key: cord-023442-4vzwc2d2
authors: nan
title: Proceedings of SCANNING 94/SEEMS 94 Charleston, South Carolina, USA
date: 2006-12-05
journal: Scanning
DOI: 10.1002/sca.4950160315
sha: 
doc_id: 23442
cord_uid: 4vzwc2d2

nan

The widespread application of Monte Carlo electron trajectory modeling has never been realized in the electron microscopy community because of the computation-intensive nature of the Monte Carlo algorithm (e.g., many hours of computer time were required to run one simulation which might improve one's ability to interpret data if sufficient statistics were obtained). Massively parallel supercomputers with multiple-instruction multiple-data (MIMD) architecture are a good platform for Monte Carlo electron trajectory codes without the complications of vectorization of the computer code required on vector supercomputers (e.g., CRAY-YMP). The NIST Monte Carlo electron trajectory simulation code has been adapted to run on a massively parallel computer. For the first time, the increased speed of the calculations has made Monte Carlo calculations a real-time tool for data interpretation. The increased speed achieved by the parallelization of the Monte Carlo code results in the ability to model small probability events due to the massively parallel Monte Carlo (MPMC) code's ability to simulate large numbers of electron trajectories quickly. Applications include both thin-film and bulk x-ray microanalysis and examples of contrast in backscattered electron images obtained in the SEM. This work was supported by the United States Department of Energy under contract #DE-AC04-94AL85000.

In electron probe x-ray microanalysis, "secondary fluorescence" refers to the emission of characteristic x-rays following photoelectric absorption of the primary x-rays generated by the electron beam. While generally a minor contribution to the total x-ray emission, secondary x-ray fluorescence can become an important issue, when the spatial resolution of analysis is of interest, because of the significantly greater range of x-rays than of electrons. Thus, for a nickel-10% iron alloy excited with a 15 keV electron beam, the range of excitation of the primary radiation will be a hemispheric volume with a diameter of about one micrometer, while the range of secondary fluorescence of iron K-shell radiation by nickel K-shell radiation will be a hemispheric volume with a diameter of approximately 60 micrometers for 90% of the total secondary radiation. Such long-range production of secondary radiation is important when the measurement of trace element distributions near phase boundaries is attempted in situations where an analyte appears at high concentration across the boundary.

Monte Carlo electron trajectory simulation provides a powerful tool for the calculation of the three-dimensional distribution of the primary radiation. To calculate the spatial distribution of the secondary fluorescence due to the absorption of the primary radiation, a second calculation is needed that integrates the x-ray absorption equation in three dimensions. The resulting hybrid calculation can give the total secondary radiation induced for a given structure, such as a planar interface between two materials.

The transport phenomenon of incident electrons in solid materials is an important subject in microscopy, microanalysis, and microlithography. It has been treated by either an analytical method, the transport equation, or Monte Carlo simulation. Because of the flexibility of modeling, Monte Carlo simulation has been developed extensively.

Various models for Monte Carlo simulation proposed so far are classified into four groups from the aspect as to how inelastic interactions are treated. The most simple model is the one which uses the inelastic mean free path. This model is applied, for example, to spectroscopic analyses of elastically reflected electrons, Auger electrons, and photoelectrons. The most popular model is the single scattering model in which the step length is taking the free path of elastic scattering of electrons, and an energy loss during their traveling is calculated by the continuous slowing-down approximation of Bethe, [dE/ds] Bethe . The most complicated model is the direct simulation model in which all inelastic interactions are taken into consideration as the discrete process. This model requires all differential cross sections for inelastic scattering and, thus, a long computational time. The last one, a compromise model, is the hybrid model which introduces partially the discrete process of inelastic scattering. In the model the continuous energy loss [dE/ds] (1) where [dE/ds] dis includes both energy losses due to core electron ionization and valence electron excitation. A variety of the cross sections has been used for inelastic scattering. The fast secondary production model uses the Moller equation for the cross section, assuming that all atomic electrons are free (Murata et al. 1981 ).

The authors have published a hybrid model by using the differential inelastic cross sections of Vriens for core electron ionization and the Moller cross section for free electron excitation (Murata et al. 1992 ). The Mott cross section is used for elastic scattering. Typical results for backscattering have shown fairly good agreement with experimental data.

The present paper gives a brief survey of Monte Carlo modeling and discusses the validity of the hybrid model mentioned above. Figure 1 shows a typical example of the calculated depth distribution of generated x-rays in an Au target in comparison with the experimental result of Castaing and Descamps (1955) . Agreement is good. Also shown is the result without the discrete processes of inelastic scattering. An appreciable deviation is seen in the vicinity of the peak. However, the effect of the energy straggling is not so significant. We have also done calculations for other x-rays and have obtained reasonable agreement with experimental data. The effect of the energy straggling will appear more significantly in electron scattering in a thin film. Figure 2 shows the energy distribution of transmitted electrons, with a comparison between two Monte Carlo results and the experimental data of Shimizu et al. (1976) . The agreement is not as good; it seems that the disagreement comes from insufficient discrete processes because we still keep the continuous energy loss process.

We also propose a new model to improve the discrepancy in Figure 2 , caused by energy straggling. The straggling is impor-tant in thin-film analysis, especially at initial energies near the ionization energy for x-ray production. Discussions of ways to improve Monte Carlo simulations have usually concentrated on topics such as the effect of the choice of scattering cross section, or the appropriate model for electron stopping power at low energies. Much less attention has been given to considering whether or not we actually have sufficient experimental data to make it possible to demonstrate by a comparison of simulation and experiment that one model, or a portion of that model, actually performs better than another alternative.

The earliest experimental data on electron-solid interactions was published 95 years ago (by Starke in Germany in 1897, and by Campbell-Swinton in England in 1898), and numerous workers since then have contributed to the literature on this topic. Unfortunately, there never seems to have been any attempt to collect and collate all of this material, and consequently workers seeking data such as the variation of the secondary electron yield with energy for silicon have been forced to conduct a random search of the literature to find what values are available. It is not surprising that such a search typically finishes as soon as a set of data plausibly matching the simulated values is found.

In order to try and remedy this deficiency, a systematic search of published data has been carried out to generate material for a rudimentary database, with the hope that this will provide some of the necessary numbers against which Monte Carlo simulations can be tested. In its current form the database, derived from 70 separate references, is divided into four segments arranged by atomic number (or compound) and comprising secondary electron yields, backscattered electron yields, x-ray ionization cross sections, and electron stopping power. Only experimental values are included; thus interpolated, extrapolated, or normalized data sets, or values not specifically indicated by the author to be experimental, have been removed. This restriction unfortunately eliminates most of the voluminous secondary electron data, since such published values have invariably been normalized in order to facilitate fitting to a yield curve equation. No attempt has been made to judge the quality of any of the data since any such assessment would be premature until a sufficient number of independent values are available to permit obviously erroneous, or possibly dubious, results to be safely identified and eliminated. The database is available on request from the author both in printed form, or as a set of CricketGraph ™ files for the Apple Macintosh.

While the quantity and quality of data available for a given element vary widely, they are mostly sparse and scattered. For about half the elements in the periodic table, no experimental values ever appear to have been published, and data are conspicuously absent for even the most common alloys and compounds. Even for an element such as silicon ( Fig. 1) , the scatter in the data is too large to make it possible to be certain to better than a factor of 2 what the SE yield is at some energies. Thus, unless a major source of quantitative results has been overlooked, it is fair to conclude that the experimental data presently available are not good enough to allow the merits of competing Monte Carlo models to be judged, or even to permit specific numerical comparisons (e.g., the backscattered yield of carbon, copper, and gold at 5 keV) to be made with any satisfactory level of certainty. On the positive side, however, the database does give some indication of interesting trends (e.g., the variation of secondary yield with atomic number) that have not been easily accessible previously.

School of Electrical Engineering and the National Nanofabrication Facility, Cornell University, Ithaca, New York, USA Two simulation programs have been developed at Cornell, SEEL [1] [2] [3] (for Simulation of Electron Energy Loss) and PYRA-MID. SEEL is a two-dimensional Monte Carlo program for the simulation of electron trajectories in complex, multi-material nanostructures; the program incorporates energy loss models that include quantum mechanical cross sections for energies down to < 100 eV. The second program, PYRAMID, 4, 5 is a very fast electron beam proximity correction program that can perform proximity correction (electron scattering correction) for complex ultra-large-scale integrated circuit patterns with 100 nm minimum feature sizes (MFSs). PYRAMID uses SEEL to generate the point spread function or radial exposure distribution (RED) as the input.

SEEL simulations have been tested against numerous published electron energy loss data and PYRAMID has been tested on ULSI density (100 nm MFS) exposures using 0.5 µm PMMA. Simulation results are compared with experimental data to evaluate PYRAMID's performance.

A more recent application of SEEL is the evaluation of signal-noise ratio (SNR) for time-resolved microscopy of microelectromechanical structures (MEMS) . For this application, we are interested in estimating the SNR for high-speed video recording 6, 7 of moving high-aspect-ratio MEMS oscillating > 1 MHz, particularly applications of nanoelectromechanical structures for metrology and time-resolved characterization. IV-4 Scanning Vol. 16, Supplement IV (1994) Simulation of image formation and detection systems in the SEM is a vital link in performing image analysis to obtain precise measurements, to provide the necessary connection between image parameters and structural dimensions, and to reflect important microscope beam and detector parameters. Monte Carlo methods allow a wide degree of freedom in specifying simulation conditions for sample composition and geometry. Published examples often limit the simulation to two-dimensional structures, with a zero-diameter electron beam and all secondary or backscattered electrons (BSE) collected. A more useful and realistic approach will take into account the effect of beam diameter, and detector geometry and gain characteristics. These effects for simulated BSE images for three-dimensional patterned structures of carbon on a silicon substrate are shown in Figure 1 .

The Monte Carlo simulation used here is based on a single-scattering procedure with the Rutherford scattering cross section and the Bethe energy loss formula modified for low-energy primary electrons. The program is a further development of a single-scattering Monte Carlo program (in Pascal) widely distributed by David Joy 1 and modified later by Russ et al. 2 The program is written in C language and runs on a Sun workstation. The program can scan the beam position in x and y directions over an arbitrary multielement and multilayer sample with topographical features to produce images.

Using the Monte Carlo procedure, it is possible to extract information about the position, energy, and direction of the electron at each scattering point. This allows tracking BSEs as a function of angle. If the dependence of detector gain on electron energy is added, then the details of signal formation can be modeled. The effect of electron beam shape and diameter on the image can be added by convolution after an ideal image is generated. Figure 2 shows the matrix representing a Gaussian electron beam used in this work. An example of a simulated BSE-SEM image is shown in Figure 3 for a structure with carbon features on a silicon substrate. 

FIG. 2. Contour plot of Gaussian shape electron beam with standard deviation of 64 Å was used to convolute the data obtained for a zero-width electron beam.

FIG. 3. BSE image simulated using a zero-diameter electron beam (left) and after convolution using the Gaussian beam width shown in Figure 2 (right). The simulation was performed for 2 keV primary electrons with 1000 trajectories for each point. The image represents 100 × 100 pixels.

Monte Carlo simulations of electron scattering in a target normally use one of two elastic cross sections, either the screened Rutherford cross section or tabulated partial wave expansions of the Mott cross section. The screened Rutherford cross section gives acceptable results for high energies and low atomic numbers, but Mott cross sections are required for low to medium incident energies (0.1-30keV) and high atomic number targets. However, computations tend to be slow using tabulated data due to the need to interpolate between data points. Empirical equations for the total and differential electron/atom elastic scattering cross sections have been found that can be substituted for tabulated Mott cross sections in predicting backscattering coefficients.

The total elastic Mott scattering cross section 1 is fitted by similar form to the screened Rutherford cross section but contains three terms in energy in the denominator. The empirical total elastic scattering cross section is valid for atomic numbers up to 92 and for energies from 100eV to 30keV: (1) The fit to the differential Mott cross sections is decomposed into two parts, one part being of the same mathematical form as the screened Rutherford cross section (σ R ), and the second part being an isotropic distribution (σ I ). The screened Rutherford part of the differential scattering cross section is first fitted to the half angle of the Mott cross sections. This fit of the differential screened Rutherford is in turn reduced to a fit of the screening parameter alone over energy and atomic number. In marked contrast to the screened Rutherford cross section, the tabulated Mott cross sections show only a small overall downward trend in half angle with increasing atomic number(Z). Implying an average Rutherford screening parameter for all Z, with E the electron energy, of:

The ratio of the total cross sections (σ R /σ I ) between the screened Rutherford part of the differential scattering cross section and the isotropic part of the distribution is fitted to the backscattering coefficients calculated directly from tabulated Mott cross sections. The ratio of Rutherford to isotropic cross sections is:

(3) Figure 1 shows a comparison of the calculated backscattering factors using the present empirical fit (solid lines) with those calculated using Mott cross sections. 1 The fit for Al, Cu, and Au is good over the entire energy range. The fit for Ag is moderate and the fit for C is high. However, most deviations are similar to differences because of the use of different atomic models in the Mott cross sections and are acceptable. There are two major reasons why the simple monotonic Eqs. 1-3 work well. First, the scattering of the electrons in a solid is a multiple scattering process. Thus, many of the complex quantum interference effects are averaged out. Second, the elastic backscattering is monotonic with atomic number. These two factors serve to smooth out the effects of the complex multidimensional cross sectional surface that is being fitted over Z, E, and θ.

Reference IV-6 Scanning Vol. 16, Supplement IV (1994) σ Rutherford σ Isotropic = 300E 1− Z / 2000 Z + Z 3 3 × 10 5 E σ T = 3.0 × 10 −18 Z 1.7 (E + 0.005Z 1.7 E 0.5 + 0.0007Z 2 / E 0.5 ) A scanning interference microscope may be constructed by allowing light which has probed the object and light which has not probed the object (the reference beam) to interfere on a suitable detector. If the detector is large in extent, a conventional scanning interference microscope results, whereas if a point detector is used we have a confocal scanning interference microscope. The image in both cases may be regarded as a superposition of three terms. The first represents a normal conventional or confocal image depending on the kind of interferometer used. The second we will call an interference term image, whereas the third represents a constant background. It is possible to design a system in which the interferometer term image is identical for both conventional and confocal scanning systems in all respects, including optical sectioning. A simple scheme permits the interference term image to be detected separately and then processed in a variety of ways.

Let us consider two specific geometries. The first is an almost common path confocal system based on a single mode optical fibre. The interferometer in this case is confocal. Simple image processing then permits surface profilometry to be performed. The second implementation is a conventional scanning interference system. The interference term image from this conventional image may again be isolated and processed to give a confocal image.

In biology, the description and precise representation of microstructural forms is of increasing importance for threedimensional (3-D) computer image understanding, in particular to enhance 3-D visualization and analysis. The intermediate level of computer vision, located between the bottom layer (signals) and the top layer (model) is best suited as a starting point to improve multidimensional image understanding. We implement the concept of 3-D topology embedded in the bottom-up structure of data processing to enhance subsequent rendering and specific quantitative analysis.

Segmented and contoured serial section images require sophisticated algorithms to generate correct surface-rendered views. We have, therefore, developed a method to evaluate connectivities between sections, based on bijective correspondance analysis of the center of gravities of contours. These connections indicate nods and branches of 3-D structures and can be interactively edited in exploded views of the contour stack. Because of the strong data reduction, such procedure can be implemented even in computer graphic systems with entry graphics.

The topological skeletons are the backbone to render correctly biological structures of free forms, in particular those featuring frequent branching patterns. Examples where such an analysis is successful include microvessel networks, dendritic trees, lung anatomy, or dental root canals (Baumann et al. 1993) . Moreover, the automatic labeling of connected structures gives rise to the analysis of topological criteria, which was only available in stereological methods until now. This includes criteria such as connectivity and branching angle, or higher representations of branching schemes. In return, topological connectivities can be used to improve the segmentation of images in the top down-process of data processing, or it can serve as embedded analytical graphics to improve volume renditions.

The mammalian central nervous system (CNS) contains at least ten times more glial cells than neurons. The three major populations of CNS glia are astrocytes, oligodendrocytes, and microglia. Astrocytes and oligodendrocytes together are often referred to as macroglial cells and arise from primitive neuroectodermal precursor cells, while microglia are derived from the mesodermal germ layer. It is thought that common precursor cells, known as O-2A progenitors, can differentiate into either astrocytes or oligodendrocytes during gliogenesis. In contrast, the origin of microglial cells remains enigmatic, although there is substantial evidence that microglial cells arise from primitive hematopoietic stem cells that gain access to the CNS at a very early stage of development.

Much of our knowledge about glia and glial cell function has been derived from histochemical and electron microscopic studies. Astrocytes can be visualized reliably using immunohistochemical methods with antibodies against the glial fibrillary acidic protein (GFAP). GFAP is an abundant constituent of intermediate filaments which can seen ultrastructurally in most astrocytes. In addition, astrocytes also contain enzymes, such as NADPH diaphorase and glucose-6-phosphatase, which are readily detected by enzyme histochemical methods.

We have focused much of our attention on developing methods for detecting oligodendrocytes and microglial cells in sections of rat brain. Lectin histochemistry has been a particularly useful tool in studying these types of CNS glial cells. Using lectins I and II from Griffonia simplicifolia seeds with carbohydrate specificities against α-D-galacatose and Nacetyl-D-galactosamine (GlcNAc), we were able to demonstrate selective labeling of microglia and oligodendrocytes, respectively. The B 4 -isolectin from Griffonia simplicifolia coupled to horseradish peroxidase (GS I-B 4 -HRP) can be used to detect a membrane-bound glycoprotein on the microglial cell surface at all stages of CNS development ranging from the early embryonic age to adulthood. In contrast, oligodendrocytes were found to express GlcNAc-containing glycoproteins in the perinuclear cytoplasm using biotinylated GSL II. The perinuclear staining was determined ultrastructurally to be associated with Golgi complexes. Biochemical analyses using tricine/SDS-polyacrylamide gel electrophoresis and western blotting with GSL II showed the GlcNAc-containing glycoproteins to be insoluble, with molecular masses ranging from 15 to 30 kD. Having available specific markers for the three major glial cell groups, we were able to combine lectin histochemistry with immunohistochemistry to perform double-labeling studies demonstrating specificity of each stain for a given glial cell type.

Following the study of glia in the normal CNS, we went on to investigate glial cell reactions that occur as a consequence of nervous system injury or disease. It became immediately ap-parent that microglial cells were the major glial cell type responding to neuron injury. Microglia not only proliferate and change their morphology in response to CNS damage, but they can also vary their membrane phenotype by expressing new molecules on their surface. We found that antigens of the major histocompatibility complex (MHC), which are largely absent from the normal CNS, are expressed de novo on microglia and related perivascular cells responding to neuron injury. Our studies, which have included various neuropathologic conditions including stroke and brain tumors, have shown that the expression of MHC antigens, as well as other related immunomolecules, is always restricted to cells of the microglial lineage. These findings strongly suggest a role for microglia as indigenous immunocompetent cells of the CNS.

JEREMIAH R. LOWNEY National Institute of Standards and Technology, Gaithersburg, Maryland, USA A scanning electron microscope (SEM) can be used to measure the dimensions of the microlithographic features of integrated circuits. However, without a good model of the electron-beam/specimen interaction, accurate edge location cannot be obtained. A Monte Carlo code has been developed to model the interaction of an electron beam with lines lithographically produced on a multilayer substrate. The purpose of the code is to enable one to extract the edge position of a line from SEM measurements. It is based on prior codes developed at NIST 1 but with a new formulation for the atomic scattering cross sections and the inclusion of a method to simulate edge roughness or rounding. The code is currently able to model transmitted and backscattered electrons, and the results from the code have been applied to the analysis of electron transmission through gold lines on a thin silicon substrate, such as used in an x-ray lithographic mask. There is provision for both transmitted and backscattered electron detectors. By comparing the predictions of the code with measured data, it is possible to obtain edge positions to the order of 10 nm, which is needed for the advanced lithography projected for the year 2000. 2 The uncertainty of these measurements is limited by the sample geometry and surface roughness and not by the measurement process.

Much of the improved code is devoted to the treatment of boundary crossings by the electrons. The present code allows for a substrate of at most three layers and one or two identical lines with a trapezoidal cross section on top. There is also provision for a symmetrical jog (i.e., a discontinuous change in the width of the trapezoid) along the edges to simulate edge roughness and rounding. The three layers that form the substrate, which are typical of an x-ray mask membrane, are silicon, polyamide, and chromium. The lithographically produced lines are gold, and the chromium improves adherence of the lines to the substrate. The code is easily modified to model other media by simply changing the atomic parameters in the input subroutine. The x-ray lithography mask, which has been used as a test sample, is a very good model system for the development of accurate SEM standards because it provides a measurement of both transmitted and backscattered electrons for comparison with the predictions of theoretical models.

A plateau in the transmitted and backscattered electron signal occurs as the beam traverses the sloping edge of the line trapezoid. This effect can be blurred by edge roughness, and the effects on the plateau of various widths and heights for the jog can be shown as well as the effects of rounding at the bottom of the lines and the calculated noise level for various numbers of trajectories. Direct comparisons with measured transmission through an x-ray mask of gold on a silicon membrane can be used to demonstrate the determination of the edge of the gold line. Figure 1 shows the transmission (measured downward) along the axis of a gold line indicating the plateaus in the data near the middle of the edges. 3 Figure 2 shows a simulation of the data with the plateaus at nearly the same location along the edges as in the data. This work shows how high-resolution metrology of the features produced by advanced lithography can be obtained with an SEM. Extensions of this code to modeling secondary signals as well as the effects of charging are planned. At Philips Research, a Monte Carlo program for electron microscopy is developed, which simulates electron-solid state interactions in the energy range 0.5-50 keV. It is based on the program published by L. Reimer (Scanning 8, 265, 1986) .

In particular, to ensure proper operation in the low-voltage region, Mott cross sections for elastic scattering are calculated by numerically solving the Dirac equation. The model for inelastic scattering treats inner shell ionizations as discrete events, described by a scaled Gryzinski formula, whereas the effect of valence and/or conduction electrons is incorporated as a continuous Bethe loss. This model, which differs from similar models presented in the literature, provides good fits to experimental stopping power data. The program is extended to include multilayer specimens, each layer composed of multiple elements.

One of our objectives is the modeling of CD linewidth measurements in a high-resolution low-voltage scanning electron microscope. Our Monte Carlo program can generate top-view backscattered electron (BSE) and secondary electron (SE) video profiles of lines with variable slope and pitch. We use the SE generation model of D.C. Joy (J Microsc, 147, 52, 1987) , which is extended to account for the regions near the corners of the line profile. Furthermore, the recollection of SE by the sample is accounted for. As a result, realistic pure SE profiles are generated. The primary beam parameters include the probe size and the depth of focus. Determination of the top and bottom linewidths via the commonly used heuristic algorithms reveals the systematic errors of these methods. A simple example is the probe size dependence of the peak-to-peak width, which can be related to the strong asymmetry of these peaks. We expect that improved algorithms can be developed, which use a modest database acquired by Monte Carlo simulation. A scanning electron microscope (SEM) fitted with a helium-neon laser interferometer is used to measure the widths of features on photomasks. In this way the magnification of the SEM can be known very precisely. Algorithms yielding good measurement repeatability, which use the back-scattered electron (BSE) signal, have been developed (Nunn 1990, Nunn and Turner 1989) but in order to be able to relate measurements made on the image to the physical dimensions of the artefact, it has been necessary to model the image formation process.

The essential details of the "plural scattering" model used in this work have been described by D.C. Joy (1988) . Typical geometries and materials of photomasks have been modelled along with a range of accelerating voltages and beam diameters to observe how the image is affected by the different parameters. More important, the modelled image intensity profiles are studied to relate the position of the physical foot of the sloping edges of actual physical lines to the broadened image intensity profile of the edges.

Although the "single scattering" model, also described by Joy (1988) , is more rigorous and more applicable to the type of samples used in this work, it was not used in the first instance because of the much greater computer time required.

