped to a pressure in the range 6xl0-8 to 2xl0-9 torr.

The image obtained in a scanning transmission electron microscope (STEM) is related to that obtained, under otherwise identical conditions, in the conventional transmission electron microscope (CTEM) by the principle of reciprocity. This principle is only applicable under rigorously defined mathematical conditions which are difficult to achieve in practice, nevertheless its general validity has been demonstrated experimentally for both amorphous and crystalline materials. The principle of reciprocity states, in part, that if equivalent beam cone angles are utilized then the STEM and CTEM images should be similar. Conversely, for a given set of electron-optical conditions in the STEM mode, the expected form of the images can be deduced from a knowledge of the CTEM image under reciprocally related conditions.

Image simulation, as a means of image interpretation, has been an important component of high resolution transmission electron microscopy for many years. By contrast scanning electron microscopists have relied almost exclusively on simple visual analogies to understand the micrographs that they produce. However with the development of SEMs capable of near atomic levels of resolution, and with the requirement for accurate dimensional metrology of sub-micron features in semiconductor devices using the SEM, a more quantitative approach to image formation and analysis is needed.Image simulation in the SEM encounters several special problems. Firstly, the signal displayed to form the micrograph is not well defined in terms of either the contrast mechanisms which give rise to it, or the energy and origin of the electrons actually collected to produce the signal.

The formation of glassy (amorphous) metallic alloys by rapid quenching from the liquid state was discovered in 1960 (1), and subsequent work revealed that certain metallic glasses are ferromagnetic (2). Recently, transition metal based alloys, which possess Curie temperatures as high as 700°K and saturation magnetizations up to 15,000 Gauss, have been shown to be both mechanically very hard and magnetically quite soft (3). We have applied the SEM technique of Fathers et al. (4) as well as electron transmission Lorentz microscopy to a study of roller quenched (5) metallic glass ribbons of composition: TM0.75P.16X.06A1.03, where TM denotes Fe, Co, Ni, or an alloy of these and X denotes either B, C or Si (6).

Electron beam irradiation of organic molecular crystals almost invariably leads to a rapid loss of crystallinity, followed by gross mass loss at higher doses. One interesting exception to this sequence of events was described, by Jones, Thomas and Williams who showed that the irradiation of p-terphenyl, at temperatures below 130°K, converted the stable triclinic phase to the characteristic room temperature monoclinic structure. For doses above 0.16 coulomb.cm-2 this transformation was permanent and could not be reversed by thermal cycling. We report here another electron-beam induced transformation, the end product of which has technologically useful properties.3,4,9,10-Perylenetetracarboxylic di-anhydride, C24H8O6 (PTCDA), the structure of which is shown in figure 1, when vapor deposited to a thickness of a few thousand angstroms forms a red transparent organic film. In this form it is stable to relatively high temperatures (300°C) does not appear to degrade after several months of room temperature storage and is electrically insulating.

In a crystalline solid the regular arrangement of the lattice structure influences the interaction of the incident beam with the specimen, leading to changes in both the transmitted and backscattered signals when the angle of incidence of the beam to the specimen is changed. For the simplest case the electron flux inside the specimen can be visualized as the sum of two, standing wave distributions of electrons (Fig. 1). Bloch wave 1 is concentrated mainly between the atom rows and so only interacts weakly with them. It is therefore transmitted well and backscattered weakly. Bloch wave 2 is concentrated on the line of atom centers and is therefore transmitted poorly and backscattered strongly. The ratio of the excitation of wave 1 to wave 2 varies with the angle between the incident beam and the crystal structure.

Electron channeling patterns (ECP) were first found by Coates (1967) while observing a large bulk, single crystal of silicon in a scanning electron microscope. The geometric pattern visible was shown to be produced as a result of the changes in the angle of incidence, between the beam and the specimen surface normal, which occur when the sample is examined at low magnification (Booker, Shaw, Whelan and Hirsch 1967).A conventional electron diffraction pattern consists of an angularly resolved intensity distribution in space which may be directly viewed on a fluorescent screen or recorded on a photographic plate. An ECP, on the other hand, is produced as the result of changes in the signal collected by a suitable electron detector as the incidence angle is varied. If an integrating detector is used, or if the beam traverses the surface at a fixed angle, then no channeling contrast will be observed. The ECP is thus a time resolved electron diffraction effect. It can therefore be related to spatially resolved diffraction phenomena by an application of the concepts of reciprocity (Cowley 1969).