Results from the modelled profiles suggest that the position of the 50% thresholds of the BSE image intensity profiles are located a few tens of nanometres away from the physical position of the edge. The exact offset depends on the angle between the sloping sides and the horizontal. The modelled offset increases from approximately 5 nm for a vertical edge to 20 nm for an edge angled at 60˚ to the horizontal.

The effect of the BSE detector size and geometry on the signal detected has also been modelled. The model shows that for best performance the detectors should collect the largest possible solid angle, but that as long as the detector is placed symmetrically, the only adverse effect of a smaller detector on the image is a loss of signal to noise.

Measurement comparisons between optical microscopy and the SEM corrected by the "plural scattering" modelled offsets have been encouraging (Nunn and Turner 1991) ; nevertheless, if time permits, a more rigorous study using more appropriate models should be pursued. Audio enthusiasts carefully read the test reports when looking for their new equipment. However, usually no one tries to find out the frequency response, distortion or signal-to-noise ratio of a very expensive, complex system such as a scanning electron microscope (SEM). SEMs were once used mainly as image-gathering devices, and in spite of certain obvious and sometimes serious problems they served as acceptable and effective instruments for many applications. These "image grade" microscopes are now being replaced with much better quality "measurement grade" instruments. The guaranteed resolution is no longer enough to ensure good performance and other factors must be considered as well. Even though these newer systems have field emission of LaB 6 electron guns, the main advantage is in the built-in, computer controlled, digital image and data-processing capabilities. There are several alternatives to upgrading an existing microscope with an external, usually desktop computer-based measuring system. Soon the SEM, similar to a scanner or camera, will just be a part of an imaging network, and automatic image enhancement and processing will be commonplace.

This presentation deals with a short description of the electrical properties of different components of an SEM; the beam scanning circuitry, the video signal chain, sampling, analog to digital (AD) and digital to analog (DA) conversion. It will relate their possible influences to the detected and observed signal to improve the imaging. This will also make possible better comparisons of measured data to the computer-modeled data.

Essentially, the main parts of an SEM essentially are the electron source, electromagnetic coils, DC power supplies, (scan) generators, detectors, amplifiers and, finally, the displays. The quality of an SEM image depends on the characteristics of the primary electron beam, the correctness of the beam scanning, signal detection, and the fidelity of the video signal chain.

The correctness of the beam scanning not only means good linearity and proper amplitude and direction of the deflection of the primary and displaying electron beam, but the displaying or data assigning has to be in correct synchrony also. The deflection coils are fundamentally nonlinear and inherently have hysteresis; hence, correcting circuitry is indispensable. The total distortion caused by improper magnification, nonlinearity, and hysteresis can be greater than 30% and it is difficult to keep it below 1% 4 .

The fidelity of the video signal chain depends on its transfer function, noise, and distortion figure. The main parts of the video chain of a modern SEM are the detector, amplifier, AD converter, computer memory (disk, imaging network), DA converter, amplifier, and the display or printer. All of these parts make their contribution and form the overall characteristics of the video chain.

To produce good quality images at a certain scan rate, the bandwidth of the video chain must be high enough to show the finest possible details of the image. The transfer function that describes the frequency or bandwidth and phase characteristics of a detector or an amplifier is very useful to characterize the video chain of an SEM. For good quality TV frequency imaging, 4-5 mHz bandwidth is required, but for slow-speed image collection this bandwidth can be surprisingly low, that is, 10-20 kHz.

Noisy signal means loss or lack of information. At a certain noise level, details that otherwise would be in the video signal cannot be seen. The visual effects of noise are closely tied to the resolution, appearance, contrast, brightness, and other aspects of the image, as well as intensity distribution and other properties of the noise itself. A 40 dB (100-1), signal-to-noise ratio gives a reasonably clean image, but more noise degrades the ability to differentiate two areas of different brightness by eye. 5 To turn an analog signal to digital data, AD conversion is needed. Usually there is a sampling and hold circuitry that keeps the analog signal unchanged for the time of conversion. Depending on the design, this results in smaller or larger signal loss because just a fraction of the analog signal is collected due to the relatively short sampling and long conversion time. Well-designed circuitry, for example, a gated integrator, can improve the signal-to-noise ratio, but the quantization process in the AD converter itself introduces noise also.

Modeling data with Monte Carlo or other techniques has to include the shortcomings of the real, nonideal measuring tool, that is, the SEM with all its associated components. By knowing the transfer function, noise, and distortion figure in digital form, it is relatively easy to obtain more accurate comparison of the measured and calculated signal (Fig. 1 The calculation of image contrast in the scanning electron microscope (SEM) can be done using Monte Carlo techniques if the electron trajectories can be calculated through the composition profiles in the specimen. This has been done for the case of a discrete one-dimensional composition variation in the incident beam direction , but most of the programs that are available are designed to calculate signal intensities from samples of uniform atomic number and density (e.g., Reimer and Stelter 1986) . Nevertheless, the calculation of electron trajectories in a three-dimensional heterogeneous microstructure has not been reported.

The basic idea behind a Monte Carlo calculation is that an electron, in penetrating a solid, will undergo a series of predictable elastic collisions (electron-atom) which change the direction and inelastic collision (electron-electron) which change the energy. In Figure 1 , which is the example of a calculation in a lead tin eutectic with somewhat unrealistic geometry, the electron enters the sample and is scattered to a new coordinate location. Since the directional change and distance scattered are sensitive to the variations of atomic number, atomic weight, and density, the electron might travel to location 2 in the block of the lead-rich position of the PbSn eutectic, scatter into the position 3, and then go on to position 4 of the tin-rich matrix. Thus, depending upon the details of the microstructure, that is, the size and location of the second phase, the image will appear different.

The anticipated backscattered intensities from this rather unrealistic microstructure are shown, for example, in Figure 2 as the block varies both in size and location below the sample surface. We clearly see that the backscattered signal in this case is much more sensitive to the position in the specimen of the sec- IV-12 Scanning Vol. 16, Supplement IV (1994) FIG . 1 The geometry of the sample and a schematic of the electron trajectory for the calculation. Intensity ond phase rather than the size distribution, but for more realistic microstructural geometries such a calculation would be useful in determining the presence of alternative phases in a complex assemblage. terference microscopy 1-6 is based on the observation that a simple single-arm interferometer can be constructed by allowing light from a laser to be back-reflected from a target and to reenter the laser resonant cavity to produce a modulation of the laser light-intensity. [7] [8] [9] The resultant light modulation is dependent on the optical phase (optical path length traveled) and amplitude of the reentry light. If the target is moved over several wavelengths, the laser light intensity displays a cos(z) dependence; for a mirror target, the light modulation index (I-I o /I+I o ) can be as great as 0.75. These observations were denoted as laser-feedback interferometry (LFI); 9 this phenomenon has been characterized and analyzed several times in the past 30 years. [10] [11] [12] [13] [14] [15] [16] The theory of LFI 2, 11, 17 provides an understanding of the harmonic content of LFI signals; they follow a Bessel-function dependence, a property that is useful in designing electronic feedback circuits necessary to produce a practical laser-feedback microscope. LFI can employ either diode or gas lasers as the interactive laser source/detector. Using a He-Ne laser with a high-reflectivity output coupler mirror (≤ 98%), it is possible to measure surface profiles of highly-reflecting surfaces (e.g., metals or high index-of-refraction materials such as silicon) with axial detail as small as 1 nm and surface vibrations up to 1-2 MHz can be measured down to picometer amplitudes. In addition, weakly backscattering materials (e.g., biological cell surfaces and cell components) give sufficient signals to provide a useful method of high-resolution imaging in biology.

A versatile laser-feedback microscope (PHOEBE) has been constructed having the following properties. The He-Ne (632.8 nm) laser incorporated is a milliwatt, linearly-polarized output unit with special mounting to minimize output mirror movement. Access to the low-power rear laser beam allows the laser intensity to be measured directly by a silicon photodetector.

The main beam is expanded to fill the back aperture of any type of microscope objective (air-, water-, or oil-immersion) suitable for the microscopic examination being undertaken. An x,y scanning stage moves the sample under the laser beam in a raster fashion; 256 pixel × 256 pixel image frames are obtained in < 20 s. Axial (z) motion, furnished by a tubular piezoelectric transducer, provides both a modulation signal input for the electronic feedback circuit (25 kHz) and repositions the sample at each point. The electronic feedback circuit maintains a fixed distance between the output coupler mirror of the laser and a point in the object being imaged. The correction (output) signal of the feedback circuit then gives a measure of the surface topography of the sample. Concomitantly, the laser intensity modulation gives a measure of the surface reflectivity of each point on the sample. These two images are digitized and stored in the computer memory. All microscope control functions have been consolidated in a dedicated small computer allowing samples to be run without interruption by the resetting or realignment of the microscope. PHOEBE scans for biological use range from 10 µm × 10 µm to 120 µm × 120 µm. When an air objective is used, surface topography images display a quantitative measure of height; for fluid-immersion objectives, the measured heights are shortened by a factor dependent on the index-of-refraction of the fluid (water or oil).

The confocal property of this scanning microscope is due to the requirement that only backscattered light reentering the cavity-resonator mode (TEM oo ) of the laser is effective; the through-focus response of LFM verifies this property in comparison with a confocal pinhole placed at the focus of the laserbeam expanding lens. The lateral resolution of PHOEBE is 200 nm as determined by imaging silicon-based resolution standards.

An important result of measurements with the PHOEBE LFM is that high-contrast images can be obtained from biological samples placed on a plastic or glass substrate (e.g., a microscope slide or Petri dish). In contrast to images of samples with sharply defined index-of-refraction boundaries, what is measured in this case is a point-by-point optical path length difference with the substrate furnishing the majority of the back-reflected, LFI-measured light. Other imaging modes include the use of interferometric optical sectioning and the buildup of three-dimensional structures from a series of twodimensional sections. Because of the coherence requirement of LFM, the use of fluorescent labels and the detection of fluorescence is precluded; however, other reflective labels may be employed to gain back some of the advantages of that method. -25, 1968-1972 (1989) 17. Sarid D, Iams D, Weissenberger V, Bell LS: Compact scanningforce microscope using a laser diode. Opt Lett 13, 1057 Lett 13, -1059 Lett 13, (1988 

University of California, Irvine, California, USA NMR microscopy is one of the newly emerging tools for the high resolution three-dimensional (3-D) imaging of live animals and plants for biological as well as medical research. More recent applications and developments include such things as porous materials and microflow, essential for oil research. One of the main interests of NMR microscopy, however, lies in the fact that the method is truly a noninvasive 3-D high resolution imaging tool with which µm resolution can be achieved.

However, NMR imaging, especially NMR microscopy, has a number of formidable difficulties, namely, small signal-tonoise ratio due to the inherently small object size, diffusion and bandwidth limitations, and other inhomogeneity effects such as chemical shifts and susceptibility.

Among others, diffusion problems due to the random Brownian motion of (water) molecules appear to be one of the fundamental physical limitations to NMR imaging, especially in high-resolution microscopy where these molecular diffusion distances are close to the resolution limits. Although the diffusion effect in NMR has been studied extensively, the effect on NMR microscopy has not as yet been observed experimentally due to the current limitations of the NMR microscope. Theory, however, suggests substantial resolution broadening in NMR microscopy if molecular self-diffusion prevails, especially when we deal with molecules with large diffusion coefficients. Resolution limits on NMR microscopy especially include diffusion limits, namely, effects of phase variation due to molecular self-diffusion during data acquisition, effects of signal attenuation and related line broadening, and some diffusion effects of molecules which, when confined in boundaries or walls, exhibit anomalies in resolution observation in NMR microscopy. The first two are based on free-water molecules, while the third is based on the model of water molecules confined by boundaries, which have significant physical consequences. Diffusion effects limit resolution in microscopic imaging, as does the phase factor which is related to the finite bandwidth of the imaging instrumentation: that is, the available gradient strength and acquisition time, both of which are intimately related to the diffusion effect.

An X-ray microtomographic system is being developed at our laboratories AMIL & ARTS. A generalized Feldkamp cone-beam reconstruction algorithm was already developed for our system (Wang et al. 1993) . The generalized algorithm is approximate, but quite accurate and computationally efficient. Under some feasible conditions, the generalized algorithm produces exact volumetric reconstruction for longitudinal invariant specimens and exact transaxial reconstruction for a point source contained in that transaxial plane. In the generalized Feldkamp cone-beam reconstruction, a transaxial slice is reconstructed using projection data collected from a scanning turn of 360˚ angular range.

In fan-beam reconstruction, there actually are two complete sets of projection data over a full-scan range. Exact reconstruction can be achieved using only projection data corresponding to a half-scan. In our cone-beam system configuration, it can be appreciated that there are "approximate redundancies" in data acquired along geometric rays that would be identical after projection onto a transaxial plane.

There are various weight functions for half-scan fan-beam reconstruction. With Gullberg and Zeng's weight function, a half-scan generalized Feldkamp cone-beam algorithm is obtained for less longitudinal blurring:

(1) where (2) ∆ is an additional angle for a smooth transition between essential and duplicated Radon regions, and φ(z) is an offset specific to the longitudinal coordinate z. Optionally, cone-beam projection values associated with appropriate pairs of opposite rays can be lineraly interpolated to synthesize needed fan-beam projections for transaxial reconstruction. At an additional computational cost, the interpolation-based cone-beam algorithm allows exact reconstruction if the longitudinal specimen variation is linear. Numerical simulation results demonstrate the feasibility of our extended algorithms.

Wang G, Lin TH, Cheng PC, Shinozaki DM: A general cone-beam reconstruction algorithm. IEEE Trans Med Imag 12 (3), 486-496 (1993) Biological specimens can be preserved by rapid freezing, a process which takes just a few milliseconds and is termed "cryofixation." It is an alternative to chemical fixatives which may take many minutes to penetrate a specimen and even longer to fully stabilise cellular components, thereby compromising their ultrastructural and chemical integrity.

When cryofixation is performed properly, the water molecules in the specimen do not have time to form ice crystals and the specimen is preserved in a near life-like state. The time-scale of cryofixation means that short-lived phenomena can be captured and preserved. When this aspect is combined with electrical or chemical stimulation in a controlled experiment, then cellular processes can be studied by freezing the experiment at predetermined time intervals, so that a series of transient steps are preserved.

Besides electrical and chemical stimulation, other methods have been combined with cryofixation; these have involved electrophoresis, electroporation, temperature-jump, and flash photolysis.

Electrically stimulated muscle was slam-frozen by Sjöström et al. (1973) and Van Harreveld et al. (1974) . Their methods held specimens in defined physiologic states: True time-resolved freezing was introduced by Heuser et al. (1979) , who showed that ultrastructural differences could be observed in synaptic events at neuromuscular junctions between 3 and 5 ms after stimulation.

Chemical stimulation is effectively performed by a chemical flow method, involving the rapid mixing of small specimens with a stimulant from different syringes and flowing them for a predetermined time along a reaction tube before spraying the reactants into a coolant, thus quenching the reaction (Knoll et al. 1991 , Rand et al. 1985 .

Electrophoresis involves the movement of particles in an electrical field and was used with freeze fracture to study the diffusion rate of intramembrane particles (Sowers and Hackenbrock 1981) . Electroporation uses a radio frequency field to induce transient pores in cell membranes (Chang and Reese 1990) . Optical stimulation has been used to cause a temperature jump to investigate temperature effects, for example, a 59˚C jump after 450 ms exposure to a xenon lamp, on the ultrastructure of lipid specimens (Chestnut et al. 1992) . Flash photolysis methods have been used to release caged photolabile chemicals in the study of the dynamics of the actomysin cycle (Funatsu et al. 1993 , Ménétret et al. 1991 .

When the available stimulation methods are considered with the available freezing methods (namely, slam, plunge, jet, and microdroplet-spray freezing), then it becomes clear that there is promising scope for the study of dynamic cellular processes using electron microscopy. The combined methodology integrates the temporal resolution of rapid freezing with the spatial resolution of the electron microscope.

Chang DC, Reese TS: Changes in membrane structure induced by electroporation as revealed by rapid-freezing electron microscopy. Biophys J 58, 1-12 (1990 189-198 (1974) 

Universität des Saarlandes, Homburg-Saar, Germany

Contrary to standard preparation at ambient temperature, there is some confusion on what is optimal in cryopreparation of biological specimens for subsequent electron microscopy (EM) investigation. This results mainly from the diversity of methods as well as instruments described and the rapid development of these techniques in the preceding years. Since the big advantages of cryopreparation in various fields are often hidden by the choice of unsuitable or even antiquated techniques, it seems to be justified to report about the recent developments in this field and to classify different methods for different purposes. This tutorial is based mainly on experience in transmission electron microscopy (TEM). Nevertheless, most of the information may also be useful for scanning electron microscopy, since a proper preparation has the same importance in both fields.

Without doubt, CF is the most important first step in most cryopreparations with a big variety of different techniques. The simplest procedure (plunging or immersion cryofixation = ICF) is best suited for vitrification of thin suspension layers ("bare grid" or "ice embedding") for subsequent cryo-TEM in the frozen hydrated state. High pressure freezing (HPF) reduces ice segregation in suitable specimens (e.g., thin plant leaves or rigid tissues such as cartilage). The main problem of HPF in soft tissues results from the indispensable dividing of the specimens into small pieces. Impact freezing on a metal mirror (MMF) is well suited for suspensions such as HPF or double jet (DJF), but MMF has an advantage for soft tissues, since it is suitable for larger specimens, that is, tissue slices > 50 mm 2 . As far as liquid cryogens are concerned (ICF, DJF), LN 2 or partially frozen N 2 -slush are not to be recommended; ethane gives the best results. Propane is well suited and less expensive for routine applications.

As an alternative to cryosectioning, these more time-consuming preparations gain continuously in importance for cytochemistry. Even element analysis is possible in certain cases. Only for simple morphology or morphometry additives (e.g., OsO4, UO2-acetate, aldehydes) are short drying times in freeze substitution (FS) or freeze drying (FD) possible. Proper drying by FS or FD without additives seems only possible for small objects (diameter < 0.5 mm) in periods between 5-10 days at −80˚ to −60˚C. Otherwise, severe artifacts by thermal collapse phenomena and redistribution effects result. A suitable instrumentation is of great importance for this long drying time. The results are excellent and the advantages are striking as long as sufficiently long drying times are employed. In addition, low-temperature embedding (LTE) in special acrylics improves the results. Care has to be taken in handling these monomeric resins, since some of them are strong allergens.

Sugar protected specimens are easy to section on the dry knife at −80 to −120(−140)˚C for subsequent histochemistry of macromolecular components. Overall morphologic preservation is rather poor in comparison with FS/FD/LTE. Larger areas are mostly not obtainable, but speedy work and results within hours instead of days or weeks are possible. Sectioning can be considerably improved by the use of cryo-diamond knives together with an ionizer (useless without antistatic tool).

Sectioning below −120˚C down to −180˚C is possible with cryo-diamond knives on an ionizer, if the specimen is well frozen (no ice segregation) and the cutting area << 0.1 × 0.1 mm 2 . Trimming with a diamond trimming tool is advantageous. Good preconditions are given after HPF or MMF. The advantage of MMF is the mirror-like, well-frozen surface, which can be easily trimmed and orientated to the knife edge. Cryotransfer to the TEM/STEM or proper FD for FEDX pose no severe problems (FD for > 10 h at −80˚C).

Most of the different actual cryomethods (except fresh frozen cryosectioning < −160˚C for cryo-TEM/STEM) allow routine work, if the most suitable method is carefully selected and modern instrumentation is available. In all cases, the additional effort and investigation are definitely justified by better and more reliable results (literature and reprints on request).

PAUL WALTHER, RENÉ HERMANN, MARTIN MÜLLER Laboratory of EM 1, Department of Cell Biology, ETH Zurich, Zurich, Switzerland

The reason for using low temperatures for preparation and microscopy is the decreased mobility of atoms and molecules reducing the danger of artefact formation. For scanning electron microscopy (SEM) cryotechniques have the following advantages. 1. Cryofixation (rapid freezing) is the fastest way to immobilize a biological sample at a defined physiologic state. Thereby, all processes in a cell are arrested within milliseconds. Chemical fixation, in contrast, takes seconds or minutes to act, leading to unpredictable osmotic effects and redistribution of cellular compounds. 2. A frozen biological sample behaves like any other bulk specimen, and redistribution of substances is almost excluded. Inner structures can be made amenable to the electron beam by cryofracturing or cryosectioning. 3. The conductive metal layers have a finer grain size when applied to cold samples, because diffusion of the metal atoms on the surface is reduced. 4. The frozen sample is analyzed in the SEM by use of a cold stage. This prevents volume changes due to drying artifacts. In addition, hydrocarbon contamination due to irradiation by the electron beam is greatly reduced at cold temperatures.

For most applications in biology, cryotechniques are only used for some but not for all preparation steps; for example, samples are cryofixed and then dehydrated by either freezedrying or freeze-substitution and afterward stored and imaged at ambient temperature. For high resolution SEM, it is advantageous to coat also dry samples at cold temperatures in order to obtain a finer grain size and to observe the samples at cold temperatures in the SEM to reduce hydrocarbon contamination.

(1) A major limitation of cryotechniques is distortion of the ultrastructure due to ice crystal formation during freezing. High pressure freezing allows for direct cryofixation of living samples within minimal or no ice crystal artefacts up to a thickness of several 100 µm. (2) The problems of water vapour contaminants that condense on the sample during preparation and microscopy were well investigated for the TEM freeze-etching technique in the 1960s and 1970s. During the last years, this knowledge has been adapted for the construction of cryo preparation systems for the low-temperature SEM.

(1) Drying artefacts such as shrinkage are omitted by observing fully hydrated frozen samples. On the other hand, the ice covers many structures of interest and therefore often needs to be partially removed, either by freeze-drying or by freezesubstitution. However, removal of water bears the risk of drying artefacts. (2) Hydrated samples are extremely sensitive to the electron beam. The high surface-to-volume ratio inherent in particulate samples, and the fact that we already live in a sea of possibly contaminating particles, necessitates special care and understanding in particle preparation. Selection of reagents and handling practices can be critical. For instance, in a hypothetical case, a particle analyst reports that his sample consists of approximately five major inorganic phases. The first has a rounded morphology with crystalline overgrowths, the light element content of the second is depleted, the crystalline structure of the next is damaged, the fourth occurs at a 5% volume level, and the fifth consists of glass shards. In this example, these results are essentially worthless. Phase one was soluble in a liquid used in the preparation process and then reprecipitated upon evaporation. The light elements of the second were leached because of the use of the same solvent. The third was attacked by HF formation during ultrasonification in a fluorinated solvent. The fourth was originally present at a 10% volume level, but was preferentially attracted to the sides of the storage container prior to sample preparation, biasing the sampling; and the fifth is not part of the original sample, but comes from the ground glass neck of one of the reagent bottles.

Decisions on reagents and methods are crucial but unfortunately cannot satisfy every concern. For instance, reagents chosen for inorganic preparation should be nonpolar when easily leached elements such as boron or lithium are of interest, or if surface oxidation or ionic dissolution is a concern. But nonpolar solvents can increase particle agglomeration problems due to the lack of charge dissipation. Therefore, wise choices in preparation methods are strongly tied to the objectives of the analyst, and sometimes multiple methods performed in parallel with additional analyses may be required to obtain truly representative results.

Contamination often is a concern, especially when performing high resolution work on a trace particle constituent. Suspended urban air particulate is typically 20-30 µg/m 3 , mostly in the 0.01-0.1 µm range and usually with less than 10 particles per cubic centimeter above 1 µm. 1 Supermicrometer-sized par-ticles can constitute 40 µg/m 3 . 2 These particles may settle out gravitationally, electrostatically, or may simply happen to intercept the surface of the sample mount or sample particles. A large background can accumulate on unprotected substrates. The use of easily charged containers such as polystyrene disposable petri dishes or polyethylene centrifugation cones can further complicate dust collection or an additional problem, sample losses, due to electrostatic forces.

The use of light microscopes for particle preparation makes possible a variety of preparation methods including micromanipulation. 3 Uses include ensuring even particle distributions on mounts, precise particle positioning, particle size reduction, washing unwanted films or residues off of particles, and many more. A light microscope can also be used to make some helpful observations such as specific gravity, population densities, refractive index, morphology, size, solubility, film thickness, and color.

"Micro world" effects often have to be provided for when performing preparation on a microscale with a microscope. For example, the polarity of solvents sometimes must be sacrificed for lower evaporation rates. Reduction in evaporation rate can be aided by the use of small transparent covers, air movement shields, or by the choice of a combination of substrate and solvent that have poor affinity for one another (the resulting bead-ing-up action decreases the surface-to-volume ratio). The insertion of a "reservoir object" such as a probe tip at 45°"holding" a liquid droplet against the substrate, or a coverslip from the edge of which a supply of underlying liquid can be obtained, are examples of creative procedures that may be used (Fig. 1) .Recommended polar solvents include heptane and cyclohexane. These avoid HFC environmental concerns, evaporate readily, and avoid halogen interactions. Isoamyl acetate, flexible collodion, and formvar in ethylene dichloride are preferred micromanipulation and mounting media. Microscale fume "hoods" protect microscope optical coatings and personnel. For containers, glass or static dissipator-coated plastics are preferred to ordinary uncoated plastics. If clean rooms are not available, static dissipators (physical or chemical), laminar flow work benches, single HEPA-filter curtained areas, and modified work practices are inexpensive solutions. Syringe filters help provide easy to handle point-of-use contamination control for reagents. Recommended micromanipulation probes include tungsten and specific animal hairs. Intermediate substrate surfaces may employ temporary teflon or paraffin coatings. Other procedures include ashing, centrifugation, filtration, ultrasonification, and the exploitation of effects observed during micromanipulation.

An approach for the indirect visualization of biological material in transmission electron microscopy (TEM) is the freezeetch or freeze-fracture/replica method. 1 The preparations steps ( Fig. 1 ) of this purely physical method include: (a) fast fixation and stabilization by quick freezing (in ms vs. min necessary for conventional chemical fixation); (b) creation of clean fracture faces with a fracturing cryotome (high vacuum required); (c) replication with electron beam evaporation (EBE, high vacuum required); (d) three-dimensional imaging of fracture faces and surfaces (with structural resolution of 3-5 nm); (e) high stability of the Pt-C replicas in the electron beam (selection of relevant details) and possible long-term storage (weeks to months). The important drawbacks are that the replicas have to be cleaned (subsequent changing of several cleaning solutions over hours or days), picked up (disintegration of highly structured replicas happens frequently), and can be viewed only as a result of one single fracturing event. Specimens providing IV-18 Scanning Vol. 16, Supplement IV (1994) FIG highly redundant structures and relatively smooth fractures, such as cell suspensions or o/w emulsions, were investigated using freeze fracture/replication and ambient temperature transmission electron microscopy (AT-TEM). 2 Freeze-drying is comparable to freeze-etching ( Fig. 1) , but with the important difference that all the water is removed from the specimen by sublimation. In TEM, scanning electron microscopy (SEM), or scanning tunnelling microscopy (STM), small cell components or particles, such as viruses, can be imaged together with the replication layer on top. 3 In freeze-fracturing for low-temperature SEM (LT-SEM), most advantages of TEM cryopreparation are kept. The preparation includes: (a) quick freezing; (b) fracturing under high vacuum conditions (c) coating with a planar magnetron sputter source (PMS, Ar as working gas); (d) 3-D imaging and observation of changed conditions (in situ etching); (e) high signal/noise ratio of Pt or W coatings for secondary electron imaging and low yield from Cr for backscattered electron detection of immuno-gold labelled specimens. The specimens can be imaged directly eventually after multiple fracturing (i.e., searching of distinct structures) in a different situation (i.e., fully hydrated or partly freeze-dried, coated or uncoated), in a "close to nature state." 4 However, new problems have to be solved in order to get thin reproducible, conductive coatings, without superimposition of specimen and coating film structures and direct imaging, reducing contamination and beam damage, 5 or long-term low temperature storage. Moreover, the coating film thickness, the ice in the frozen bulk specimen, and, most important of all, the type of SEM are now limiting the structural resolution. The biological relevant details of results obtained with low-temperature SEM in a conventional fieldemission SEM can be compared with those achieved with routine freeze fracture/replication and TEM. 6 According to Honig and Hook, 7 the physics of water in high vacuum play an ultimate role in the application of these techniques (Fig. 2) ; therefore, only this knowledge in combination with cold stages and high vacuum technology enable the application of state of the art preparations of biological material for electron microscopy.

The cryo-jet freezing technique 1 provides a means for the in situ preservation of diffusible ions in mouse spinal cord explants after trauma. Using the technique, the effects of trauma, in particular the intracellular shifts of calcium and other diffusible ions, can be analyzed and recorded by qualitative and quantitative electron probe microanalysis (EPMA). Chemical fixation must be avoided if one intends to attempt the EPMA localization of diffusible substances in cells and tissues. The goal of the cryo-jet method is to confine axonal components and chemical elements to a biologically natural position. Trau- matized mouse embryo cord explants were compared with the intact rat spinal cord trauma model, 2 using EPMA studies of perturbations of calcium and other ions that are reported to be located in the axonal cytoskeleton. 3 ICR mouse embryo spinal cord explants were grown to 14-21-day maturation in Maximow chambers. Cords that survived stringent elimination examination criteria and exhibited minimal degeneration or spontaneous necrosis were randomly divided into four groups. Group I served as experimental controls; Group II was traumatized by dropping a 25 mg weight through a 4.5 cm glass pipette positioned directly over the culture and cryojetted 1 h after trauma. Representative explants from I and II were freeze-substituted and embedded in epoxy for comparison. By light microscopic examination of the whole mount, living culture, the site of impact could be determined within 10-30 min. The impact zone became markedly dense when compared with the surrounding, uninjured tissue. Toluidine blue of the freeze-substituted sections revealed the site of impact on the surface and path of the trauma damage through the explant. Control cultures with spontaneous necrosis could be distinguished from the experimental group, because the traumatic necrosis was confluent and often even wedge-shaped. Within the traumatized zone, nerve fiber alterations similar to those observed in in vivo spinal cord trauma and calcium toxicity could be found, that is, granular degeneration of axoplasm, pleomorphic spheroids, tubulovesicular profiles, and myelin showing adaxonal, periaxonal, and intramyelinic vacuolization. Light microscopy sections from the freeze-substituted explants were noted to be well preserved at the surface and into the culture for 100-200 µm (Fig. 1) ; however, some ice crystal damage was noted deep within the body of the culture. Previous studies that compared metal mirror and plunge freezing results with the propane jet methodology reported here confirmed that propane jetting is the method of choice for preserving organotypic spinal cord tissue cultures.

X-ray microanalysis of cryosections of the cytoskeletal tubules, cross bridge structures, mitochondria, and the myelin sheath of in vitro spinal cord explants, compared with the in vivo model of spinal cord trauma, has yielded unique data concerning the ionic changes that occur as a result of trauma or experimental manipulation.

After trauma, populations of axons were found that could be distinguished by their elemental compositions. One group was similar in elemental composition to controls, although the morphology was dissimilar, while a second group had 80-90% lower K and 5-10-fold elevated Ca (Fig. 2) . The response of axons and the associated myelin was paired, that is, loss of K and gain of Na within the axon was always accompanied by comparable changes in the myelin and vice versa. The morphologic appearance of axons was not a predictor of the elemental composition. The cardiomyopathic (CM) hamster manifests cardiac dysfunction from an early age, with the disease ultimately progressing to congestive heart failure. On the basis of significant elevation in the Ca 2+ content of the CM hearts, as well as significantly increased Ca 2+ in the mitochondrial (MT) fractions and an increased density of voltage-sensitive Ca 2+ channels, previous studies suggested a Ca 2+ overload in the CM hearts (Sen et al. 1990 ). Evidence has also been presented for an increased sensitivity of cardiac muscle cells of CM hamster hearts to an external Ca 2+ "stress," that is, any manipulation which increases Ca 2+ influx into the cells (Hano and Lakatta 1991, Sen et al. 1990 ). On the other hand, ultrastructural studies have shown localized focal lesions from an early age, believed to be due to local cell injury. The cardiac myocytes within these areas suffer irreversible damage, becoming Ca 2+ overloaded as demonstrated by measurements of MT and A-band (AB) Ca 2+ content by electron probe microanalysis (EPMA) in a previous study from our laboratory (Bond et al. 1989 ). This study also demonstrated that the vast majority of cardiac myocytes throughout the CM hearts not only had a normal ultrastructural appearance, but also showed no elevation of subcellular Ca 2+ content.

An alternative explanation for impaired contractile function may be a decreased amount or availability of Ca 2+ stored in the sarcoplasmic reticulum (SR) for stimulated release and activation of contraction. To investigate this question, we have utilized EPMA to measure Ca 2+ content directly in the junctional SR (jSR), as well as in MT and AB of the CM hamster heart, either under control conditions or, alternatively, after pretreatment with the Ca 2+ channel agonist, BAY K8644, in order to increase Ca 2+ influx into the cardiac muscle cells.

Isolated papillary muscles from normal and CM hearts at 110 days of age were stimulated to contract electrically, and parameters of isometric contraction were recorded at L max . Muscles were randomly assigned to one of three protocols: (1) Eleven CM and 12 normal papillary muscles were used to construct dose/response curves to the Ca 2+ channel agonist, Bay K. After stabilization at L max , cumulative doses of Bay K (10 -9 M to 10 -5 M) were added to the muscle bath. All contractile parameters were recorded from the muscle at each dose. (2) For EPMA, five CM muscles were frozen, after stabilization at L max , at peak +dT/dt and five muscles during relaxation. (3) Ten CM papillary muscles were incubated with single dose of 10 -5 M Bay K in order to elicit a maximal inotropic effect. Once the contractile response to drug addition had stabilized, the muscles were frozen either at peak +dT/dt (n=5) or during relaxation (n=5). Ultrathin cryosections were cut from the surface of the frozen muscles and freeze-dried overnight. Subcellular Ca 2+ content (AB, MT, and jSR) of CM muscles was measured by EPMA in STEM mode in a Philips CM12 scanning transmission electron microscope (Fig.1) .

Measurements of baseline contractile function revealed a significant decrease in develop tension (DT) (from 0.61 ± 0.08 g in CM muscles to 0.40 ± 0.04 g in normals), +dT/dt (8.09 ± 0.91 g/s to 5.35 ± 0.56 g/s) and a decrease in −dT/dt (4.56 ± 0.53 g/s to 3.23 ± 0.37 g/s) in CM versus normal hamster. There was no significant difference in values of resting tension (RT). The inotropic response to increasing doses of Bay K was markedly blunted in the CM muscles compared with controls, suggesting that even when Ca 2+ entry into the cardiac muscle cells is increased, force development is still impaired.

A comparison of elemental content (Na, Mg, P, S, Cl, K, and Ca) of AB and MT between experimental groups revealed no statistically significant differences. In addition, no differences in elemental composition of AB and MT were observed compared with our previous measurement on normal hamsters frozen during contraction and relaxation (Moravec and Bond 1991) .

The amount of Ca 2+ stored in the SR of the CM muscles that were rapidly frozen during relaxation (in absence of Bay K) was 5.1 ± 0.5 mmol/kg dry weight ( Fig. 2 ) (left panel, REL); however treatment with the Ca 2+ channel agonist Bay K 8644 significantly increased the size of the store to 6.8 ± 0.6 mmol/kg dry weight (right panel, REL). Thus we conclude that Ca 2+ uptake into the SR of CM muscles was enhanced as a result of Bay K treatment. The contractile data show that the amount of Ca 2+ that can be released from the jSR of CM muscles during a cardiac twitch is very small (equivalent to the difference between the relaxed and contracted values) in "control" (untreated CM hamsters, left panel); however, treatment with Bay K increases the SR Ca 2+ load in the relaxed muscle, with little demonstrable effect on the amount of Ca 2+ remaining in the SR at peak +dT/dt (right panel, CONT), resulting in a significant increase in the amount of releasable Ca 2+ in the SR store. The total Ca 2+ content measured in the relaxed Bay K treated muscles was, nevertheless, considerably less than previously measured in normal papillary muscles rapidly frozen during relaxation.

In summary, these data suggest that (1) a Ca 2+ deficit, as opposed to a Ca 2+ excess or Ca 2+ overload, may be an important factor contributing to the cardiac dysfunction in CM hamsters, and (2) that, specifically, impaired Ca 2+ regulation by the SR may result in this Ca 2+ deficit. The application of automated scanning electron microscopy (SEM) to the analysis of particulate populations is in many ways a unique application of microanalytical instrumentation. When applied to the analysis of large numbers of particles, the SEM is used as both a microanalytical and a macroanalytical instrument. As a microanalytical instrument, the SEM provides single-particle compositional and morphologic information that is not available from the conventional macroanalytical techniques used in particle analysis such as atomic absorption and instrumental neutron activation analysis. As a macroanalytical instrument, the SEM, when used for automated particle analysis, provides information from which population characteristics can be inferred. For example, automated SEM information often is used to determine the percent of an aerosol that originates from a given source by extrapolating the single-particle results (e.g., the number of particles containing major Fe) to the entire particle population, that is, the air filter. This value is then extrapolated to the sampled air volume.

While providing the analyst with a wealth of information, the dual role of the SEM in automated particle analysis has several limitations that must be considered when doing an analysis. These limitations involve, among other things, the spatial dispersion of the particle sample; particle size distribution; analytical parameters such as magnification, accelerating voltage and electron dose; and the algorithms for determining particle composition and particle groupings.

At NIST we have been conducting a series of experiments to study the limitations associated with the quantitative elemental analysis of particles during an automated run. Specifically, we have been evaluating the relationship of measured x-ray intensity to the accuracy, precision, and detection limits of automated x-ray analyses. Analytical accuracy, precision, and detection will have a pronounced effect on the ability to separate particles with similar but different elemental compositions into groups. For this experiment, we developed an analytical glass series containing six glasses with varying amounts of uranium and lead ( Table I) . Samples of the different glasses were prepared as bulk-polished specimens and as particles. The results from the analyses of the bulk samples represented the "best case" that could be expected under a given set of analytical conditions since there were no particle effects involved. All analyses were done on an electron probe and a SEM at 15 keV and 1 nA beam current. Dead times for the bulk and particle analyses were between 15 and 20%. Two separate counting times were selected for the experiment , 200 s and 15 s. These times resulted in the x-ray peak intensities for Pb and U M xrays that are shown in Table II . Glass K-3111 was used as the standard for the quantitative analyses of the different runs. For the particles, both bulk and particle forms of K-3111 were used as standards. The concentration of oxygen was determined by stiochiometry.

The results of the experimental runs on the bulk and particle forms of the glasses are shown in Figure 1a and b as plots of the U versus Pb concentrations in wt.% (normalized to 100% total analysis) for the 200 s data. The bulk plot shows a complete separation of the six different glasses, while the particle results show an overlap of adjacent glasses even at the 200 s count time. The 15 s data show a much stronger overlap between adjacent glasses for both bulk and particle morphologies.

The analytical data were also processed with a clusteranalysis algorithm to determine the average silhouette width (ASW) for both bulk and particle forms of the glasses at the different counting times, (Fig. 2 ) (Kaufman and Rousseeuw 1988). The ASW is a robust measure of the cluster strength with a value approaching 1 representing the maximum association among cluster members. Since there are six glasses in the series, the ASW for six clusters should be the highest. Of all the different runs on both bulk glasses and particles, only the 200 s data on the bulk glasses has the highest ASW for six clusters.

The results of this study indicate that the uncertainties associated with the shorter counting times in ASEM analysis may severely limit the ability to distinguish correctly between similar groups of materials. In addition, the uncertainties associated with particle analyses are considerably greater than those from bulk analyses due to absorption and mass effects. These greater uncertainties for particles underscore the need to define the strengths and limitations of ASEM analysis and to design automated methods that will maximize the information that can be obtained from a sample.

Botany Department, University of Georgia, Athens, Georgia, USA The Monoblepharidales (Chytridiomycetes) produce asexual motile reproductive structures known as zoospores through the cleavage of cytoplasm in the zoosporangium. Although different aspects of zoosporogenesis have been studied in a number of zoosporic fungi, little is known about the events of zoospore formation in the sporangium of the Monoblepharidales, or the Chytridiomycetes in general. Earlier studies of other zoosporic fungi proposed various spore formation events that have been challenged recently (Hyde et al. 1991) , especially with respect to the relationship of vesicles, golgi, and cleavage furrows. Hyde et al. (1991) concluded that all eukaryotic cleavage events may need reinvestigation.

We are studying the various aspects of zoosporogenesis in Monoblepharella using both chemical fixation and cryofixation methods to determine whether the cleavage events in this Chytridiomycete are similar to those found in the study by Hyde et al. (1991) . Questions that remain include how the nuclei accumulate in the sporangium, how the cytoplasmic domains are established, and how each zoospore obtains its usual complement of cellular components.

Mycelia were induced to form sporangia by transferring the cultures to distilled water. Sporangia were chemically fixed at different stages of development using a sequential aldehyde/ osmium fixation protocol. The resulting tissue was embedded in Araldite/Embed 812. Freeze substitution fixation was also used for comparison of vesicle and cleavage furrow profiles. When the tissue was to be used for cytochemical localizations, osmium was omitted and LR White was used as the embedding medium for both fixation protocols. Various dyes, lectins, and antibodies were used to localize organelles within the sporangium.

Nuclei were close to the plasma membrane as they moved from the subtending hyphal strand to the swelling tip of the developing sporangium. The centrioles were closely associated with the nuclei and oriented toward the periphery of the sporangium. Nuclei then moved a short distance from the centriole toward the center of the sporangium. Microtubules, originating from the centriolar region ( Fig. 1 ), were seen around the nucleus and extended into the cytoplasm, possibly delimiting the boundary of the forming zoospore. Membranes forming the cleavage furrows appeared around the forming flagella near the kinetosome (Fig. 2 ) and also at specific sites along the periphery of the sporangium. The furrows continue to expand at both sites as extending sheets of membrane. Dictyosomes with large numbers of vesicles were closely associated with the nuclei, suggesting a source of the membrane needed for furrow extension. ER was first seen in large sheets in the central region of the sporangium. Later the strands of ER surrounded the nuclei prior to ribosomal aggregation. IV-24 Scanning Vol. 16, Supplement IV (1994) FIG. 1 Nucleus (N) with associated centriole (arrow) and microtubules (arrowheads) at periphery of sporangium.

FIG. 2 Developing cleavage furrow and flagellum (F). Note coated region of furrow (arrowhead) and radiating microtubules (arrows).

Hyde GJ, Lancelle S, Hepler PK, Hardham AR: Freeze substitution reveals a new model for sporangial cleavage in Phytophthora, a result with implications for cytokinesis in other eukaryotes. J Cell Sci 100, 735 (1991)

Center of Ultrastructural Research, Barrow Hall, University of Georgia, Athens, Georgia, USA Cryptocaryon irritans, a parasitic ciliate of many species of seawater fishes, has a complex life cycle consisting of feeding, resting, dividing, and infective stages. The disease, termed white spot disease, can cause extreme fish loss in marine aquaria and mariculture environments. Cryptocaryon irritans attaches to the epidermis of the fish and feeds on epidermal cells. This form of the ciliate termed the "trophont" appears as large white spots on the fish. The trophonts grow in size and eventually leave the host. The free-living trophont settles down to the substrate and secretes a cyst wall. While in this resting stage known as the "tomont," the cell undergoes a series of unequal palintomic divisions to form daughter cells called "tomites." The cyst wall ruptures and free swimming ciliates known as "theronts" are released. The theronts represent the infective stage as they search for new hosts.

Different procedures were used to prepare tomite, theront, tomont, trophont, and cyst stages. Organisms were fixed with glutaraldehyde and osmium tetroxide for transmission electron microscopy (TEM). Parduc's fixation (OsO 4 +saturated HgCl) was applied for the scanning electron microscopic (SEM) work. The parasite's cytoskelatal framework was stained by using indirect immunofluorescence technique. For this purpose monoclonal anti-α-tubulin was used as a primary antibody, and rhodamine-labeled goat antimouse IgG as a secondary antibody. Fluorescently labeled cells were examined by laser scanning confocal microscopy.

The trophont has an elongate body shape with a broad anterior end, a tapered posterior end, and is completely covered with somatic cilia. The cytostome is apically located and trophonts often were observed moving with their mouth part leading. Kineties were arranged in a parallel fashion along the longitudinal axis of the cell and terminated in a ring around the cytostome. There is no oral membrane at the oral region. There are cirri-like structures around the oral opening as described by Cheung et al. (1981) . These cilia are shorter (3-4 µm) and wider than somatic cilia. The cytopharynx is surrounded by ridges or oral ribs (nonciliated lining). Large bundles of microtubules support the oral region.

The theronts are oval to teardrop-shaped and completely covered with cilia. They have a ventral mouth with a slit-type structure in the middle. The cytostome covers almost 1/3-1/4 of the body. The cirri-like structures are found around the mouth and are similar to those in trophonts. The oral ribs are present but they are layered. Based on SEM and TEM results, it was found that there were some structural differences in the cytostomes of trophonts and theronts. The mouth probably does not become functional until the theront has entered a host. The short and stiff cilia seem to be used for burrowing and the gathering of food particles. The role of the pellicle and cyto-plasm is discussed in relation to penetration into fish epidermis.

We found that there are mucocysts in both trophont and theront, but not in tomont. Theront mucocysts are concentrated around the slit type structure of the mouth. The secretion of mucocysts might contain enzymes that help the theront to penetrate into fish epidermis. They also might aid trophonts in feeding on fish tissue and have a function in cyst wall formation at the later stage. Penetration into the fish tissue probably is started by mucocyst secretion that either enables the parasite to stick to the tissues or to help it enter the tissue by enzymatic reaction over the irritated areas. The theront may then utilize its relatively stiff oral cilia for burrowing into the these irritated areas. After burrowing into the epidermis of fish, the theront develops its oral apparatus and increases its size. Trophonts eventually leave the host when they reach a certain size (350-450 µm).

This study is supported in part by the National Aquarium, Baltimore, Md. Over the 20 years that the FBI has practiced SEM/EDXA, the technology has grown in importance to be considered an essential tool for investigative forensic exams. Because of the wide variety of applications, there is a corresponding vast array of preparation methods and analytical techniques. Analytic techniques and sample preparation methods are inextricably linked, and although many are routinely applied, often only imaginative approaches serve to fulfill the desired analytic result.

The scanning electron microscope (SEM) practitioner is expected to be knowledgeable about the applications and proficient at the methods of preparation in order to utilize the SEM to its greatest advantage. The main types of analysis practiced at the FBI include (1) visualization of structure, including surface features and internal structure, (2) inorganic elemental characterization, (3) particle analysis, and most recently (4) the use of a compositional data base for identification and association.

SEM is a powerful complement to light microscopy (LM) for low magnification morphology characterization and is unsurpassed for applications requiring greater depth of field than are available with LM. Toolmark and fracture exams can be enhanced by stereo analysis. Preparation can be minimal, and several electronic signals (BEI, SEI) are available.

The preparation of cross sections permits the study and comparison of heterogeneous materials. Embedment usually is necessary to support the object during cross sectioning. Hard materials generally are polished by an adaption of metallurgical polishing methods, and soft materials generally are microtomed. Sectioning often is possible by manual methods, without the use of a microtome. This method can reveal complex structures such as plating layers and document laminates.

The most routinely applied application of EDXA is the qualitative exam. It is part of the inorganic analysis scheme for material characterization. The qualitative exam frequently is combined with elemental distribution mapping to provide "compositional pictures."

Particle populations often are indicative of an environment and can be used to associate an item or individual with an activ-ity. Too small for individual manipulation, they are most easily sampled by adhesive lift. Additional methods involving separation and concentration often are effective.

Compositional characteristics of materials are stored in a database to permit comparison and identification of a questioned material. Standard spectra are collected and a specific peak for each element is integrated above background and ratioed to the sum peaks from all elements. This value representing % x-ray counts is stored in a "periodic table" data base including standard information. In addition, the original spectrum is stored on disk and a hard copy is filed. The standard files include metal alloys, building materials, paints, tape adhesives, fingerprint powders, and cosmetics. The reference list can be queried for comparison to an unknown for alloy matching, identification of an unknown material, or manufacturer identification. The effectiveness of this method depends upon the compositional uniqueness of the material and the variation of composition within the class of materials to which it belongs. This project is in its infancy, with only several hundred entries to date. Data entry currently is manual, although software currently is being developed to extract required data from spectra automatically and to export it to the database.