The image produced by the SEM is not an intrinsic property of the object viewed but is also a function of the electron-optical conditions used to generate the image as well as of the chemical and geometric characteristics of the specimen itself. The image is therefore not a simple representation of the specimen, but must be interpreted to yield the desired information. A general purpose method for image interpretation is to perform a detailed contrast simulation for the specified object and to then examine the influence of various factors such the beam energy or signal mode on the form of the image so as to deduce the rules which relate the form of the image to the form of the specimen.In the computations shown here a Monte Carlo simulation is used to model the interaction of the incident beam. The generation of the secondary electron signal is incorporated using the method proposed by Salow, and the escape probability of these secondary electrons is computed using a simple exponential approximation.

A variety of standard specimens is needed in order to systematically investigate the instrumentation, specimen, data reduction and quantitation variables in electron energy-loss spectroscopy (EELS). Pure single element specimens (e.g. various forms of carbon) have received considerable attention to date but certain elements of interest cannot be prepared directly as thin films. Since studies of the first and second row elements in two- or multicomponent systems will be of considerable importance in microanalysis using EELS, there is a need for convenient standards containing these species. For many investigations a standard should contain the desired element, or elements, homogeneously dispersed through a suitable matrix and at an accurately known concentration. These conditions may be met by the technique of implantation.Silicon was chosen as the host lattice since its principal ionization energies, EL23 = 98 eV and Ek = 1843 eV, are well removed from the K-edges of most elements of major interest such as boron (Ek = 188 eV), carbon (Ek = 283 eV), nitrogen (Ek = 400 eV) and oxygen (Ek = 532 eV).

Micro-computers conveniently provide the basis of a powerful data collection, display and storage system for an electron microscope. In the example described here a Vacuum Generators HB501 STEM has been interfaced with an APPLE II micro-computer. The APPLE, with 48k of available core memory and a single mini-diskette system of 256k. byte capacity, was fitted with 16 input channels of analog-to-digital conversion (Interactive Structures AI-Ø2) and 8 output channels of digital-to-analog conversion (Interactive Structures AO-Ø3).Figure (1) shows a block diagram of the system and its interconnections to the microscope. ADC channels Ø to 5 monitor the excitations of the six lenses (two condenser, one objective and three post specimen) of the HB501. These inputs are taken directly from the feedback shunt resistors in the lens power supply and fall directly in the range 0 to +5 volts required for the ADC. Channels 6 to 9 measure the X,Y coordinates and X,Y tilt-coordinates of the specimen stage, by monitoring the voltages derived from the position-sensing potentiometers on the stepping drive motors. These signals in the range ±0.5V are converted to the required 0 to + 5V level by operational amplifiers.

High-resolution images of the surface topography of solid specimens can be obtained using the low-loss technique of Wells. If the specimen is placed inside a lens of the condenser/objective type, then it has been shown that the lens itself can be used to collect and filter the low-loss electrons. Since the probeforming lenses in TEM instruments fitted with scanning attachments are of this type, low-loss imaging should be possible.High-resolution, low-loss images have been obtained in a JEOL JEM 100B fitted with a scanning attachment and a thermal, fieldemission gun. No modifications were made to the instrument, but a wedge-shaped, specimen holder was made to fit the side-entry, goniometer stage. Thus the specimen is oriented initially at a glancing angle of about 30° to the beam direction. The instrument is set up in the conventional manner for STEM operation with all the lenses, including the projector, excited.

High resolution scanning electron microscopy is still a relatively new and unfamiliar concept because in the early days of the SEM it was expected, that secondary electron imaging would be limited to a resolution of between 5 and 10nm at best. Now, however, because of improvements in instrumentation and technique based on advances in the understanding of electron beam interactions with solids current SEMs can demonstrate spatial resolutions below 1nm, rivaling those obtained by transmission instruments.High performance scanning electron microscopes always incorporate two advanced items of instrumentation. Firstly they use field emission guns (FEGs). The high brightness, low energy spread, and small source size of the FEG makes it possible to produce an electron probe of sub-nanometer size which contains sufficient current for secondary electron imaging (i.e 10-12 amps or more) and which can maintain this performance over a wide energy range (3 to 30keV). Secondly, the new high performance instruments place the specimen within a high excitation, immersion, probe forming lens.