The need for a vehicle for information exchange has been expressed within the community of SEM users in crime labs.

Since an electronic medium such as Internet was not feasible because most forensic laboratories are not electronically linked, a "newsletter" was produced to link laboratories involved with SEM in forensic science and related areas, as well as individuals in industry and academia. Timely and informal, it augments the professional publications and attempts to bring practical methods, reprints from obscure journals, translations from foreign publications, and questions/answers directly to the user.

Hamilton County Coroner's Laboratory, Cincinnati, Ohio, USA Many examinations in the crime lab involve comparing questioned material from a suspect to known material from a victim. By characterizing the material it may be possible to establish a link between the suspect and victim. Our Trace Evidence Section characterizes materials by using a combination of analytical instruments. The infrared microspectrometer is used to determine the organic constituents, and the scanning electron microscope-energy dispersive x-ray spectrometer (SEM/EDXA) is used to analyze the inorganic composition. This approach is routinely applied to paint particles because paint formulations include both organic and inorganic constituents. Analysis by SEM/EDXA is very valuable when more than one layer is present in the particle. The instrument can then be used in line scan mode to analyze each layer individually without separation.

In addition to comparisons, the SEM/EDXA is used to identify materials. In bombing cases it may be difficult to identify the explosive as well as other components. Large amounts of potassium and sulfur in residues from a pipe bomb indicate black powder as an explosive. If chlorine is also present then "Pyrodex," a black powder substitute, may have been used.

Other applications, such as matching small fractures, make use of the superb imaging capabilities of the SEM. The capability of maintaining excellent depth of field at high magnifications is particularly important when matching the ends of wires that have been pulled apart. This is exactly what was done in a case of tape players that were jerked out of victims' vehicles. Small fractured surfaces also have been encountered in the investigation of hit-and-run accidents when pieces of chrome trim were knocked loose and left at the scene.

Clearly modern instrumental means of analysis, such as the SEM/EDXA, are critical to the work of the forensic scientist. The increased sensitivity of the instrumentation, however, may raise questions as to the relevance of the evidence found. If a single, very small paint particle is found on the jeans of a pedestrian struck in a hit-and-run accident, could the particle be from the striking vehicle, or merely from the roadway debris at the scene? Such questions indicate that increases in instrument sensitivity require sensitivity on the part of the analyst to questions of contamination and weight of evidence.

Research Division, Office of Laboratories and Scientific Services, U.S. Customs Service, Washington, D.C., USA The U.S. Customs Service laboratory system provides a variety of analytic services to control the commerce and assure enforcement of numerous regulations at the border. Any item which is imported into the United States may need to be analyzed to determine the answer to any number of questions. What is the item? Is the item correctly described? Does the item infringe upon a U.S. patent? The scanning electron microscope (SEM) and EDS X-Ray system can be utilized to answer some of the questions which arise. Several examples follow.

A U.S. company holds a patent on a feature incorporated into an electronically programmable read-only memory (EPROM) cell. They allege to the International Trade Commission (ITC) that another company is incorporating this patented feature in their EPROMs without the patent holder's permission. They win their case and the ITC issues an exclusion order. It is at this point that the U.S. Customs Service becomes involved. We are charged with enforcement of the exclusion order. The incorporated feature is exceedingly small on a visible scale. How will we know which shipments of EPROMs should be excluded from entry into the U.S.? Now the ability of the SEM to produce images easily at very high magnifications comes into play. With the help of the SEM, the laboratories are able to determine if the infringing feature is present or not.

Another U.S. company holds a patent for a denim (textile) finishing process. They bring a complaint before the ITC that their patent is being infringed upon. The complaint is upheld by the ITC and an exclusion order is issued. The U.S. Customs Service laboratory system must now develop a method to differentiate among various finishing processes in use. Research at the headquarters laboratory found that a combination of imaging with a stereomicroscope and an SEM could make the differentiation. The SEM samples were Au-Pd sputtercoated with a Hummer VI A Sputtering System (Anatech Ltd.). The sputtercoated denim samples clearly showed distinctive features not easily seen with an optical microscope.

A sample purported to be eelskin was submitted to the headquarters laboratory for conformation of its identity. It was thought that the product might be embossed plastic or possibly leather of mammalian origin. The top surface of the sample was Au-Pd sputtercoated and then examined with the SEM. The features displayed by this examination were convincing evidence that the sample was indeed leather. A cross section of the sample was then prepared and Au-Pd sputtercoated. The SEM images showed cell patterns consistent with reptilian or marine origin.

Another sample arrived courtesy of a foreign customs service. Their inspectors had seized a large statue of a Roman gladiator and his horse and chariot. The original reason for suspicion was that the declared value for the statue was much higher than one would expect for an object of this type; it appeared to be a cheap plastic statue. It had been dismantled and tested for the presence of drugs. The results were negative. Using our Princeton Gamma-Tech EDS X-Ray system, an elemental analysis of the exterior covering of the portion of the statue we received was performed. It showed the presence of large amounts of silver. This would explain the high declared value of the statue.

When the U.S. Customs Service ran tests on narcotics particle detection systems, it was noted that sampling for heroin was more difficult than sampling for cocaine. A brief study of samples of each narcotic using the SEM showed that in general the average particle size for heroin is less than that of cocaine. This may help to account for the sampling difficulty for heroin.

ALLAN N. WALTERS U.S. Postal Inspection Service, Forensic Laboratory, Dulles, Virginia, USA Currently, the two main applications of SEM/EDAX are explosive residue analysis and alloy quantitation.

Most explosives encountered are from improvised explosive devices (IEDs) and commonly are low explosives such as black powder, smokeless powder, Pyrodex, and flash powder. Pyrodex and smokeless powder are commercially manufactured, while black and flash powder may be either of commercial or improvised (homemade) manufacture.

The following table illustrates the composition of the common low explosives: Device components are placed in the SEM, and EDAX is performed before disturbing the residue. The resulting elemental profile is then used to guide further analyses with other instrumentation such as XRD, FTIR, HPLC, TLC, and chemical spot tests.

Sputter coating of samples is not performed so as to avoid modifying the sample and interfering with subsequent analyses and examinations.

Quantitative analysis is performed on alloys which are of interest to the U.S. Postal Service Engineering and Development Center and is required to ensure that the materials meet the re-quired specifications. Samples have included lock bodies, lock springs, keys, and lock tumblers.

Reverse engineering using quantitative analysis of current collectors for mobile electrification systems (mail sorting machines) has been conducted.

Quantitations are performed using a standardless method. Scanned probe microscopy has evolved significantly over the last 10 years. Beginning with the first commercial scanning tunneling microscopes (STM) and continuing through the sophisticated "multifunction" microscopes of today, the "probe" has been one of the most critical and, in some instances, the least understood component of these systems. Depending on the type and geometry of the sample, the properties of a probe can be optimized to reduce imaging artifacts. Examples of this will be given using two well-known techniques, scanning tunneling microscopy (STM) and atomic force microscopy (AFM). Also, new innovations in probes for these two techniques (and others) will be presented.

Scanning tunneling microscopy, as first introduced, used a probe constructed of a chemically sharpened tungsten wire. 1 Later, it was found that a suitable probe could be formed from a mechanically sharpened wire (usually Pt/Ir). Both of these probes proved that they could produce self-consistent images under specific sample and environmental conditions. However, problems arose with the tungsten probe while imaging in air because of the formation of native oxide. In addition, the mechanically formed probe tended to produce significant imaging artifacts if used on samples with topography > 100 nm. In many cases the distortion produced by these artifacts made the data uninterpretable.

A solution to these problems was to use a chemically etched probe (to control the shape) made from an inert material with physical properties suitable for the samples involved. Musselman et al. developed the techniques necessary to chemically etch Pt/Ir wire into tips with a controlled geometry. These probes were capable of imaging surface structures greater than 1 µm peak-to-valley depths with significantly reduced tip-related artifacts. 2 This controlled geometry shape also made it more advantageous for use as a coated tip for electrochemical imaging.

Still, the aspect ratio of this particular probe was not suitable when imaging high-aspect ratio features such as pits in optical discs, contact vias in ICs, fracture surfaces, etc. To obtain images from these types of structures, it was necessary to "machine" (in a controlled way) the probe described above. By using a focused ion beam (FIB) as a machining tool, it was possible to create a probe to image the above features ( Fig. 1) . 3 This FIB "nanomachining" technique has opened up many possibilities for specialty probes.

The majority of atomic force microscopy still utilizes the basic silicon nitride (Si 3 N 4 ) triangular cantilever and pyramidal probe combination. 4 The base of the pyramid is on the order of 5 µm with the sidewalls extending upward at an approximate 55 o angle to the apex. Again, for samples with features < 100 nm, such as mica, these probes have provided very good image repeatability. In general, however, for structures much greater than 100 nm, a convolution of the probe and the sample surface will again occur. These probes can be modified, using the FIB technique, to increase their aspect ratio significantly. Because of their "hollow" design, these modified probes are still limited to topographies of ≤ 0.3 µm.

Most of the advances in probe manufacturing (on a wafer level) have come about by utilizing silicon as the probe material. Several silicon probe types are now available, which provide significantly sharper tips with aspect ratios on the order of 5 to 1. 5 For imaging structures of higher aspect ratios, such as vias or deep trenches, longer and thinner probes are needed. These are made possible by using a combination of FIB and electron beam techniques to "grow" a thin probe using an existing Si 3 N 4 pyramidal probe as the base (Fig. 2 ). 6 These probes are available in lengths up to 2 µm with aspect ratios of ≥ 10 to 1. Utilizing a combination of FIB milling of existing structures and electron beam growth, probes with lengths of 8-10 µm are feasible. An obvious problem with probes of this type (and with any long, thin structure) is "flexing" during imaging. Analysis of one such probes' physical characteristics has shown that the elastic modulus is relatively low (E~1.5GPa), while the coefficient of friction on most surfaces is extremely low (µ<0.05). Thus, probes of this type should be kept as short as possible while exceeding the maximum peak-to-valley distance to be imaged. 7 Other scanned probe techniques now in the prototype phase include thermal, magnetic force, near field optical, and probes capable of imaging undercut sidewalls. Probes for all of these methods have been shown to be feasible, although manufacturing techniques to provide large quantities, reliably and repeatably, have yet to be developed. When imaging with an atomic force microscope (AFM), the image resolution is a complex function of the relative tip and sample geometries. When imaging or measuring high-aspect ratio features, sharp and slender tips offer the possibility of probing down into extremely small topographical features. The most commonly used contact mode AFM tips are batchfabricated Si 3 N 4 thin-film cantilevers with an integrated pyramidal structure used as the tip. 1 It has been shown that microtips, which are fabricated by electron beam-induced growth of carbonaceous material on the apex of the pyramid, can reduce the artifacts associated with integrated pyramidal AFM tips. 2 graph of a whisker of electron beam-grown contamination, or microtip, grown on the apex of an integrated pyramid, is shown in Figure 1 . An obvious problem with the use of a long slender microtip is the lateral deflection of the microtip as it is scanned across the sample surface. The proper use and interpretation of artifacts associated with electron beam-grown microtips demands an understanding of the mechanics of microtip deflection. It has been observed that long, slender microtips scanned over flat surfaces (rms roughness of < 1 nm) produce hysteresis in the fast-scan direction. A model has been developed to explain the observed hysteresis loop in terms of a mechanical cantilever beam deflecting because of lateral frictional forces induced by repulsive imaging forces. This model is shown schematically in Figure 2 . By applying cantilever beam mechanics to the model of microtip deflection, a method of calculating the elastic modulus of microtip material has been developed. To find the elastic modulus of microtip material, a series of experimentally determined microtip deflection distances and the respective microtip lengths are required. Microtip deflection distances have been experimentally determined for 18 different microtip lengths on two different flat surfaces; fusion deposited boro-silicate glass and polished silicon. The elastic modulus of the microtip material has been determined from the deflection data to be approximately 1.4 GPa. Once the elastic modulus has been determined, the coefficient of friction between microtip material and a sample sur-face can be calculated. The coefficient of friction between a microtip and the sample surface will indicate if the sample material is suitable for AFM imaging with electron beam-grown microtips.

The elastic modulus of microtip material and the coefficient of friction data lead to a better understanding not only of microtips but of electron beam-induced contamination in general. The low elastic modulus rules out the possibility of the material being diamond-like and suggests a polymeric material. The use of microtips can greatly improve image resolution; however, it is important to note that since microtip deflection increases with increasing microtip length, the microtip used to image a sample surface should be as short as possible while remaining long enough to image the largest peak-valley structure on the sample surface. The magnitude of observed microtip deflection should be reduced substantially by the use of AC mode microscopes where surface friction is less of a concern. Because of their ability to achieve high resolution simultaneously in all three dimensions in a wide range of ambient conditions, scanning probe microscopes are promising candidates for performing measurements of surface topography. Crosssectional and perspective views can be generated, nondestructively, at any location once an image has been acquired. Surface topography measurements fall into two basic classes: position (or pitch) measurement and size (or critical dimension) measurement. The ability of a microscope to perform position measurement depends more on the quality of its design and construction than on the fundamental interaction of probing beam or stylus with the sample. Size measurement, on the other hand, depends strongly on the probe-sample interaction. Modern manufacturing, especially semiconductor lithography, often produces high-aspect ratio, submicron structures whose size and shape must be known with tiny uncertainties. Surprisingly, stylus profilometers in the guise of scanning probe microscopes can perform some of these measurements at a level unmatched by any other type of microscope. To obtain size and shape measurements in semiconductor manufacture, we have developed a scanning probe microscope with several refinements, depicted in the Figure. 1 We use capacitance-based sensors for probe force sensing IV-30 Scanning Vol. 16, Supplement IV (1994) Cantilever and microtip at rest and for probe position measurement. 2, 3 Many of the samples that we scan are at least partially electrically insulating. For this reason, our microscope is used primarily as a scanning force microscope, although it is capable of operating as a tunneling microscope also. Our force sensor employs a small silicon beam that pivots in one dimension about a pair of magnetically constrained ball bearings. The beam forms a pair of capacitors that both sense the position of the beam and maintain its balance with a suitable servo loop. This force-balance technique allows high-force sensitivity without sacrificing the stiffness required to resist surface forces. Capacitors are also used to measure the position of the probe tip. The piezoceramic tube used as a scan actuator exhibits strong hysteresis and creep, so the drive voltage is an unreliable measure of probe position. The capacitors monitor the probe position in all three dimensions during the scan, and these data are collected along with the topograph.

The most important factor determining the quality of the measurements is the shape of the probe tip. Geometry alone makes the probe-sample interaction strongly nonlinear. In surface roughness measurements, a blunt probe can severely limit the range of spatial frequencies that can be detected. In scans of high-aspect ratio features, the probe shape determines what parts of the features can be measured. When scanning deep trenches and holes on a patterned surface, we use either a conical probe or a cylindrical probe, depending of what part of the feature is most important. The conical probes are made using focused ion beam sputtering of iridium. 4 A chemical etch is used to form the cylindrical probes. If the probe shape is well known, then it is possible to determine what parts of a scan were distorted by the probe tip and, in some cases, this distortion can be removed. 5, 6 The probe microscope itself can be used to determine the probe tip shape if a suitable structure is available for probe characterization. 7 Since the pioneering work of Binnig and Rohrer in the early 1980s on scanning tunneling microscopy (STM), the STM has evolved into a powerful tool for spectroscopy, metrology, electrochemistry, and nanolithography. Many other instruments have also evolved from the STM technology under the family of scanning probe microscopes (SPMs) with applications in atomic force, electric potentiometry, and magnetic force imaging.

In the magnetic force microscopy (MFM) mode, the technique has been applied to the imaging of magnetic bit patterns in recording media (Grütter et al. 1990 , Mamin et al. 1988 ) and the mapping of static and dynamic magnetic fields of recording heads (Martin and Wickramasinghe 1987) . MFM is typically performed in the noncontact atomic force microscopy (AFM) mode using a silicon cantilever which is coated with a thin film magnetic material, usually Co, Ni-Fe alloy, or Co-Pt-Cr alloy. The force exerted on the magnetic tip by stray fields from the sample causes the deflection of the cantilever which is subsequently measured. Using a novel variation of the STM technique with a flexible iron tip, Rice and Moreland (1991) have also performed MFM imaging on magnetic bit patterns on a hard disk in a tunneling-stabilized MFM (TSMFM) mode. Standard MFM images, however, reflect both topographic and magnetic information, with the relative strengths of each signal depending on the tip-to-sample spacing. Using a differential interferometric technique, Schönenberger et al. (1990) have shown that the topographic and magnetic information can be reasonably separated.

In this abstract, results of some applications of the SPM for topographical and magnetic force imaging of magnetic materials are presented. As the critical dimensions of magnetic devices are getting smaller, the surface topography and magnetic morphology of the recording head and media are becoming increasingly important with respect to optimization for best performance. The ability of the SPM to obtain submicron topographical and magnetic information makes this an invaluable metrology and failure analysis tool for the magnetic recording industry. Compared with other techniques for highresolution magnetic imaging, such as Lorentz microscopy, electron holography, scanning electron microscopy with polarization analysis, MFM has the advantages of ease of sample preparation and the ability to operate in air. Typical resolutions of 40-100 nm are obtainable in the MFM mode, although resolutions of 10-25nm have also been achieved with considerable effort. In our work, topographical imaging was performed in the contact mode using a V-shape silicon nitride cantilever of length l = 200 µm, width w = 18 µm, thickness t = 1 µm and force constant k = 0.64 N/m. The sensing tip at the end of the cantilever is pyramidal in shape, with a 4 × 4 µm square base and 1:1 aspect ratio. The tip radius is typically smaller than 500 Å. For magnetic force imaging, the instrument was operated in the noncontact amplitude resonance mode by oscillating the cantilever and measuring changes in its resonant frequency using either phase or amplitude detection. This method of detection is responsive to the force gradient. For each point of the recorded image, the cantilever was first brought to a large tipto-sample distance of 100 nm or greater to acquire the magnetic force image and subsequently moved close to the sample surface to acquire the topography image. This allows the simultaneous acquisition of topography and magnetic force images. A -10 V specimen bias was also applied to provide a linearizing and stabilizing force to the servo feedback loop. The cantilever used for the MFM imaging is a diving boardshape probe made of (100) silicon. The geometry of the cantilever are l = 450 µm, w = 35 µm, t = 6 µm and k = 1 N/m. The sensing tip is conical in shape with a height of about 15 µm, an aspect ratio of 3:1, and a tip radius of < 200 Å. The cantilever, which was sputter-coated with a 600 Å thin-film cobalt material and magnetized inside a 200-400 Gauss solenoid field, has a typical resonant frequency of 72 kHz. Figure 1 shows a three-dimensional topographic profile of a defective thin-film recording head, in which the pole tips protrude slightly from the surface, using the contact AFM mode. The dimensions of the two rectangular pole tips are about 8 µm × 4 µm and 11 µm × 4 µm. The gap width was measured to be about 0.35 µm. Dirt particles can also be seen and these are due to the cleaning process during sample preparation whereby the head was lightly swabbed with a cotton tip soaked in methanol. Figure 2 shows the magnetic force image of the surface of a recorded computer hard disk. The bright and dark lines represent regions of different magnetization or bits on the hard disk surface. The separation between each pair of bright-dark lines was measured to be about 2.3 µm and the track width is about 20 µm. The striations in the topographic image (not shown), representing the texture lines, are approximately perpendicular to the recorded bit patterns. The line profile of the bit patterns in Figure 2 (not shown) is very similar to the calculated force gradient contours by Mamin et al. (1988) in which both the horizontal and vertical magnetization components of the tip are are being sensed. IV-32 Scanning Vol. 16, Supplement IV (1994) FIG. 1 Three-dimensional topographic profile of a defective thin-film recording head using contact AFM, in which the pole tips protrude slightly from the surface. In a good recording head, the pole-tips are supposed to be slightly recessed from the surrounding region. Image resolution in traditional far-field microscopy is limited by diffraction caused by the primary aperture in the optical system of the microscope being used. In practice, diffraction limits lateral resolution to approximately λ/2, or about 0.5 micron for optical microscopes.

To image smaller cellular structures and biological materials which are smaller than 0.5 micron, biologists have used microscopies which "illuminate" the specimen with radiation of smaller wavelength, such as an electron beam (i.e., using an electron microscope).

The nearfield scanning optical microscope (NSOM) breaks the diffraction limit to lateral resolution by illuminating (or collecting light from) the specimen through a sub-wavelength aperture which is held a small fraction of a wavelength above the surface of the specimen. Using this nearfield technique, resolution far better than λ/2 has been demonstrated on biological specimens and resolution better than λ/20 has been demonstrated on standards (Betzig and Trautman 1992) .

With NSOM we have recently obtained optical images of tobacco mosaic viruses (18 nm diameter) and of photoreceptor rod outer segments at a sub-wavelength resolution.

Betzig E, Trautman JK: Near-field optics; microscopy spectroscopy and surface modification beyond the diffraction limit. Science 257, 189 In June 1993, the initial incident involving a report of a syringe found in a canned Pepsi ® product received nationwide publicity. During the following 6 months, law enforcement authorities investigated hundreds of additional claims of tampering from nearly every state in the United States. More than 200 cases involving foreign objects allegedly found in soft drink containers were processed by the National Forensic Chemistry Center (NFCC). To date, the forensic and law enforcement efforts of the FDA have resulted in numerous arrests and convictions for charges related to product tampering and felonious reporting of product tampering. Scanning electron microscopy (SEM) has been valuable, and in several cases crucial, in providing conclusive evidence of fraudulent reports of product tampering.

The items reportedly found inside of suspect cans were primarily hypodermic syringes (with and without needles), but other reported objects included nails, screws, bullets, pins, sewing needles, tacks, glass and plastic shards, and rodent carcasses. Since most of the cases involved syringes, primarily insulin-type syringes, the first analyses were designed to determine if the syringes were contaminated and if they contained human blood, tissue, drug, or insulin residue. Syringe/needle rinses were examined by a number of techniques including SEM, which was used to find any intact blood cells and/or tissue. Energy dispersive x-ray analysis (EDXA) was performed on residue preparations. EDXA was able to detect a small K αl peak for zinc in several syringe rinses which was consistent with EDXA analysis of dried residues containing some types of insulin. Type identification of insulin from syringe rinsing was performed by LC/mass spectrometry in a method developed at NFCC.

In another case, a broken pin was reportedly found in a soft drink can. A cross-sectional analysis of the suspect pin by EDXA showed the pin was a nickel-plated, iron-core straight pin. The absence of observed iron corrosion provided evidence that the object had not been in contact with the soft drink from the time the canned product was produced until the time the pin was allegedly discovered.

The most common question requested by investigators was "How long had the syringe been in the soft drink can?" To answer part of that question, a number of experiments were conducted at NFCC. One SEM/EDXA method investigated the corrosion of the aluminum crimp (found on some syringe needles) submerged in diet cola. New aluminum crimp needles were sealed in diet cola and removed at 24-h intervals. Stereoscopic light microscopy was initially used to examine each needle crimp. It was discovered that the submersion produced a brown to black "corrosion flower" in less than 24 h of soaking. Of specific interest, the submersion repeatedly produced a single point of corrosion on the crimp. The single point of corrosion suggested possible electrolysis of the aluminum crimp in the soft drink. This corrosion point continued to enlarge and penetrate more deeply into the aluminum crimp over time. Backscattered electron (BSE) imaging and x-ray mapping were used to highlight the region of corrosion in the SEM. Figure 1 shows the secondary and backscattered electron image (SEI and BEI) of an aluminum crimp on a syringe needle after 24 h submersion in diet cola. The SEI shows the corrosion of the aluminum crimp as pitting in the crimp surface. The BEI shows the corrosion area as a darker region caused by surface oxides blocking BSE generation from the aluminum below.

The study also demonstrated that the corrosion of the aluminum crimp was dependent upon the amount of time the crimp was submerged in the pressurized soft drink container. Figure 2 shows a point of corrosion on an aluminum crimp after 3 weeks of submersion. At this time interval, several points of corrosion often developed. The pitting of the aluminum mater-ial immediately beneath the corrosion was extensive. SEM/EDXA analyses have shown that the aluminum crimp on syringes submerged in diet cola produced a characteristic single point of corrosion after only 24 h. The corrosion resulted in damage to the crimp surface and the production of a metal oxide. The location of the oxide was identified and plotted by x-ray mapping for oxygen (via EDXA).

The analysis of samples related to the Pepsi "Tampering" cases of 1993 are continuing to date. In many cases, the circumstances and sample condition are unique and generated specific questions which require further forensic research using additional time-related studies, multielement x-ray mapping, and/or image analysis. Confocal microscopy can be used to investigate the properties of rough surfaces. Based on the Kirchhoff approximation (Sheppard et al. 1993 a,b) , the three-dimensional (3-D) confocal image can be modelled using the 3-D coherent transfer function (CTF). The form of the 3-D CTF for confocal reflection and transmission have been derived using a high-angle scalar theory (Sheppard et al. 1994) . In both cases the CTFs can be expressed analytically. Using these allows the profile of the rough surface to be reconstructed.

In many cases we need to know only the statistics of the rough surface, rather than the actual profile. Ways in which the statistical properties can be extracted directly have therefore also been considered (Sheppard 1993 IV-34 Scanning Vol. 16, Supplement IV (1994) FIG . 

A. CASTENHOLZ

In previous in vivo studies based on vital microscopic and fluorescence microscopic techniques it was possible to observe lymph flow in initial lymphatics and regional lymph nodes. [1] [2] [3] [4] As an expression of immunological mechanisms, the studies carried out in the rat tongue under various issues showed some remarkable phenomena such as cell traffic along lymphatic pathways and phagocytotic activity of the lymphatic endothelium. Since these phenomena could not be defined with traditional fluorescence microscopy on the cytological level, confocal laser scanning microscopy (CLSM) was applied to the living tissue and fixed specimens). 5 Moreover, in morphologic studies on the lymphatic and blood vascular system, CLSM has proved a very suitable tool also for the representation of resin-injected tissue, which is commonly processed as corrosion casts for scanning electron microscopy. 6 All these approaches, also including the application of special fluorescent markers, should be outlined here.

In vivo studies with the CLSM have been designed for the functional morphologic analysis under normal conditions and in a state of inflammation and experimental edema. After staining the lymphatic endothelium with dimethyl-carboxfluorescein, CLSM revealed a distinct pattern of the so-labelled initial lymphatics consisting of bright (cytoplasm) and dark zones (nuclear portions). 4 For labelling tissue macrophages passing the lymphatic pathways and other phagocytotic cells, fluorescent microbeads (latex standard particles and liposomes) were applied by interstitial injection. Thus, CLSM enabled a certain identification of moving particle-laden cells and also gave evidence of phagocytotic activity of the endothelium of initial lymphatics. In long-term experiments, phago-cytosis of cells lining the sinuses of lymph nodes could be clearly recognized.

Tissue (tongue, skin) and lymph nodes from different sites of the rat were examined in the CLSM as unfixed or fixed (glutardialdehyde) thick sections. Cells labelled with fluorescent microbeads or stained with fluorescent dyes thus could be easily detected and identified. Thereby, it was possible to determine the location of single beads at the endothelial surface or within the endothelial cytoplasm (Fig. 1 ). If the tissue was conventionally stained with hematoxilin eosin or acridine orange, optimum information could be obtained from unlabelled structures in these specimens as well. By means of the two-detection system, CLSM also was able to distinguish two or more different fractions of microbeads incorporated by macro-phages ( Fig. 2) or lying in tissue spaces as free elements. In hemal lymph nodes of rats, erythrophagocytosis related to sinus macrophages was successfully represented by the CLSM after vital staining of erythrocytes with the fluorescent celllinker PKH 26.

In this approach, Mercox (acrylate) stained with rhodamin and fluorescent yellow was used to create high fluorescence in the lumen of blood or lymphatic vessels. If the resin was injected into the tissue, the interstitial spaces were filled up by resin. CLSM of sections from such resin-injected specimens enabled exact distinction between casts related to the blood or lymphatic system and those of the interstitial spaces, when two differently stained resins were simultaneously injected from the arterial system and into the interstice. CLSM of resin-injected specimens was also applied as a suitable method for the control of corrosion casts of SEM. 6 Proceedings of SCANNING 94/SEEMS 94 IV-35 

Today, CLSM has become an established tool in many fields of biological sciences. Because of its abilities to produce images with optimum resolution and clarity, the application in experimental lymphology is striking as well. Some experiments in techniques related to in vivo experiments, supra vital and fixed tissue, and resin-injected specimens have been reported here. These approaches, also comprising the application of new fluorescent markers for living cells, may demonstrate how wide the spectrum of application is spanned for CLSM in experimental lymphology and immune research.

This study was supported by a grant of DFG (Deutsche Forschungsgemeinschaft, Bonn) 

DANIEL CHIN, PH.D Agouron Institute, La Jolla, California, USA

The regulated metabolism and distribution of nucleic acids is required for endogenous antisense or RNA regulatory systems. Recent interest has focused upon using exogenous agents as antisense therapeutics to treat viral infections or metastatic diseases. Both endogenous and therapeutic antisense regulation of gene expression requires the formation of a hybrid between the antisense molecule and the message or gene sequence. It is not surprising that detection of such hybrids in living cells has not been reported to date.

We have used fluorescence resonance energy transfer (FRET) to study hybrid formation and dissociation after microinjection of oligonucleotides (ODNs) into living cells. Two systems were examined: one system characterized the kinetics of hybrid dissociation of two synthetic ODNs while a second system examined the distribution of hybrid complexes formed by hybridization of injected ODNs with endogenous mRNA. In the first system, a 28-mer phosphodiester ODN (+PD) was synthesized and labeled with a 3′ rhodamine (+PD-R). The complementary, antisense 5′-fluorescein labeled phosphorothioate ODN (-PT-F) was specifically quenched by addition of the +PD-R, as detected by both absorbance and FRET in solution. Rapid and specific hybridization between the ODNs occurred at 17 µM within 1 min and the preformed hybrid slowly dissociated (T1/2 ≈ 3 h) in the presence of a 300-fold excess of the unlabeled ODN. Upon microinjection into the cytoplasm of cells, preformed fluorescent hybrids dissociated with a halftime of 15 min, which is attributed to the degradation of the phosphodiester. Formation of the hybrid from sequentially injected ODNs was detected by FRET transiently in the cytoplasm and later in the cell nucleus, where nearly all injected ODNs accumulate. Diffuse, specific nucleoplasmic FRET could be distinguished from nonspecific FRET in punctate nuclear structures which may result from concentration effects similar to FRET seen in late endosomes or lysosomes between endocytosed, inert FITC-dextran and +PT-R ODN.

A second set of experiments with two adjacent 25-mer PT ODNs complementary to mouse b-actin mRNA was performed in mouse 3T3 fibroblasts or N18 neuroblastomas. The upstream ODN was 5′ labeled with FITC and the downstream ODN was 3′ labeled with rhodamine. When hybridized to synthetic actin mRNA or a complementary 40-mer PD ODN, quenching of FITC and FRET was observed in solution. Injection studies in 3T3 cells showed transient ODN accumulation and FRET in filopodia and extended processes. In N18 neuroblastomas, FITC quenching was dependent upon cytoplasmic compartmentalization. After 30 min, both ODNs accumulated into the nucleus. In summary, these experiments suggest that antisense ODNs can hybridize to intracellular targets in both the cytoplasm and the nucleus. The goal of this work was to increase the sensitivity of the electron backscattering pattern (EBSP) technique by introducing an electron energy filter to increase contrast and improve pattern visibility. Energy filtering previously has been shown to increase contrast in selected area channelling patterns (Joy 1986) . Energy filtering differs from arithmetic background subtraction in signal processing, because filtering before detection has a physical basis and eliminates the unwanted background without increasing the noise component in the signal.

The electrostatic retarding filter used in this experiment was constructed as a cone with five coaxial electrodes. The primary purpose of the electrodes was to create symmetric fields that would not distort the pattern, yet allow only backscattered electrons with the least energy loss to contribute to the pattern. Analysis (courtesy E. Munro, Micro Electron Beam Software, Ltd.) of the electron trajectories indicated that for 11 keV electrons, filtering voltages of up to 10 kV could be used before chromatic aberrations became unacceptable. The angular field-of-view of the filter was approximately 38 o .

Another feature of the detector is the use of a microchannel plate as the initial sensing component. Microchannel plates have high gain and are very sensitive to the low-energy electrons that pass through the retarding field filter. In a configuration similar to that of Venables (1973) , the output of the microchannel plates was proximity-focussed to a phosphor screen that displayed the microdiffraction patterns. The gain of the assembly exceeded a simple phosphor by more than 100 times.

The light output of the assembly was focused on a Kodak Megaplus 1.4 CCD camera using either a macrolens alone or a hybrid fiber optic-lens combination. No vignetting was observed across the field. The digital output of the camera was displayed and analyzed on a Macintosh IIx using NIH Image.

Results were obtained on a variety of materials including single crystal silicon, Al-SiC metal matrix material, alumina, aluminum foil, and 3004 aluminum used in aluminum cans. The high voltage design limited the maximum retarding potential to −10 kV so that meaningful data were obtained with incident electron beam energies between 6 and 10 keV. Filter potentials above 75% of the primary beam accelerating voltage increased the Kikuchi band contrast relative to the background. Figure 1 contains a very low contrast pattern for alumina obtained with essentially no filtering. The effect of filtering to within 1 kV of the beam accelerating voltage is shown in Figure 2 for the same specimen. The best patterns were obtained when the filter voltages were between 80 and 90% of the accelerating potential. Filtering above 90% degraded pattern resolution while less filtering usually produced less contrast. Up to four-fold improvements in the contrast ratio were achieved on some specimens.

The detector also had the ability to obtain energy-filtered backscattered electron images in SEM raster mode. One advantage of this configuration is that the SEM characterization of a given feature and the EBSP data obtained from it can be obtained from the same perspective. Another is that filtered im-ages appeared to have improved surface detail and greater crystallographic contrast. Venables (1973) showed that a detector of this kind could be used for "dark field" SEM imaging.

This work demonstrated the feasibility of employing an energy-filtering detector with high gain to increase the sensitivity of microdiffraction measurements in the SEM. Further work is planned to optimize the filter and improve pattern quality and the range of filtering voltages.

The authors gratefully acknowledge the contributions of Munro Electron Beam Software, Ltd. and of Mark Vaudin at NIST, as well as the support by the National Science Foundation under award number III-9260017, and the Department of Commerce under contract number 50-DKNA-3-00127. 

Submicron-sized elements are of great interest in physics and technology. There exists a variety of methods to produce them by using conventional microelectronics technology as well as new methods for deposition of such elements directly from the desired material onto a substrate. The current onestage maskless techniques, for example, laser-induced CVD or laser-induced etching 1 , do not ensure resolution below 250 nm because of their physical limitations imposed by the laser wavelength. Focused ion beams, which generally allow the formation of submicron-sized elements, often are unacceptable as they can cause radiation damage in the material. Moreover, the cost of the necessary equipment is high. These circumstances stimulate the development of techniques for direct formation of submicron-sized patterns with the necessary geometry from an arbitrarily chosen material directly at substrates based on electron-beam-induced CVD 2 .

The CVD experiments were performed in a TEMSCAN JEM-100CX electron microscope at an accelerating voltage of 100 keV and a beam diameter ~3 nm. The electron microscope was equipped with a specially designed pressure cell and an oil-free system for pumping and reagent vapor inlet. This allows to maintain a well-defined atmosphere of chemical compounds (metal carbonyls and halides, freons) around the specimen. Two electronically controlled needle valves are capable of supplying two different reagents simultaneously into the specimen holder. The sample itself is placed inside a heater which provides temperatures up to 250˚C required for the local electron-beam-induced etching. Contacts are made to the sample, which allow in situ measurement of its electrical characteristics. The holder is supplied with two differential apertures (through which the electron beam passes) placed above and beneath the sample. These apertures keep the gas pressure around the sample up to 1 Pa without breaking the vacuum in the microscope column. W(CO) 6 and Rez(CO) 10 were used as reagents. Structure and composition analysis both were performed in a JEM-2000FX electron microscope equipped with a Link AN-10/95S system for EDX analysis.

Self-supporting rods of 50-100 nm in width, containing tungsten or rhenium, respectively, were grown up to 3000 nm long at a speed of the electron-beam movement 1-4 nm/s and at a vapor pressure near the sample of 0.1 Pa. Provided a constant speed of the beam is maintained, the thickness of the produced rods was found to be inversely proportional to the beam speed but decreasing towards the rod's end. This result is similar to that obtained by electron-beam-induced fabrication of self-supporting carbon-containing rods 3 Electron microscopic observation of the inner structure of these self-supporting metal-containing rods revealed many individual fibers mutually aligned in parallel. Thus, the mor- IV-38 Scanning Vol. 16, Supplement IV (1994) FIG. 1 TEM image (a) and SAD pattern (b) from a self-supporting tungsten containing rod section after annealing in vacuum at 400˚C for 15 min. The accelerating voltage is 200 kV, the diffraction length on the (b) is 130 cm. phology of these rods is similar to that of the well-known carbon rods 4 . This observation suggests similar formation mechanisms independent from the rod material. Selected area diffraction (SAD) patterns show the prepared rods as being amorphous. In situ annealing of the tungsten-containing rods at 400˚C for 15 min in the heating holder inside the microscope column transforms the amorphous structure into a nanocrystalline one (Fig. 1) . SAD patterns reveal the presence of a set of different phases: in addition to pure crystalline tungsten there are various tungsten-containing compounds. Among them, in particular, two new cubic phases with lattice constants 696 ± 3 pm and 587 ± 3 pm could be identified. After annealing chemical and phase compositions of the rods grown from W(CO) 6 and Rez(CO) 10 were found to differ depending on the conditions of their formation (reagent vapor pressure, electron current, beam speed). The rods of nano-sized width, grown from Rez(CO) 10 are considered to be promising as tips for scanning tunneling microscopes because of their chemical stability. An important requirement for developing high-speed highpower devices is fabrication of thermally stable ohmic contacts with low resistance and smooth interfaces. Metal semiconductor interface inhomogeneities, such as lateral interface phases and spiking protrusions, lead to nonuniform current flow and consumption of the GaAs substrate. This type of interface morphology is not acceptable for device applications where a large electric field or shallow contact is required. In particular, devices such as a heterojunction bipolar transistor (HBT) cannot tolerate contacts with lateral and vertical interface inhomogeneities. Recently, we have introduced a novel metallization scheme, Pt/Ti/Ge/Pd, which yields thermally stable and low resistance ohmic contacts to both n and p + -GaAs (Han et. al. 1993) .To insure processing reproducibility and contact reliability, one must identify and understand the mechanisms responsible for this contact's electrical per-formance. This study employed cross-sectional transmission electron microscopy (TEM), Auger electron spectroscopy (AES), and electrical measurements (transmission line mode) to investigate the structural, chemical, and electrical properties of this contact. The interface morphology, phase composition, and elemental diffusion were examined and correlated with the measured contact resistances at annealing temperatures which yielded the best slight degradation and the worst electrical performance.

Annealing at 450˚C yielded the lowest specific contact resistance, ~6.4 × 10 -7 Ω-cm 2 . The metal-semiconductor interface was planar and structurally abrupt. The Pd and Ge reacted to form a PdGe phase. Directly beneath the PdGe was a thin, discontinuous, Ga-rich Pd-Ga-As ternary phase. The presence of this Ga-rich ternary compound has important implications for contact formation on the n-GaAs substrate. Formation of this interface phase creates excess Ga vacancies in the GaAs substrate. Upon heating, Ge diffuses into the GaAs, occupying Ga vacancies at dopant levels, to form an n + surface layer, thus allowing considerable tunneling at the metal-semiconductor interface. The existence of the Ga vacancies is crucial, because otherwise Ge would not readily diffuse into the n-GaAs to form this n + -GaAs layer. For the p + -GaAs, this Ge indiffusion occurs also, but the Ge concentration level is not such that it has a detrimental effect, that is, it does not approach that of carbon (the p-type dopant) ~5 × 10 19 cm -3 . The Ti/Pt layers remained stable, hence no surface degradation was present.

Annealing at 550˚C resulted in a slightly higher specific contact resistance, ~2.1 × 10 -6 Ω-cm 2 . There was significant elemental diffusion within the contact metal and minor elemental diffusion into the substrate. The interface exhibited a roughness on the order of 10 nm and possessed large areas where spiking single-phased PdGeGa protrusions spatially dominated the metal-semiconductor interface region. The nonplanar nature of the interface lends itself to several explanations: (1) nonuniformity in the heating associated with the annealing process, (2) material defects or pitting in the GaAs substrate such that the contact metallization covered and filled the pitted area as it would a flat GaAs surface, and/or (3) new phases being formed as a result of the Ti layer beginning to break down as a diffusion barrier. Apparently, this interface nonuniformity (spiking) does not cause significant deterioration in the specific contact resistance, since the contact resistance maintains reasonable electrical integrity. It is speculated that the compositional uniformity of the protrusion phase, which is close to that of the PdGe phase formed during the 450˚C anneal, is responsible for this behavior.

Annealing at 600˚C proved to have a detrimental effect on the specific contact resistance, ~10 -4 Ω-cm 2 . This degradation was accompanied by strong chemical intermixing between the contact and the substrate, resulting in laterally continuous and vertically deep (>100 nm) multiphased protrusions spiking into the GaAs substrate. The surface of this contact possesses surface anomalies (~2.5 µm in size) with a density of ~10%. The anomalies are somewhat enriched in Pt and As and depleted in Ge. This nonuniform surface morphology demonstrates that Pt no longer represents a smooth surface for bonding.

Our results demonstrate that annealing temperatures between 450-550˚C are of practical interest for HBT device processing. The thin base region (1000 Å) and narrow distance between the contact and emitter (0.2-1.0 µm) make the 600˚C anneal impractical for the HBT. Specifically, the composition, extent, and magnitude of the interface spiking would be totally detrimental to this device design, that is, spiking areas would penetrate through the base region, into the collector, and short the device. The present paper is a review of diagnostic availabilities as well as of physical data taken by color cathodoluminescence scanning electron microscopy (CCL-SEM) used for investigation of structural, polytypic, and impurity inhomogeneities of SiC crystals, epitaxial layers, and devices (Saparin and Obyden 1988) .

As examined objects, the "alpha" and "beta" SiC crystals, different polytypes (6H, 4H, 3C, 21R) SiC epitaxial layers and devices have been studied. The epitaxial layers were grown by the sublimation "sandwich" method (Vodakov et al. 1979) in the vacuum or Ar-media under temperatures between 1700 and 2500˚C.

From results of experimental studies it is possible to enumerate the following that are impossible to prove by other techniques: (1) Space distribution of impurities; (2) static and dynamic characterization of polytype transformation. Multitransformation of polytypes during the growth process was observed by way 21R → 3C → 4H. It was concluded that the growth conditions of epitaxial layers and surface perfection of the substrate markedly influence polytype variations. (3) Space distribution of radiation defects on dependence of annealing temperature; (4) action of mechanical defects on the spatial distribution of luminescence centers; (5) availability to observe the transient layers in SiC devices; (6) 3-D analysis of epitaxial layers with polytype transformation.

Saparin GV, Obyden SK: Colour display of video information in scanning electron microscopy: Principles and applications to physics, geology, soil science, biology, and medicine. Scanning 10, 87-106 The examination of chemical vapour deposition (CVD) diamond films by the scanning electron microscope (SEM) (secondary electron mode with spatial resolution 5-10 nm) shows that the morphologies of films are strongly affected by the synthesis conditions, especially the substrate temperature, the methane concentration, total pressure in the deposition chamber, etc. Studies demonstrate that the changes of CH4 for concentrations in the range of 0.2-10% vol. create the shape variation of crystals determined by the ratio of the apparent growth rates of the (111) and (100) faces R(111)/R(100) in cubo-octaedrons interval from 0.577-1.732. The exact cubooctahedral shape observed very frequently is determined by the value of ratio 1.1547 and preferable orientation of the (111)-or (100) faces. Thus, the secondary electron mode of the SEM is a very useful diagnostic technique for diamond films. Less frequently researchers use the second diagnostic availability of the SEM (Saparin and Obyden 1994) : Cathodoluminescence (CL) mode (spatial resolution 0.2-0.5 mkm) with monochromatic and panchromatic (real color) images and local CL spectra. Diagnostic possibilities of this mode identify the quality of diamond films in comparison with natural diamond. The CL spectra of undoped epitaxial films show the dependence on the face upon which the epitaxy was done. CL discriminates between different types of diamond. There are distinct spectral differences between natural, synthetic diamonds and CVD films. The CL emission of the A-band was associated with donor-acceptor pair recombination (blue and green regions); vacancy-rich material emits in the red spectral region (575-633 nm). We would like to note that the CL spectra of natural diamonds (types Ia, Ib, IIa, and IIb) lie in the range of 420-460 nm. Variations in these CL spectra were attributed to differences in the defect structures formed during the growth of material. CL images and local spectra allow one to recognize, for doped and undoped films, the small amounts of impurities, such as Al(30ppm) and B(0.3ppm) and nitrogen remaining in the gas, which could be a possible source of donors. Thus, the SEM investigations of morphology, crystallinity, and CL emission of films led to the correlative estima-tion of diamond film quality in comparison with the natural diamond as standard. The etching process of a crystal surface during its dissolution is well-known. The process of etch pits formation, as a result of crystals dissolution, usually is explained by scientists by the presence of dislocations inside the crystal structure. However, the same process occurring near the crystal edges is studied less, although such investigations can give additional scientific knowledge for a better understanding of the dissolution mechanisms of solids. This brief communication continues the series of investigations (Dorozhkin 1993 , Dorozhkin et al. 1992 The results show that here can be various surface structures taking place under geometric interaction among the growing etch pits and dissolving FAP crystal surfaces. As the FAP has a hexagonal crystal structure, the growing etch pits mainly have a hexagonal shape also (Dorozhkin 1993 , Dorozhkin et al. 1992 ). On the other hand, the investigated FAP crystals were obtained from natural FAP-containing rock after its disintegration and concentration stages. Therefore, the crystals had a very irregular shape and a lot of dislocations inside. Figure 1 shows an example of a typical hexagonal etch pit having only three faces instead of the necessary six; three lacking faces (left side) have already been dissolved by the acid. A very similar situation is presented in Figure 2 . This is another example of a former hexagonal etch pit having only three faces (center).

It is easy to restore events which occurred with the pits some minutes before. The pits began to form and grow from the moment when the outputs of dislocations on the FAP crystal surface began to interact with the acid solution. As the pits are always faced only by fast dissolving faces, their dissolution rate is greater than the one for the crystal surface in common. On the other hand, the growing of etch pits and the FAP crystal surface dissolution are always occurring simultaneously. As a result, a layer of FAP substance between the growing pit, which is situated close to any vicinal dissolving crystal face, and the vicinal crystal face itself were getting increasingly thinner. Finally, this layer disappeared completely, resulting in the pits' formation having only three faces (Figs. 1, 2) . If one had dissolved the FAP crystal for a few more seconds, its surface would have reached the bottom of the pits and they would have disappeared entirely. Figure 3 shows an example of interaction between etch pits with an irregular shape and a surface of the FAP crystal. For this purpose a very thin and sharp spallation fragment of the FAP crystal was chosen which was then etched by phosphoric acid solution. Inasmuch as the spallation fragment was equal to a random and unknown crystallographic face, the irregular etch pits were obtained. It is easy to see four different states of such pits (Fig. 3) : they point to a pit which (1) (4) is only running through the FAP crystal but is situated relatively far from the dissolving crystal face. Finally it should be noted that all the above-mentioned cases of interaction among pits and crystal surface may only take place when directions of dislocations inside the dissolving crystals are close to lines parallel to any nearest vicinal crystal face.

K. HABIB AND P. G. CACERES* Materials Applications Department; *Central Analytical Laboratory KISR, Safat, Kuwait A fundamental study of Co-based metallic glasses has been conducted. The study focused on understanding the changes of the properties and structures of an Fe-B-Si glass as a function of Co, Co-Ni, Co-Mn, and Co-Ni-Mo additions. The separate addition of Co, Co-Ni,Co-Mn, and Co-Ni-Mo elements was successful in a way four new metallic glasses were produced. The compositions of the new glasses are Fe66 Co18 B15 Si, Co66 Fe4 Ni B14 Si15, Co76 Fe2 Mn4 B12 Si6, and Co69 Fe4

Ni Mo2 B12 Si12. Consequently, an evaluation of the physical and magnetic properties was determined.

Furthermore, the internal and surface structures of the glasses have been characterized by a transmission electron microscope and a scanning tunneling microscope, respectively. A comparison between the internal and surface structures of the glasses was carried out on both amorphous and crystallined forms. As a result, a correlation between the properties and structures of the glasses is established.

For instance, Figures 1a and 2a show surface structures of the Co69 Fe4 Ni Mo2 B12 Si12 metallic glass in the amorphous and annealed crystalline forms, respectively. On the other hand, Figures 1b and 2b show the corresponding x-ray diffraction patterns of the amorphous (Fig. 1a) and the annealed structures (Fig. 2a) , respectively. Figures 1a and 2a are basically three dimensional line plots of the surface profile. It is clear that the amorphous structure (Fig. 1a) represents a complete rough surface along 200 nm × 200 nm scanning area. In contrary, the annealed structure (Fig. 2a) exhibits a surface profile with less surface roughness than the annealed structure. This observation is in agreement with work done by the author on other metallic glasses cited elsewhere (Habib et al. 1992 ). The current status of external funding for most academic and research facilities throughout this country is meager at best. Many institutions are being forced to seek financial support from sources other than the conventional governmental agencies, private funding, or contractual agreements. This facility has, for the past 15 years, derived the bulk of its operational budget through third-party payments for services rendered as hospital charges for diagnostic pathology services. While we have been fairly successful in maintaining a relatively consistent level of service, the overall cost of this operation continues to rise. As cost containment became the buzz word and hospital admissions declined, the number of requests for diagnostic procedures also began to decline. This pattern was observed not only in the electron microscopy (EM) lab but in many other specialized laboratories throughout the hospital and our affiliated clinics. As the health care debate accelerated and new concepts such as HMOs, emergency care clinics, managed competition, and regional alliances all geared to assure "more health care for the dollar came into being," it became apparent that an entire new concept was needed to provide high-quality diagnostic services for these newer group practices and smaller clinics that were being created to meet these new demands.

The Diagnostic Referral Service * has been established in the Department of Pathology, Medical College of Georgia, to offer a range of state-of-the-art diagnostic services to external referral sources, including private practitioners, group practices, pathology laboratories, and hospitals. These specialized services are designed to supplement other routine analyses and assist in the diagnosis, prognosis, and clinical management of patients with a wide range of diseases. The laboratories involved are fully CAP-accredited, and all diagnoses are evaluated by a pathologist certified by the American Board of Pathology. The following is a listing of the laboratories contributing to this endeavor and a brief summary of some of the diagnostic offerings.

This laboratory provides cytogenetic analyses to identify specific chromosomal abnormalities. These are useful not only in establishing a diagnosis of malignancy, but also in classifying certain malignant disorders, monitoring remission and progression, deciding on treatment regimen, and estimating prognosis. Chromosomal analysis can be conducted on bone marrow aspirates, peripheral blood, and lymph node biopsies.

This laboratory offers standardized transmission electron microscopic analysis of biopsies from any organ or tumor, as well as specialized procedures for examination of cell suspensions such as bone marrow and lung aspirates and blood samples. EM analysis is particularly useful in conjunction with light and/or immunohistochemical microscopy for the diagnosis of tumors which cannot be classified by conventional light microscopy. Working in close collaboration with the histology laboratory, the immunohistochemical laboratory, and the renal biopsy service, the EM laboratory provides standardized, reproducible ultrastructural analysis which allows a consistent comprehensive approach to the interpretation of biopsy pathology.

Immunophenotyping by flow cytometry details the presence or absence of surface antigen markers of cellular maturation which define cellular subsets present in specific forms of leukemia and lymphomas. This technique provides precise information to aid in the diagnosis, prognosis, and clinical management of patients thought to have varying forms of lymphoproliferative disorders. In addition, quantitative DNA ploidy and cell cycle analysis in combination with standard histopathologic and cytochemical methods provides the most comprehensive assessment of clinicopathologic status, tumor aggressiveness, and the likelihood of disease progression for patients with breast, colon, and ovarian cancer.

This facility performs state-of-the-art diagnostic immunohistochemical and in situ hybridization techniques for the morphologic analysis of cellular and molecular events. These techniques are adjuncts to histopathologic examination and can be applied to the study of neoplastic, infectious, and other diseases. The laboratory offers more than 65 immunohistochemical markers as special decision-making tests to resolve differential diagnoses. Special histochemical stains and molecular probes for colorimetric in situ hybridization are also provided by this laboratory.

This laboratory offers a full service for examination and consultation on kidney biopsies involving a wide range of disease processes. In addition to routine histopathology, all renal biopsies are examined by immunofluorescence for the identification of immunoglobulins, albumin, and C3 deposits, and by transmission electron microscopy for the detection of early or submicroscopic abnormalities.

This laboratory presently offers complete DAU screening and GC/MS confirmation for pre-employment/employee drug testing. Upon completion of certification as a Forensic Urine Drug Lab, this will be the only such certified lab in this area.

Department of Pathology, Duke University and VA Medical Centers, Durham, North Carolina, USA Microprobe analysis in biomedicine is usually done on an electron microscope (EM) equipped with an energy-dispersive x-ray detector (EDX). This type of analysis is commonly referred to as electron probe microanalysis (EPMA). There are also other fundamentally different new techniques for microprobe analysis which involve the use of laser or ion beams. However, these are not yet commonly employed for diagnostic studies (Ingram et al. 1989) .

Whereas biomedical EPMA primarily used to be a research tool, that is no longer the case. EPMA findings now often have diagnostic, therapeutic, and/or legal significance for the patient (Baker et al. 1985 , Shelburne 1992 ). In addition, since much of the current work involves conditions such as the pneumoconioses, the findings frequently have public health and/or industrial medicine implications as well as implications for single patients (Shelburne et al. 1993) .

Currently the most commonly studied clinical conditions include the pneumoconioses, especially asbestosis and related conditions, "hard metal" pulmonary fibrosis, and other min-eral-induced pneumoconioses (Roggli and Shelburne 1994). A second major application is the use of this technology for the analysis of stones, particularly renal stones. Microprobe analysis can be more sensitive than x-ray diffraction or chemical techniques, particularly for the identification of small components of complex stones. Another major application is the use of microprobe analysis to identify unexplained pigments or deposits and to study unexplained granulomas (Kupke et al. 1984 , Pickett et al. 1980 .

As is evident from the foregoing discussion, most applications involve the study of insoluble particulates. Accordingly, conventional histologic processing with chemical fixation and paraffin or plastic embedding is acceptable. An obvious limitation of this approach is that electrolytes cannot be studied. Currently we are exploring the feasibility of utilizing flash freezing techniques to permit studies on electrolytes utilizing cryoultramicrotomy.

One way to gauge the usefulness of this technology is to study the use of microprobe analysis in a single hospital system, that of the Veterans Administration Medical Centers. Currently there are 171 VA Medical Centers in the United States. Within these hospitals there are 47 diagnostic electron microscopy laboratories. Of these, only five currently utilize EPMA as a diagnostic technique. At each of these laboratories, conventional transmission electron microscopic studies are the predominant type of analysis. EPMA studies constitute less than 5% of our electron microscopy laboratory workload. Nevertheless, as the chemical information available from EPMA is better understood by clinicians, and as cryotechniques are shown to be useful, we anticipate increased usage.

In conducting these studies, it is important to be aware of several artifacts that are common problems. The major types are those caused by poor specimen fixation. Not only does traditional chemical fixation remove electrolytes from the tissue, it is common in electron microscopy laboratories to add heavy metals such as osmium, uranium, and lead. These elements may produce peaks in the final spectrum that can obscure important elements of physiologic significance. For example, the M-alpha line for lead obscures the K-alpha line for sulfur. In conducting an EPMA study, it is important to identify all peaks obtained so that the investigator is not mislead by a contaminant. In addition, it is important to utilize several controls. The investigator should not only probe the feature of interest, but also the cytoplasm adjacent to that feature and the blank stub. Only in this manner can artifacts contributed by, for example, metal in the microscope column be understood and eliminated from consideration. All living things are infected/affected by viruses. Whether the subject is a tissue culture, an animal being used in research, or a human, it behaves differently when infected with a virus. In research subjects virus identification is important to prevent erroneous data due to the presence of a foreign organism. In the case of human viral illness, it is increasingly important to identify pathogens so that appropriate viral therapy can be initiated. Several antiviral agents are already on the market, and many more are presently in clinical trials.

Advantages of using electron microscopy (EM) in viral diagnosis are that it is rapid; specific standards and reagents are unnecessary; and infectious particles are not required. Disadvantages include the facts that a fairly high concentration of virions must be present in liquid samples to visualize them, and that solid tissues may have focal infections that must be included in the sampling.

Identification of viruses by direct EM is performed by two techniques: negative staining of liquid samples and thin sectioning of tissues, cells, and tissue cultures. Negative stains most used in virology are phosphotungstic acid (PTA) and uranyl acetate, although many others have been described. 1 For thin sectioning, any fixation method used successfully for EM of tissue will preserve viruses; this includes some form of glutaraldehyde and osmium fixation, usually followed by uranyl acetate. Detailed preparatory techniques have been described. 2 In negative stains, morphology questions used in identification 3 are: Is the virus naked (always icosahedral) or enveloped (pleomorphic); if naked, what is its size, and does it have distinguishing capsid (outer) markings; if enveloped, does it have visible spikes or fuzz around the outside; what is its size, and is the nucleocapsid (the core) visible in particles that have been penetrated by stain; what is the shape of the nucleocapsid, if visible?

Naked human viruses are all icosahedral; these pathogens fall into three size ranges: 20-35 nm, 45-55 nm, and 70-90 nm. The small viruses may or may not have surface substructure; those that do not are not morphologically distinguishable and have been referred to as small round viruses (SRV) (Fig.   1A) if smooth, or small round structured viruses (SRSV) if rough. Others may have characteristic surface markings that permit precise morphologic identification. The medium-sized (Fig. 1B) and large (Fig. 1C, D) viruses are identifiable.

Enveloped viruses (those that have a pliable covering) are harder to identify, especially if mixed together with cellular debris. If they have surface fuzz or spikes (Fig. 1E) , they are more readily distinguished. The genetic material inside is sometimes packaged into a distinct form such as an icosahedron, similar to the naked viruses (Fig. 1A-D) or helical filament (Fig. 1F ). If the negative stain penetrates the membrane, this nucleocapsid may be recognizable. However, some viruses do not have a morphologically characteristic nucleocapsid.

In thin sections of infected cells, DNA viruses are usually seen in the nucleus ( Fig. 2A, B) where they are constructed, and RNA viruses are usually found in the cytoplasm where they are formed (Fig. 2C, D) , but there are exceptions. Enveloped viruses can be seen associated with or budding through cell membranes; the membrane type is a further clue to identity. Finally, the shape of the nucleocapsid within enveloped viruses is a key. Possibilities are icosahedral (round in sections, Fig. 2A, B) , helical or filamentous (like worms in sections, Fig. 2C ), complex (pox viruses), or morphologically nondescript.

Recognition of viral particles and differentiation from cellular components and debris is paramount. Once the presence of a virus has been determined, one may consult an atlas to confirm identification. [1] [2] [3] [4] [5] For specific concentration or identification of viruses, some antiviral antibodies are available. These reagents can be used to aggregate, to coat, or to gold-label viruses. 1 Use of antibodies requires an a priori hint of the identity of the potential pathogen for selection of the proper reagent.

CHARLES D. HUMPHREY, CYNTHIA S. GOLDSMITH, LUANNE H.

Division of Viral and Rickettsial Diseases, CDC, Atlanta, Georgia, USA An unexplained acute pulmonary illness resulting in the death of previously healthy individuals was recognized in May 1993. The cause of the illness was quickly identified serologically, pathologically, and genetically as a close relative of Prospect Hill virus (a rodent-transmitted hantavirus). 1 Recently, we isolated the virus from trapped rodents near the homes of patients, cultivated it in Vero E6 cells, and determined that it was identical genetically and morphologically to the causative infectious agent. 2 Preparations for electron microscopy (EM) were made by extracting, clarifying, and concentrating the virus from unfixed and 0.25% glutaraldehyde-fixed, supernatant fluids of infected Vero E6 cells3. Uninfected supernatants were prepared similarly as controls. Concentrated virus suspensions were applied to glow-discharge treated formvar-carbon grids, blotted, and stained with 0.5% uranyl acetate (UA) or with 2% phosphotungstic acid (PTA), pH 6.5. Infected and noninfected cells were prepared for thin section by washing with 0.1 M phosphate buffer (PO4) pH 7.3, fixing in situ with 2.5% glutaraldehyde in PO4, scraping, and pelleting. Cell pellets were postfixed en bloc in PO4 buffered 1% osmium tetroxide, stained in 0.5% UA, dehydrated, infiltrated, and embedded in Epon-substitute araldite. Thin sections were stained with UA and in lead citrate.

The extracted negatively stained virus resembled a hantavirus. 3 The 90 nm -200+ nm viruses were spherical, generally nonelongated particles with typical hantavirus grid-like surface features to which surface projections were attached (Fig. 1) . IV-46 Scanning Vol. 16, Supplement IV (1994) FIG Our previous studies have demonstrated that application of scanning electron microscopy (SEM) imaging for examination of transplanted hearts gave new possibilities for evaluation and early detection of acute rejection (Jakobczak et al. 1992) . Continuation of our studies confirmed the benefits of SEM for diagnostics and basic investigations in cardiac transplantations. The aim of this study was to investigate coronary vessels during acute rejection in experimentally transplanted hearts.

Allogeneic heterotopic heart transplants were performed in ether anaesthetized rats, using microsurgical techniques according to the Ono-Lindsey method. Male rats of the inbred Long-Evans strain were used as recipients, and male inbred Sprague-Dawley rats served as donors. The anastomoses-the donor aorta to the recipient abdominal aorta and the donor pulmonary artery to the inferior vena cava of the recipient (end to the side)-were performed under the operating microscope (Wild M691) with 10-0 microsutures (Davis-Geck). Cardiac allograft survival was estimated daily by electrocardiogram and palpation of ventricular contractions. Rejection was considered complete at the time of cessation of a palpable cardiac beat. Rejection was confirmed with laparotomy and histologic examination. The animals did not receive immunosuppression. Allograft survival ranged from 7 to 8 days.

Transplanted hearts were perfused with 0.9% NaCl and 6.25% glutaraldehyde solution in 0.1 M sodium cacodylate buffer (1150 mOsm; pH 7.4). Hearts fixed in glutaraldehyde were cut into thin sections, 1 mm thick slides orientated perpendicularly to interventricular and interatrial septums. Slides were treated with 1% osmium tetroxide for 1 h at room temperature. Then the specimens were dehydrated in an ascending series of ethanol solutions, ending in rinses in 100% ethanol and in pure acetone. Thereafter, slides were criticalpoint dried using CO 2 , mounted on aluminum stubs with conductive silver paint and coated with a thin layer of gold. For examination a JEOL JSM35C SEM was used.

Microcorrosion casts of the coronary arteries and veins and myocardial microvasculature of transplanted hearts were prepared by infusion of methacrylate casting medium (Mercox). Following infusion of approximately 1 ml of casting material for one heart, the preparations were left for 60 min to allow polymerization to occur. Then, the tissue was macerated, leav- Empty 90 nm particles were often seen. Virus particles were seen by thin section EM as shells coated with barely distinct surface projections and enclosing hair-like strands of nucleocapsid material (Fig. 2) . Particle size (80-120 nm) was smaller than that seen by negative stain EM. Spherical particles with elongated tubules (Fig. 2) were often observed by both negative stain and thin-section EM. We conclude that deer mice trapped near the homes of humans with unexplained acute pulmonary illness harbor hantaviruses that likely are the causative agent for the human illness.

ing a microcorrosion cast. Dried preparations were mounted on stubs, coated with gold, and examined with SEM (Murakami 1971 , Potter et al. 1992 .

Investigations of the coronary vessels of the right and left ventricular walls, septum, and the apex region were performed. Tissue fragments were taken from various heart regions each day during the first postoperative week for studies with SEM, and remnant tissues were used for histologic procedure. Grading of the severity of cardiac graft rejection was based upon the Stanford classification. In addition, the coronary vessels of transplanted hearts (of the other experimental group) were reproduced with a casting medium and examined with SEM.

The observations were compared with views of nontransplanted and syngeneically transplanted control hearts.

The applied vascular casting method enables one to study the three-dimensional architecture of the vascular network of the rejecting myocardium. Besides the large areas with minimal damage and readily distinguishable impressions made in the cast material by endothelial nuclei, there were the regions where various signals of vascular pathology were present: changes of diameter of coronary arteries and veins; unusual size and shape of the vessels; indentations on the cast surface caused by adhesion of blood cells to the vessel wall; characteristic occlusion of the vessels; irregularity of capillary network with changes in diameter and cast surface; capillary destruction with accompanying extrusion of casting material into the interstitial tissue. The observed vasculopathies varied in mild, moderate, and severe rejection and in several areas of the hearts.

Studies of coronary vessels of transplanted hearts using SEM imaging have made significant advances in the understanding of rejection vasculopathy possible. SEM observations of sectioned transplanted hearts correlate with results obtained by application of the microcorrosion casting technique. Data showed that the highest rate of acute rejection occurs in the interventricular septum, and the development of rejection in various heart regions differs significantly. Observed differences in localization and dynamics of development of vasculopathies in the septum and left and right ventricular walls of transplanted hearts confirm the unequal character of rejection. SEM imaging is a valuable method for investigation of vascular pathology during acute rejection in transplanted hearts.

The authors are very grateful to Prof. A. Miodonski, director of the SEM Laboratory of ENT Dept. of the N. Copernicus Academy of Medicine, Cracow, for the enabling of SEM observations. Preservation of the ureteral blood supply in kidney transplant surgery and ureteral reconstruction is of paramount importance in preventing urological complications of the ureter. 1 The incidence of complications in ureteral surgery has declined in recent years, in part due to greater understanding of the ureteral blood supply and improved surgical technique. 2 However, the ureteral vasculature has been described primarily at the gross level and only recently received attention at the microvascular level. 3 Because of its critical role in the success of renal transplants and its potential vulnerability to surgical trauma, the ureteral vasculature merits further investigation.

We studied the microvascular anatomy of the ureteral vasculature (UV) in 29 male New Zealand rabbits using transmission electron microscopy (TEM) and scanning electron microscopy (SEM) of vascular corrosion casts and alkalitreated tissue samples. The UV was perfuse-fixed with buffered glutaraldehyde via the abdominal aorta, washed in buffer, and treated with alkali according to the method of Takahashi-Iwanaga for SEM. 4 For TEM, fixed tissue was embedded in epon, sectioned, and stained with uranium and lead. For vascular corrosion casts, the UV was washed free of blood with buffered saline and filled with plastic resin (Mercox•/methylmethacrylate, 4/1) at physiologic temperature and pressure. Resin-filled tissues were digested with 10% KOH, washed in water, critical-point dried, mounted on stubs, and sputtercoated with gold palladium for SEM examination.

The blood supply to the male rabbit ureter originates primarily from ureteric branches of the renal artery, the testicular artery, and the vesicular artery. These intrinsic vessels run within the wall of the adventitia the full length of the ureter. Perforating arterioles and venules pass through the muscle wall and divide further in the lamina propria to supply a dense plexus of continuous capillaries in the mucosa. Although numerous arterioles and venules populate the lamina propria, the ureteral musculature does not possess a rich capillary bed.

The capillary plexus is positioned between the transitional epithelium and the lamina propria and uniformly extends the entire circumference of the ureter. Capillary orientation occasionally follows the longitudinal axis of the ureter, but is primarily plexiform and exhibits multiple "Y" and "T" shaped anastomoses (Fig. 1) . At the points of junction with the underlying arterioles and venules of the lamina propria, the capillaries commonly exhibit "kinks" and "bends." Alkali treatment revealed that the capillary bed is supported by a dense, fibrous network of collagen (Fig. 2) . Pericytes were occasionally associated with the capillary network. Intercapillary distance ranged from 10-80 µm and capillary diameters typically were 5-15 µm. Venous valves were not observed in the ureter.

The combination of vascular corrosion casts, alkali treat-ment, and TEM provides a clear and comprehensive perspective of the microvasculature of the rabbit ureter. The limited number of intrinsic vessels supplying the ureter wall and capillary bed emphasize the importance of understanding the blood flow in this organ in related surgical procedures. Rabbits were perfused with glutaraldehyde and Mercox ® via the abdominal aorta and subsequently processed for preparation of vascular replicas according to a previously described method. The angioarchitecture of the tibia was studied with use of the scanning electron microscope. At low magnification the pattern of vascularisation to the tibia could be established. They are richly supplied with arteries which pass into the bone substance proper from the periosteum. The main supply is from the nutrient arteries, which divide longitudinally in a dichotomous manner after passing through a foramina in the compact bone, major branches running both proximally and distally. This is the general case in all long bones. These branches reach the epiphysis and diaphysis of the tibia. In addition, separate arterial entities supply specific regions of the tibia: I. periostal vessels, II. epiphyseal vessels, and III. diaphyseal vessels.

Arteries and veins can be readily distinguished due to their characteristic nuclear impression present on the surface of the replicas. Arterial nuclear impressions are oriented longitudinally to the long axis of the vessel, whereas in veins they are situated circumferentially (Miodonski et al. 1981) . For further details about methods, see Syed Ali et al. 1991. At higher magnification it was possible to demonstrate the microangioarchitecture of the epiphyseal region, which consisted of a very fine meshwork of capillaries arising from a main epiphyseal artery. A connection between arteries of the epiphysis was observed and consisted of medium-sized arteries. The capillaries of the epiphyseal meshwork show a rich anastomosing network building open, sinus-like dilatations. Capillaries are also richly present in the areas of the growth plate and stressed areas of the epiphysis of the tibia. The so-called sinuses are situated under the cartilaginous plate. The venous drainage takes place through the vessels running parallel to the corresponding arteries. The animals were premedicated with chloroform and were then anesthetized with Ketanest (0.3 ml/kg body weight) via intraperitoneal injection. After examining the abdominal reflex the rabbits were fixed on an operating table. All preparations were carried out at room temperature. The abdomen was shaved and opened cranially at the xyphoid process down to the pubic symphysis. After gently displacing the intestines and mesenterial convolutes, the abdominal aorta and the inferior vena cava were exposed very carefully caudal to the kidney vessels. A knob cannula was inserted through a small incision in the abdominal aorta and held in place by a ligation around the cannula distal to the point of incision. These procedures should be performed very quickly, otherwise the blood pressure will fall. The hindlimb vessels were washed through the cannula with an isotonic NaCl solution containing heparin (15000 I.E.) and sodium nitroprusside (280 mg/l) in 1000 ml. The inferior vena cava was opened with an incision as exit for fixing fluid and clamped when required. Soon thereafter, 0.5% glutaraldehyde in phosphate buffer (0.135 m, Ph 7.2) was passed through the same cannula. After completing the perfusion, about 60 ml Mercox (methacrylate) was applied through the cannula and observed until the Mercox came out of the inferior vena cava. It was then clamped to avoid unnecessary backflow.

The animals were left overnight at room temperature and then placed in a water bath for 1 h for rapid polymerization. The bones were dissected free from other tissues and left in a 30% KOH solution, which was changed every day. The specimens were washed with distilled water very carefully; it is very important to avoid every kind of mechanical disturbance. The solution and water can be changed with the help of a small water pump. The best results were achieved with an alternative change of 30% KOH and 5% trichlor tetra acetic acid solutions. After complete maceration, the preparation was checked under a stereo microscope to eliminate all rest tissue. It was then frozen at −20°C and freeze-dried. The preparation was contrasted with 1% Os O 4 vapour in a desiccator for about 24 h to increase the contrast in the microscope. The specimens were fixed on scanning stubs with conducting silver and sputtered with gold (3 min, 30-40 mA). They were then viewed in a PSEM (Philips) microscope and photographed with Kodak film. IV-50 Scanning Vol. 16, Supplement IV (1994) FIG. 1 Low magnification view of a rabbit tibia from the epiphysial zone with its specific end capillaries arrangement, the proliferation zone, and the connecting capillaries between the epiphysis and metaphysis ( Apart from the ability to capture three-dimensional (3-D) images of microscopic structures, a confocal laser scanning microscope (CLSM) equipped with multiple detectors allows one to add an extra dimension to the data acquisition process. A typical example are the ongoing studies in our laboratories. 1 In this paper, we discuss the channel spill-over problem associated with acquiring two-color CLSM images and present image processing techniques that analyze multichannel image data.

For double labeling of DNA replication sites, mouse 3T3 cells, exponentially grown on cover slips or synchronized at specific times in 8-phase, were pulsed for brief times (2-5 min) with CldU (chlorodeoxyuridune). The pulsed cells were then fixed 2 and processed for fluorescence microscopy using monoclonal antibodies, appropriate extraction conditions, and fluorochrome-conjugated secondary antibodies which enabled differential recognition of sites of CldU versus IdU incorporation into newly replicated DNA. 3 Under the conditions used, there is no measurable cross reaction between the antibodies for CldU-versus IdU-labeled replication sites.

Optical sections of labeled cells were collected with an Olympus LSM GB-200 CLSM with a 25 m W Ar ion laser. The microscope is operated in high resolution mode (1024 × 768 pixels) with three fluorescence channels and a transmission channel. The microscope is controlled by a 80486 -33 MHz computer with 16Mb RAM and 1.7 Gb hard drive. The digital confocal optical sections were transferred via Ethernet to a dedicated Sun Sparc 10/40 with 64Mb memory for further analysis.

During image capture, a combination of band-pass and high-pass filters are used to mask unwanted emission at each detector. A major problem with the dyes used is their overlapping emission spectra. This seemed to contribute a significant amount of spill-over from the green channel (FITC) to the red channel (Texas red or rhodamine) and only an insignificant amount in the other direction.

Because of the unidirectional nature of the spill-over between red and green channels, we were able to apply a correction factor to the red channel based on the green-channel intensity. To obtain the correction coefficients, a set of calibration images were obtained by using a sample which is identically labeled in both color fluorescent dyes and captured at different gain and offset parameters. For each of these images, the mean spill-over value in to the red channel (R i ) is calculated for each pixel value (i) in the green channel as follows: (1) where I R (x,y,x) and I G (x,y,z) denotes the image intensities of the red and green channels, respectively, and N i indicates the number of pixels in the green channel having value i. A third order polynomial is fitted to the data set (<i, R i >,i=0,1, ... 255) and the corresponding coefficients (a o a 1 ,a 2 ,a 3 ) are estimated using the least square error criteria. The correction factor for each pixel of the red channel is then computed as follows and subtracted from the original intensity of the red channel to obtain the corrected image.

∆I R (x,y,z)= a o + a 1 I G (x,y,z) + a 2 I G (x,y,z) 2 + a 3 I G (x,y,z) 3 (2) Figure 1 shows the channel spill-over for different filter sets. The corrected images were analyzed using a set of image processing routines developed at our laboratory 4 to obtain the boundary of replication sites in individual channels. These boundary data are then used to detect the overlapping particles between channels, and the overlap area is calculated for each overlapping particle. The results are presented in several forms including a two-dimensional histogram of particle volume and overlap percentage.

The image processing routines are able to generate boundary data of a 600 × 512 × 26 image with 1800 particles in approximately 30 mins and the overlap estimation using boundary data takes approximately 15 s. Currently we are working on visualization techniques of raw and processed multichannel 3-D images. 

these fringes, these dark, granular, linear structures shed dark particles which then fused with opposite lines. A stream of particles trafficking between cells in contract, particularly at overlaps, was noticed. This pointed to a rather intense exchange of very small particles between cells of the same origin on contact. We conclude that interference reflection mode, video-rate, laser scanning confocal microscopy is a useful tool for intravital analysis of the intracellular structural dynamics in relationship to cell type, function and pathophysiological state.

Department of Plant Pathology and Department of Botany, University of Georgia, Athens, Georgia, USA Powdery mildew diseases of plants are caused by a group of obligately parasitic fungi belonging to the order Erysiphales, class Ascomycetes. The somatic hyphae of most of these fungi grow exclusively on the surfaces of infected plant organs, most typically leaves. These hyphae form specialized structures known as appressoria that attach to the host surface. Each appressorium gives rise to a tiny penetration peg that penetrates the wall of the underlying epidermal cell and invaginates the host cell plasma membrane. The penetration peg then develops into a specialized structure known as a haustorium that absorbs nutrients from the host cell. In this study transmission electron microscopy was used to examine the haustoria of the powdery mildew fungus Erysiphe lagerstromemiae and the relationship of these structures to epidermal cells of infected leaves of crape-myrtle (Lagerstromemia indica).

The haustorium of E. lagerstromemiae (Fig. 1 ) possesses a slender neck region and an expanded body that contains a single prominent nucleus. Much of the neck is surrounded by a collar of host cell wall material. A single septum with a tiny central pore with which Woronin bodies are associated is present in the distal portion of the neck near the haustorial body. The haustorial body is divided into numerous small, coiled branches. The entire haustorial apparatus is separated from the host cell cytoplasm by an extrahaustorial membrane (Figs. 1, 2) . Though continuous with the host cell plasma membrane, the extrahaustorial membrane is much thicker than the plasma membrane and is highly convoluted in certain regions. The haustorial branches are separated from the extrahaustorial membrane by a region known as the extrahaustorial matrix. At this point it is unclear whether the finely fibrillar material comprising this matrix originates from the fungus, the host cell or both.

Although the haustorial apparatus of E. lagerstromemiae occupies a considerable portion of the overall volume of an invaded cell, the host cell organelles appear normal. However, numerous Golgi bodies and many structures resembling microbodies are concentrated in the host cytoplasm very near the extrahaustorial membrane (Fig. 2) .

PAVEL VESELY AND ALAN BOYDE* Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic; *Department of Anatomy and Developmental Biology, University College, London, U.K.

Video-rate laser scanning confocal microscopy in the reflection interference mode (IRM) enables the visualisation of fine intracellular structures in living cells in vitro and the observation of rapid changes of shapes, the trafficking of very small particles, and exchange of material. The dynamics of this intracellular activity can be established by optical sectioning with objectives of high magnification and numerical aperture, from within the bottom level of the cell contact to the substrate through to the top cell surface: these levels may be up to 15 µm apart for large and/or tall cells.

In the present studies, three video rate confocal microscope systems (Lasertec 1LM11, Noran Odyssey and BioRad DVC250) were used to compare the ultramicroanatomy and dynamics of motion of intracellular structures in established cell lines (K4, K2, T15, and A87ON). K2 cells are clonal descendents of spontaneously in vitro transformed LEW/CUB rat fibroblasts which give rise to low metastatic sarcomas in vivo: K4 cells are Rous Sarcoma Virus transformants of the K2 line and both T15 and A87ON cells developed from K2 by neoplastic progression to a higher metastatic capacity. Previously observed differences in the incidence of cells with rapidly oscillating and trafficking particles between K4 and the other cell lines were analysed at improved resolution.

Using the Noran Odyssey system, with zoom factor 5 and a 100/1.4 Nikon lens, this phenomenon was observed in almost all the cells of all the cell lines compared. This finding provided an explanation of the previously described difference between K4 and the other cells which was obtained using a Nikon 40/0.55 WI lens (but not with an Olympus 150/1.25 WI using the Lasertec system, zoom factor 1) from the level of the cell periphery adjacent to the culture surface only.

The present observations indicate that there may be a difference in the extent of this type of intracellular activity towards the periphery of the cell. More accurate mechanical targetting of the optical probe will be needed before it will be possible to measure this effect. At the free, top surface of some K4 cells, small pinocytic opening and closing was seen. Similar images of opening and closing near the bottom sides of some cells in all cell lines were evidently produced by vertical oscillatory movements of particles in small clusters. When a 2.4 mm pinhole diameter (Odyssey) was used to improve confocality, the interference fringes were observed to move, during focussing, along the steep slope of the high bodies of the K4 cells, optically transforming granular, linear, or randomly oriented structures inside the cell into concentric lines. Above and below

Florida Institute of Technology, Department of Biological Sciences, Melbourne, Florida, USA Acid phosphatase (AP) is well known as one of the representative lysosomal enzymes. AP activity is recognized at the light microscopic level as intensely stained granules. Although the localization and distribution of AP activity in the mammalian inner ear were researched by several investigators, we found only one report about the occurrence of AP in the avian inner ear (Marmo 1965) . The objective of the present study is to demonstrate the occurrence of AP activity in the membranous labyrinth of the chick's inner ear, using the simultaneous coupling azo dye method (Barka and Anderson 1962) , and to discuss the significance of these findings.

After newly hatched chicks were sacrificed, their ears were fixed for 2 h in 2.5% glutaraldehyde (GA) or 2.5% paraformaldehyde (PFA). The membranous labyrinths were dissected out, dehydrated in either a graded series of acetone (used only for 2.5% GA fixation specimens) or N,N-dimethylformamide (DMF), and embedded in JB-4 (Polysciences, Warrington, Penna.), a methacrylate plastic. These sections were incubated for 1 or 2 h at 37°C in an incubation medium (pH 5.0) containing naphthol AS-BI phosphate as substrate and hexazonium pararosaniline as a coupler that used the azo-coupling method. In some instances, intact, fixed whole specimens were incubated in an aliquot of the same incubation medium as that of sectioned specimens and were then dehydrated with DMF, embedded in JB-4 or LR White, and sectioned. For control specimens, either a substrate-deficient medium or an incubation medium containing cupric sulfate (0.002 M) was used. Methyl green was used as a nuclear stain.

Intense AP activity, as represented by dense dye deposits, was detectable in the supranuclear area of almost all hair cells in the basilar papilla and vestibular sensory hair cells (Fig. 1 ). This result was in agreement with the finding of Marmo (1965) . There were no differences in the distribution pattern of AP activity of the hair cells from the distal to the proximal region in the basilar papilla. In particular, hair cells of lagenar macula were most often characterized by intense AP activity in the subcuticular area (Fig. 2) . The functional significance of high levels of AP in the hair cell is still a matter of conjecture. For example, Ishii and Balogh (1966) demonstrated no morphologic evidence for phagocytic or secretory activities in hair cells, although their electron micrographs of hair cells reveal lipofuscin-like substances within lysosomes. It has long been contended that nonsensory hair cells such as stria vascularis, external spiral sulcus cells, and those around the macula and cristae ampullares help to regulate the ionic composition of endolymph (Kimura et al. 1964, Kikuchi and Hilding 1966) . In the present study, the columnar cells and the cells of the tegmentum vasculosum showed moderate to strong AP activity, and the transitional epithelia of the cristae ampullares also showed strong AP activity . The dark cells help to regulate the ionic composition of endolymph and perilymph (Kimura 1969 ). In the present study, intense AP activity of dark cells was concentrated in the supranuclear area or diffusely in the cytoplasm. These results show that AP activity is highest in cells with highest levels of transport activities on the production of endolymph (Marmo 1965) , although the metabolic linkage between AP and the transport enzymes is uncertain. The wall cells and supporting cells of the vestibular labyrinth showed no enzyme reaction. The statoacoustic and vestibular ganglion cells showed various degrees of AP activity. The sections of statoacoustic and vestibular ganglion cells of specimens that had been embedded in LR White displayed more intense AP activity than those sections from specimens embedded in JB-4. AP activity was also stronger in specimens fixed in 2.5% PFA than in 2.5% GA.

The syncytiotrophoblast layer covers the fetal villous tree which is in direct contact with maternal blood. It contains many microvilli, resembling a carpet surface, which are responsible for the absorption, excretion, and synthesis of many key hormones which are important for fetal development, secretion, and exchange of gases; of course, many earlier scanning electron microscopy (SEM) studies of normal and pathologic placental villi have been described and many investigations on the histology, ultrastructure, and three-dimensional features of human placental villi have been carried out. [1] [2] [3] [4] Technical handicaps causing structural deformations on the placental villi 5 have been indicated. All of these studies depended on the normal development of placental villi but were not found to be comparable with tubal pregnancies.

The purpose of this study is to describe the development of the human placental villous trees emerging from chorionic plate during the early periods of uterinal and tubal pregnancies, and to compare their three-dimensional structures using SEM technique. IV-54 Scanning Vol. 16, Supplement IV (1994) FIG Samples of 12 human placentas have been studied. Six early specimens obtained by curettage or by hysterectomy have been staged as described in our previous study. 6 Four additional specimens dated according to anamnestic data in comparison with the measurement of fetal weight and length at 5, 6, 7, and 8 weeks p.c. were obtained from clinically normal pregnancies interrupted by legal curettage or by hysterectomy, and two samples aged 8 and 10 weeks p.c. from tubal ectopic pregnancies. All specimens were fixed and prepared for SEM. The specimens were dehydrated in ascending concentrations of ethanol, critical-point (Bio-Red E 3000) and air dried, and coated with a layer of gold particles using a Bio-Red SC 502 super coater. Observations were made with a JEOL-JEM 5200 SEM.

Ectopic tubal placental villi trees are not as well developed as normal uterinal villi trees. Villi formation and ramification were very rare depending on the physiologic causes. Some villi shoots were thinned gradually from base to its apex, and at the top these developing villi showed very interesting properties; they were curved, thinned, and crossed each other (Fig. 1) . The predominant villi types were immature, intermediate, and mesenchymal villi. [7] [8] [9] The surface of these villi trees was completely covered by small and deformative curved microvilli and by fibrinoid particles compared with the normal uterinal villi trees. The trophoblastic shell of some villi was peeled off and degenerated. Developmental retardation of placental villus trees is clearly seen. Some villi were curved and crossing, and gradually thinned to the apex; on the villi surface, many wrinkles and furrows resembling very old skin were observed (Fig. 1) . It is well known that the placental development shows a parallel regulation due to the desidual properties. Cellular and extracellular interactions take place between uterine and trophoblast during initial stages of placentation. 7, 10-12 Khong and Robertson 14 suggested that the deficient decidua is the main reason why tubal pregnancies do not reach to term. The findings they describe may be explained as being due to absent decidua rather than being its cause. Moreover, another reason for the usual failure of ectopic pregnancies to reach term is placenta accreta which becomes placenta percreta with bleeding. 14 According to our SEM observations, in tubal pregnancies the formation and development of placental villi trees were found in some patterns observed in uterinal placental samples. Villi formation patterns-buds, tendrils, and shoots-are very rare on the surface of villi trees (Fig. 2) . Placental villi shoots do not show uniform composition compared with to the uterinal villi samples. These are very thinned to the apex of villi, and are compressed in a common configuration, resembling a racket which does not appear as an alive condition. These observations are very interesting and original. The University of Iowa Central Electron Microscopy Research Facility (CEMRF) was established in 1981 to support all faculty, staff, and students needing this technology. Initially, the CEMRF operated with one transmission electron microscope (TEM), one scanning electron microscope (SEM), and three staff members, and it supported about 30 projects annually. During the past 13 years, all instrumentation predating 1981 has been replaced and now includes two TEMs, two SEMs, two EDS systems, a scanning probe microscope, a laser scanning confocal microscope, and all necessary supporting equipment. The facility presently consists of 14 staff members and supports over 150 projects yearly from 45 departments in 5 colleges and 12 industrial laboratories. One of the unique strengths of the CEMRF is that both biomedical and physical scientists use the facility.

The development of the University of Iowa CEMRF was made possible due, in large part, to the Central Administration's support of equipment acquisition over the past decade. Of the $2,000,000 invested in equipment, more than 70% was provided by the Graduate College and the Vice President for Research, with the other 30% obtained through equipment grants and user fees. Of the operating expenses for the facility, 75% are recovered from a large and well-funded group of investigators who are charged on an at-need, fee-for-service, first-come, first-served basis. Faculty recognize that they have a facility available to them that provides immediate access to state-of-the-art equipment and techniques, as well as one that provides training and supervision. In addition to supporting research, the CEMRF offers four formal courses, as well as assisting with sections of eight other classes. The facility annually organizes at least two workshops and provides about 100 tours for visiting scientists, faculty and student recruitment, high school students, business groups, and politicians. The facility also serves as the Business Office for the Iowa Microscopy Society.

Concurrent with the development and success of the CEMRF, several departments voluntarily divested themselves of their own electron microscopy (EM) equipment (28 EMs on campus in 1981 compared with 11 EMs in 1994). This represents a savings of more than $250,000 annually (1994 dollars), as well as the release of 17 or more rooms for other use. In addition, the interaction between the remaining six EM laboratories is more positive as a result of this downsizing and centralization. The availability of user-friendly quality equipment in the CEMRF assures that EM is now accessible to all University faculty, staff, and students. In addition, with many investigators sharing instruments, it is easier to justify acquisition of new equipment for the CEMRF. Accurate dimensional metrology of the submicrometer gold absorber structures of x-ray masks can be accomplished in the scanning electron microscope (SEM) with the use of electron beam modeling (Postek 1993) . Accurate metrology is possible because the x-ray masks present a unique measurement object from most other semiconductor structures viewed in the SEM. This occurs because the silicon support membrane is x-ray transparent by design. This characteristic can be used as a distinct advantage in electron beam-based mask metrology since, depending upon the incident electron beam energies, substrate composition, and substrate thickness, the membrane can also be essentially electron-transparent. The areas of the mask where the absorber structures are located are essentially x-rayopaque, as well as electron-opaque. Viewing the sample from a perspective below the mask by placing an electron detector beneath the mask provides excellent electron signal contrast between the absorber structure and the base membrane. The present technique utilizes a broad acceptance angle detector which is different in concept from other transmission electron (TE) detectors used in the SEM. In this case, the broad angle is used to detect as many of the transmitted electrons as possible (i.e., whether scattered or not) that have an energy above some predetermined threshold which is usually several kiloelectron volts. Then, the electrons are physically filtered both by the signal threshold characteristics of the detector and an by an electron energy filter in front of the detector. Energy filtering of the transmitted electrons excludes the highly scattered and thus lower-energy electrons from entrance into the detector. This greatly improves the contrast level over the conventional transmitted-electron detection mode for this type of application, and greatly simplifies the required electron beam/sample interaction modeling necessary for edge determination. It is, in fact, this change in electron detection philosophy that makes the present TE approach so attractive for dimensional metrology and inspection of x-ray masks.

Monte Carlo modeling of the transmitted electron signal was used extensively to support this work in order to determine the optimum electron detector position and characteristics, as well as to determine the position of the edge in the image profile. The Monte Carlo modeling is more accurate in this work, in contrast to the secondary electron detection mode, because in the transmitted electron detection mode the modeling of the electron beam/specimen interaction becomes far less difficult than in the modeling of typical secondary electron images of other opaque objects. The generated low-energy secondary electrons (which are complex to model) are excluded from the detector and, therefore, do not need to be included in the calculation. Combining all of these factors provides a modeled signal that is extremely sensitive to wall slope. Wall slope variation can result in large differences in the modeled profiles. The comparison between the data from the theoretically modeled electron beam interaction and actual fitted experimental data is shown in Figure 1 for a wall angle of 4˚. The theoretical profiles were shown to agree extremely well with experiment, particularly with regard to the wall slope characteristics of the structure obtained from the SEM images and video profiles. A plateau in the transmission is seen in the modeled profile as the beam traverses the edge, which can be used to identify the loca-tion of the edge of the absorber line and thus allows accurate measurements to be made.

This work provides an approach to improved x-ray mask linewidth metrology and a more precise edge location algorithm for measurement of feature sizes on x-ray masks in commercial instrumentation. The transmitted electron detection mode is also useful in both mask inspection and mask repair, because the high contrast of the image allows for rapid determination of mask defects and high-density contamination particle detection because the transmitted electrons simulate transmitted x-rays. This work represents the first time that electron beam modeling has been used to determine the accurate edge location in an SEM image. This also represents an initial step toward the first SEM-based accurate linewidth measurement standard from NIST, as well as providing a viable metrology for linewidth measurement instruments of x-ray masks for the x-ray lithography community.

Postek Using Monte Carlo models that take into account the gaussian beam width of the incident electron beam, a noticeable broadening of the linewidth measurements is simulated for aluminium linewidths under glass. This simulated effect correlates closely with measurements of linewidth using energy dispersive x-ray analysis of an aluminium Ka line under glassivation. Detailed simulation of a full linewidth measurement has been performed using a point source and gaussian incident electron beam for the multiple scattering and single scattering models. The modeling effort was undertaken to better understand the effect that incident electron beam width would have on the accuracy of scanning electron microscope measurements.

The multiple scattering (MS) model was adapted from PAS-CAL code for a single material 1 to simulate layered materials, specifically, SiO2 over aluminium. Further, application-specific code was added to simulate an electron beam scanning across a linewidth on a fabricated microelectronic circuit. The linewidth selected for these simulations was one micrometer and one micrometer thick with a selectable SiO2 thickness for the overcoat. In this case, the SiO2 thickness was 0.7 micrometers. The single scattering (SS) model was adapted also from available PASCAL code for layered materials 2 with the same application-specific code used in the multiple scattering model.

The differences in the results from the two models used for the simulation are shown in Figures 1 and 2. The results for the MS and SS models at 10 K trajectories are shown in Table I .

The gaussian beam width was selected to be 0.02 micrometers. The gaussian beam noticeably broadens linewidth measurement over point source, based on the results in Table I . The linewidth measurement was taken at FWHM by extrapolation between points. The large differences between the SS and MS models is under further study. The calculation of the image contrast from samples with surface topography can be done using Monte Carlo techniques as long as the electron trajectories can be calculated through a surface profile. The image simulations that are described here were done using the same methodology that has been applied to the calculation of the electron backscattered signal from samples where the composition variation is taken into account (Ly and Howitt 1992 and these proceedings) . The design of the program is such that the scattering cross sections are reassessed as the electron passes from one region of the specimen into another, which in this case includes free space. The program takes longer to run than that for a homogeneous specimen with a flat surface because the parameters, such as the atomic number and atomic weight and density, need to be constantly updated at each point in the calculation.

Depending upon the image resolution that is required, we divide the three-dimensional specimen into blocks of either specimen or vacuum. The modifications to the conventional Monte Carlo approach to such calculations include not only the specimen geometry but also the determination of the energy of the electron that is backscattered and the direction in which it travels. In this way the signals at the backscattered and secondary detectors can be distinguished because almost all the low-energy electrons find their way to the secondary electron detector, whereas only those in the line of sight to the backscattered detector contribute to this signal.

In most practical situations, where the geometry of the specimen is difficult to predict, it is useful to have specimens of standard shape, such as a sphere cone or box, to compare directly with the images. The calculation of the image at the secondary electron detector for spheres of the same size but of different atomic weight is shown as an example in Figure 1 . The spheres are assumed to be on a graphite substrate and the electron-beam energy is 10 keV. The signals are displayed as they would appear in a micrograph, that is, in an intensity scan across the image and as a two-dimensional profile of the image intensity. We have also found that pseudo color images, where the various gray levels are replaced with different colors, are very useful when it comes to comparing calculated and experimental images. 1 Calculated secondary electron images from 400 nm spheres of silicon, titanium, nickel, and molybdenum on a carbon substrate at an electron beam energy of 10 keV. The signals are displayed as they would appear in a micrograph, that is, in an intensity scan across the image and as a two-dimensional profile of the image intensity.

PIERRE HOVINGTON, DOMINIQUE DROUIN, RAYNALD GAUVIN, DAVID C. JOY * Department of Mechanical Engineering, University de Sherbrooke, Sherbrooke, Québec, Canada; * EM Facility, University of Tennessee, Knoxville, Tennessee, USA Quantitative analysis of resolution < 100 nm can be achieved at low accelerating voltages. However, to exploit all the emitted signals (x-rays, secondary, and backscattered images) fully, specialized programs have to be used. In this paper we present some results that can be achieved with such programs. It is shown that our single scattering Monte Carlo program can effectively predict the effect of surface oxidation on aluminum (Al), model the backscattered coefficient even at very low energy (< 1 kV) and accurately predict the backscattered profile around a spherical inclusion.

At low voltages, Mott elastic cross section has to be used. The tabulated Mott scattering cross section of Czyzewski et al. (1990) , combined with the energy loss of Joy and Luo (1989) , is used as the physical basis of the program. Hence, the program can be used accurately for simulation even at very low energy (E < 1kV) (Fig. 1) . Numerical experiments can be made with one or several regions of different composition and shapes. The regions can be defined as horizontal (i.e., layered samples) or vertical (i.e., grain boundary). In addition, based on the results of Gauvin et al. (1992) , spherical inclusions can also be modeled.

The program is written in C language and makes use of a fully graphic environment both for input and output of data. In Figure 2 , we present a typical output of the program of a simulation on a multilayer electronic component [a 400 nm Ga (28% at ) Al (22 % at) As (50% at) on a GaAs substrat] covered with 10 nm of contaminated carbon. The effect of the backscattered electrons on the resolution can be clearly determined since their paths are marked with a different color. In addition, at the end of the simulation, the interaction volume ranges at 68, 90, 95, and 99% are plotted.

To increase the range of applicability of our program, backscattered, secondary, and x-ray images (K, L, and M lines) can be generated. Generated and experimental images can then be used for metrology and microanalysis. We present in Figure  3 a backscattered linescan of a 10 nm diameter beam (Eo = 5 kV) with a hypothetical 200 nm diameter hemispherical MnS inclusion in a Fe matrix. It is important to note in Figure 4 that the difference between the "screen dimension" and the "real dimension" is over 12% . In Figure 5 , we present an experimental backscattered profile taken across an MnS inclusion in steel at 15 kV. We note the similarity of the theoretical and the simulated backscattered profile (slight decrease between the center and the end of the inclusion and the presence of peaks at the interface between the matrix and the inclusion). IV-60 Scanning Vol. 16, Supplement IV (1994) FIG . When low voltages are used, the geometry and composition of the surfaces greatly influence the resulting signals. In Figure  4 , we present an experimental spectrum from a pure aluminum (Al) sample taken at 5 kV with a windowless EDS detector and a theoretical generated spectrum (Hovington et al. 1993) . The difference between the two spectra is mainly due to the presence of oxide (Al 2 O 3 ) at the surface of the sample. In Figure 6 , we compute the j(rz) curves for an oxide thickness of 50 and 20 nm and for a nonoxidized sample. The decrease of the Al K intensity for the 50 and 20 nm compared with the nonoxidized sample is 55 and 51 %, respectively. The experimental to theoretical ratio found on the spectra (Fig. 2) is approximately 60%, indicating that the oxide layer is < 20 nm thick. It is important to note that because the standard and the unknown may not have the same thickness of Al 2 O 3 , the use of theoretical stan-dards may become critical in quantitative x-ray microanalysis at low voltages.

Material investigation methods based on the interaction of the electron beam with a solid requires detailed information about the electron distribution. Among the various approaches which allow one to calculate the electron distribution, the Monte Carlo simulation seems to be most suitable. This is due to the fact that the Monte Carlo method generally can be applied to any target. The main problem of this approach is the accuracy of the electron distribution approximation. One important criterion of the quality of the approximation is the coincidence of the calculated and experimentally observed x-rays in depth distribution. Note that the correct calculation of this function plays the essential role in EPMA data interpretation.

As have been shown in the Monte Carlo process developed by Murata et al., 1 which is based on the Mott cross section for elastic and on a knock-on model for inelastic interactions, it provides a rather good agreement between the calculated and experimental φ(ρz) function for CdLα line in Al and Au targets at the electron beam energy E = 25 keV. But a more careful comparison of φ(ρz), obtained according to this model with the experimental data 2 for SiKα in Al, Ni, Ag, and Au at E = 6, 8, 10 keV, respectively, and CuKα at E = 12, 15 keV, reveals some discrepancy, especially in the tail part. This discrepancy increases when the atomic number grows and the electron beam energy decreases. The origin of this discrepancy can be connected with the errors of the approximation of the differential cross sections for elastic and inelastic interactions, as well as with the ionization cross section.

As the Mott formula for elastic cross section uses the model of the atomic potential, then the approximation of this potential can influence the results of the scattering. To avoid the errors connected with the choice of the atomic potential approximation, we have used the elastic differential cross section obtained by Riley et al. 3 with the help of the static approximation theory (SAT). The main features of the developed Monte Carlo program are as follows: 1. The elastic interaction is calculated according to the SAT cross-section 2. The energy dissipation process is described by the knockon model. The Bethe formula for dE/ds is used in the high-energy region (E > 6.338J i ). In the low-energy region (E < 6.338J i ), the equation used in [1] is exploited instead of the Bethe law 3. The fast secondary electrons produced by electron-electron interactions in an inelastic collision are taken into account 4. The mean ionization potential is chosen according to the Berger and Seltzer formula 5. To calculate the ionization cross section, the formula of Gryzinski is used. We have calculated the energy and angular spectra of the electrons transmitted through the films of the various compositions and thicknesses. Comparison of these results with the data obtained with the help of the Mott cross section shows that the use of the SAT approximation for the elastic interaction does not influence essentially the electron distribution and cannot improve the accuracy of its determination. The comparison of φ(ρz) functions which are found in accordance with these two models confirms this conclusion. Therefore, one can state that the details of the Mott cross section together with the errors in atomic potential approximation cannot influence the electron distribution in targets.

The factor which can affect the φ(ρz) in targets independently of electron distribution is the ionization cross section. To take into account this factor, we have used the experimental results obtained by Long et al. 4 During differentiation of the heart there appears to be a sequential pattern to the formation of individual muscle fibers. Phenotypic changes result in the expression, synthesis, and organization of complex proteins to form the terminally differentiated myocytes. The formation of the basic pattern of myofibers ultimately determines the physiologic performance of the final form of the heart. Pattern formation involves both intracellular events associated with myofibrillogenesis and extracellular events associated with cell:cell and cell:matrix interactions. This basic pattern is repeated to form complex layers which results in the final form of the heart. It is essential to understand the sequence of expression associated with these patterns of fibers in order to understand errors that may result in congenital defects in heart formation.

In this study time pregnant (9.5-15.5 days of gestation, ED) Sprague Dawley rats were obtained from Harlan Sprague Dawley Laboratories (Indianapolis, Ind.). Animals were anesthetized with 37% chloral hydrate in normal saline solution and the individual embryos, including uterine tube, were removed and placed in 0.01 M phosphate buffered saline solution containing 0.02 % azide (PBS-A) at 22°C. The embryos were removed from the uterine tube under a dissecting microscope and fixed for 4-8 h in 2% paraformaldehyde in 0.1 M HEPES (N-2-Hydroxyethylpiperazine-N'-2-ethanesulfonic acid) at pH 7.2. The embryos were scored and gestational ages were assigned.

The embryos (9.5-11.0 days of gestation) were stained using a modification of the procedure described by Tokuyasu and Maher (1987) . After fixation, the embryos were placed in PBS-A for 1 h and immersed in 0.1% Triton X-100 in PBS-A for 1 h. After washing for an additional hour in PBS-A, the embryos were incubated in 1500 units/ml bovine testicular hyaluronidase for 45 min, washed in 0.1% Triton X-100 with 0.01 M glycine in PBS-A for 1 h and immersed in 1% bovine serum albumin (BSA) in PBS-A for 2 h. The individual embryos were incubated overnight with rhodamine-labelled phalloidin from Molecular Probe, Eugene, Ore. (10 µl of stock solution/embryo), or control buffer. Following incubation, the embryos were washed in PBS-A at 4°C for 6 h and immersed in 1% BSA for 2 h.

Hearts from 11.5-15.5 day gestation embryos were removed and fixed as above. After fixation, the embryos were encased in 13% polyacrylamide gel using a modification of the procedure described by Hansen and Dryer (1981) . After fixa-tion, the hearts were washed in PBS-A for 1 h and immersed in 0.1% Triton X-100 in PBS-A for 1/2 h. The hearts were embedded individually in blocks containing 13% polyacrylamide with 0.8% bis-diamine which was polymerized with 2% ammonium persulfate. The polyacrylamide blocks containing the hearts were sectioned at 200 µ with an Oxford vibratome model G before being stained as described above. The tissue was examined using a Bio-Rad MRC-600 confocal laser scanning microscope (Biorad, Cambridge, Mass.) using 4× (NA=0.2) and 10× (NA=0.5) objectives.

At ED 9.5 to 10.0, the cells of the myocardium are round and tightly packed with no discernable pattern. The first indication of cell orientation into fiber patterns occurred in the outflow tract and ventricle. The outflow tract myofibers developed circumferentially and maintained this pattern throughout the study time. The ventricular myofibers also developed circumferentially; however, they gradually changed into a net-like pattern ( Fig. 1) followed by a thickening of the ventricular wall and protrusion of trabeculae. With continued maturation, the ventricular trabeculae appeared to coalesce especially in the re-gion of the muscular interventricular septum (Fig. 2) . The myofiber pattern developed later in the atria. The atria expressed the same early pattern seen in the ventricles; however, they retained the net-like appearance throughout development. The results indicate that there are region-specific differential changes in the orientation of myocytes that result in the unique myofiber patterns of the heart. These regional pattern changes may be correlated with mechanical forces and hemodynamic alterations during development.

The preparation of thick, optically clear sections of fragile tissue structures greatly aids the power of confocal laser scanning microscopy in imaging three-dimensional structures. Hale and Matsumoto (1993) have presented confocal images of agarose-embedded, vibratome-sectioned tissues; earlier workers have embedded soft tissues in polyacrylamide gels for frozen sectioning and lectin or immunohistochemical staining of structures in 5-10 micron sections (Bronson et al. 1991 , Hausen and Dreyer 1981 , Johnson and Blanks 1984 . We report here conditions for embedding, sectioning, and staining embryos in polyacrylamide gels for a variety of confocal imaging techniques.

In general, infiltration of 1-8 mm specimens for 1-3 h in a mixture of 10-15% acrylamide monomer (1 part bisacrylamide cross linker to 40 parts monomer) yields, upon polymerization (with 0.1 volume of 2% ammonium persulfate), blocks that cut easily by vibratome between 50-1000 microns. These conditions work well for tissues previously stained (with fluorescent probes or DiI tattoos) or for staining gel sections water-soluble fluorochromes with low molecular weight (e.g., propidium iodide, phalloidin). For immunostaining after sectioning, the acrylamide concentration must be reduced to 2-3% acrylamide to facilitate access of immunoglobulins to antigenic sites, and the gel must be supplemented with 1% agarose to aid sectioning and handling.

Illustrated below are confocal images from acrylamide sections of a Stage 11 chick embryo, fluorescence-immunostained in whole mount for cadherins. This specimen was fixed in 20% DMSO/80% methanol (Dent and Klymkowsky 1987) and incubated overnight with a 1:50 dilution of commercial antiserum against a synthetic peptide common to all known cadherins (Sigma, Pan Cadherin, #C3678). Rinsing preceded overnight reaction with TRITC-conjugated secondary antibody, brief formaldehyde fixation, and embedment in 15% acrylamide. A survey ventral view was taken before transverse vibratome sectioning across the entire embryo at 100 micron intervals, yielding eight sections that were spanning the thoracic region and were mounted in 50% glycerol for closer inspection. Figure 1 illustrates a stereo pair of projections from a series of confocal images across the atrioventricular junction. This gel-embedding and vibratome-sectioning method yields abundant, optically clear, and easily handled sections for confocal examination of fluorescent structures in water-miscible media. Greater detail concerning procedures and technical problems with this technique are provided elsewhere (Germroth et al. in press) .

Department of Anatomy and Cell Biology, SUNY Health Science Center, Syracuse, New York, USA Endothelial cells arise in the early embryo from precurser cells called angioblasts. In the quail embryo, the emergence of these cells can be observed as an epithelial to mesenchymal conversion of cells from the mesoderm which may be observed by scanning electron microscopy (SEM) after removal of the endoderm 1 or by immunolabelling the whole embryo or sections with a monoclonal antibody (QH-1) which labels angioblasts. 2 The process of angioblast migration and assembly into the solid cords, which are the rudiments of the earliest blood vessels, is called vasculogenesis. 3 Simple embryological manipulations have been used to distinguish the role of cell migration in the early vessel rudiments. The dorsal aortae arise ventrolateral to the forming somites, and inserting blockages revealed that little cell migration is involved in their formation. The endocardial cells which form the endothelial cell lining of the heart, in contrast, undergo an extensive migration from more lateral regions of the embryo which has been studied by blockages and the construction of quail/chick chimeras. 4 IV-64 Scanning Vol. 16, Supplement IV (1994) The first rudiments of veins in the embryo are the cardinal veins. These form beneath the ectoderm of the embryo in an area with few angioblasts. The early stages of cardinal vein formation have been studied by using immunogold-labelled secondary antibodies after QH-1 labelling and imaging in the SEM backscatter mode after silver enhancement. These studies revealed that small groups of angioblasts appear over the lateral mesoderm, possibly by migration from the angioblastrich mesoderm adjacent to the endoderm, and then migrate medially to assemble into a solid cord of cells in association with the developing Wolffian duct. 5 The ability of scanning probe technology to image atomic topography of a surface, to manipulate individual atoms, and even to probe the internal structure of a molecule confirms the significance of these super-resolution microscopies used at the nanometer-scale analysis of molecular systems. So far, a number of interactions between the sample and the probe tip gives access to a variety of local properties (Wickramasinghe 1992) . The latest applications of the local probe microscopies on organic molecules include photoemission, tunneling potentiometry, electrostatic, elastic and tribo logic properties, or near-field thermal measurements and as many other phenomena on which various contrast formation mechanisms are based. Topics include imaging of individual biomolecules, highly ordered molecular assemblies, and polymeric materials under various environments. The development of additional imaging techniques,based on near-field electromagnetic interaction between the probe tip and the surface (Betzig and Trautman 1992) , or even by means of magnetic resonance (Rugar et al. 1992) , indicates that new technologies with subnanometer spatial resolution could be achieved in principle.

Recent advances in near-field optical microscopy (NSOM) confirmed spatial resolution in the range of 30-50 nm but still limited by the diameter and optical penetration depth of the aperture. Here, we demonstrated a new concept for an aperture near-field scanning optical microscope which combines force microscopy and optical scattering for imaging the sample at subwavelength resolution without the use of an aperture. The end of a silicon tip is illuminated in transmission mode by a laser beam through a transparent substrate. Both the tip and the sample undergo perpendicular motion, each at a different frequency, with amplitudes chosen comparable to the desired measurement resolution. The scattered light from the tip and the surface is detected at the difference frequency for imaging and sample at sub-wavelength resolution without the use of an aperture. We describe the novel experimental scheme and present the most recent results obtained from our system. The resolution demonstrated to date is 3 nm using helium-neon wavelength and optical line scans are shown in Figure 1 . Finally, a consideration of the basic theory demonstrates that much higher resolution can be easily anticipated. These preliminary results firmly establish the great potential of this new near-field optical microscopy for biological research. High-resolution scanning electron microscopy (SEM) studies of enamel crystals from remineralized enamel have provided clues as to the changes in crystal diameters within specific zones of artificially created carious-like lesions. 1, 2 These data have supported the evidence obtained through polarized light microscopy of zones of demineralized and remineralized enamel. The dark zone, viewed in polarized light, has been identified as the zone of remineralization. 3 These initial studies were conducted using sputtered coatings of gold/palladium (Au/Pd). High-resolution images of the crystals within the zone of remineralization revealed large crystallites in excess of 120 µm, often with what appeared to be remineralized "nodules." However, it was difficult to determine whether these were one crystal, or two crystals which had fused during the remineralization phase. A study was conducted using an ultrathin coating of sputtered chromium (Cr) film and SE-I imaging to provide topographic contrasts of crystal surfaces. Since the majority of the SE-I information should be produced close to the surface, the resultant image is specimen-surface specific. 4, 5 Artificial caries-like lesions were prepared in noncarious human teeth (second and third molars) and sectioned into 100 µm sections which were then viewed in polarized light to identify the dark zone of remineralization. The sections were then fractured through the dark zone producing a spicule of enamel with an edge of fractured enamel free from sectioning artifacts. Each specimen was then mounted onto supports with silver paste with its fractured edge facing up. Measurements were then made to determine the location of the dark zone so that it would be easily located within the SEM. Specimens were degassed and then sputtercoated with 10 nm Au/Pd and 2 nm Cr. 6, 7 Specimens were imaged at high magnification in the SE-I mode by placement on the condenser/objective (c/o) lens stage of an ISI DS-130/LaB 6 SEM at 30 kV and a Hitachi S-900 cold cathode field emission (FE) SEM operated at 20 kV.

High magnification SE-I images (S-900 FE SEM) of Au/Pd coated enamel crystal surfaces within the dark zone, revealed particulate features decorated by metal giving it a continuous granular appearance (Fig. 1) . Analysis of a specimen coated with a 1.5 nm Cr film revealed crystal surfaces with well defined particulate features that were clearly delineated and separate from one another.

This study documents the effect Au/Pd coating has upon the examination and evaluation of enamel crystals at high magnifications. Results obtained document that an ultrathin coating of sputtered chromium film of 2-1 nm does not produce the coating artifacts found with conventional sputtered Au/Pd. An ultrathin coating of Cr, together with the use of SE-I imaging, allows surface features to be more accurately imaged and measured. IV-66 Scanning Vol. 16, Supplement IV (1994) FIG . Previous results 1 have shown that the ability of intercalative dyes to modulate the antiviral activity of poly r(A-U) was related to the groove through which the dyes intercalated into the poly r(A-U). When poly r(A-U) was combined with the minor groove intercalating dyes such as ethidium bromide (EB) or the minor/major groove intercalating dyes, optimum enhancement of antiviral activity was observed at a dye/ribonucleotide ratio of 1/4. No enhancement of antiviral activity was observed when poly r(A-U) was combined with the major groove intercalating dyes. When EB was combined with poly r(A-U) and then added to human foreskin fibroblast (HSF), the 50% effective dose of the poly r(A-U) was 154-fold lower. The results of additional studies 2 demonstrated that the enhanced antiviral activity was not due to superinduction of interferon, direct viral inactivation, or host cell cytotoxicity. Phase contrast, fluorescence, and confocal micrographs of HSF cells following a 3-h exposure to 50 µM EB alone or a 50 µM EB/200 µM poly r(A-U) combination stained the nucleolus, but not the chromatin.

Negatively stained transmission electron microscopy (TEM) preparations (Fig. 1a) and replicas (Fig. 1b) (Fig. 2a,b) which may be associated with the enhanced antiviral activity of this combination. Under the conditions employed in this study, poly r(A-U) exhibits an elongated conformation (200-300 nm in length) that possesses a number of hairpin loops as well as single-stranded and double-stranded domains (Fig. 1a,b) . The double-stranded domains are found predominantly at the base of 30 nm hairpin loops. In contrast to the poly r(A-U) alone, micrographs of the EB/poly r(A-U) combination illustrate the presence of condensed structures with diameters ranging from 15 to 85 nm. Results from scanning force microscopy corroborate the results of both TEM preparations. TEM of unstained and uranyl-stained EB/poly r(A-U)-treated HSF cells illustrate the endocytosis of electrondense material into acidic compartments of the HSF cells. 3 Subsequently, the electron-dense material escaped from the acidic compartment and formed electron-dense bodies with dimensions that closely approximate the dimensions of the EB/poly r(A-U) combination visualized in the negative staining preparations. These electron-dense bodies are detected near the nuclear membrane and in the nucleolus. The nucleolus of an unstained, EB/poly r(A-U)-treated HSF cell demonstrates the segregation of the nucleolar components so that the fine fibrillar component, which comprises the nucleolar organizer region, is being peripheral to a dense granular material occupying the major central area of the nucleolus. 3 The results of the current study and our previous work suggest that the elevated antiviral activity of the EB/poly r(A-U) combination may be related to the ability of the EB to complex with poly r(A-U) and condense it into a conformation with dimensions that can be accommodated by endocytotic vesicles. Exit from the acidic compartment is promoted by unbound EB that induces endosomal or lysosomal swelling. Subsequently, changes in the topology and surface charge of the poly r(A-U) induced by the EB allow increased access to the nucleolus which results in the modulation of additional cellular processes, especially rRNA synthesis and processing. Aromatic constituents in plant cell walls are associated with other wall components and affect strength and other characteristics of plant cell walls. Ultraviolet (UV) absorption microspectrophotometry has been useful in characterizing aromatics within specific cell types in plant organs. 1, 2, 3 This technique was applied to walls of bran and endosperm cells in a series of hard and soft wheats.

Kernels from hard and soft cultivars were fixed in glutaraldehyde (4% in 0.1M cacodylate buffer at pH 7.4), and sections were cut with a microtome at 4 µm thickness. The sections were mounted on quartz slides in glycerin and covered with a quartz cover slip. Cell types in the sections were analyzed for UV absorption using a computer-controlled Zeiss UMSP80 microspectrophotometry system. Transmitted illumination was provided by a xenon lamp (XBO 75W) with a connecting grating monochromator using a bandwidth of 5 nm. A 32-× quartz objective lens with a final aperture of 1.56 µm, which was delimited within about onethird of the area of a field-limiting diaphragm to reduce stray light, was positioned over walls of selected cell types. Absorbance of transmitted UV illumination was recorded over a range from 230-350 nm at 2-nm increments. The system was standardized at 350 µm. Spectra were collected, displayed, and evaluated using the Zeiss Lambda Scan software.

UV absorption maxima or distinct shoulders occurred near 240, 280, and 324 nm. While the absorption near 240 nm is not defined, absorption near 280 nm is believed to be due to lignin (i.e., polymerized phenolic constituents) and that near 324 likely is due to ester-linked ferulic acid. 2 Synthetic lignin, from polymerization of coniferyl alcohol with horseradish peroxidase, has a strong absorption at 280 nm and no absorption at higher nm. Further, ferulic acid linked to arabinoxylans, which has been isolated from bermudagrass cell walls, gave a shoulder near 290 nm and a strong max at 324 nm. Other work, using the addition of cinnamic acids to milled lignins 3 or removal of phenolic esters by alkali treatment, 1 supports the above spectral interpretations.

The kernel of wheat consists of several distinct cell types (Fig. 1) . Spectra were obtained of walls of these various cell types (Fig. 2) . Epidermal and parenchymal walls gave the highest absorption and both had spectra indicative of lignin (near 280 nm) and also of ester-linked ferulic acid (near 324 nm). Nucellar epidermis walls have spectral patterns similar to the cell types above, but absorption was much less. Aleurone walls gave spectral patterns indicative of mostly ester-linked ferulic acid and less lignin than previous cell types; the side walls gave a considerably higher absorbance than upper or lower walls. Endosperm cell walls, which enclose the starch/protein matrix, were totally lacking in UV absorption maxima, indicating no aromatic constituents in these walls.

Variations occurred in UV absorption and spectral patterns for walls of various cell types in wheat kernels. However, no consistent differences occurred among the various kernels that might relate to hard-or soft-rated wheats. IV-68 Scanning Vol. 16, Supplement IV (1994) FIG . Food microscopists are continually searching for noninvasive methods of examining the internal structure of foods, especially those with delicate labile structures. X-radiation is of considerable potential because of the great penetrative power and the minimal need for sample preparation. Examination can be carried out at atmospheric pressure, and there is no intrinsic reason why temperature-controlled or temperature-ramped studies are not possible.

The term x-ray microscopy covers several techniques and can be defined as the use of x-rays to produce a magnified image, or to give information about specific microscopically identifiable parts of a specimen.

Our current equipment consist of an electron gun, two electromagnetic lenses (condenser and objective), and x-ray target. X-rays are generated from a 10 µm spot on a tungsten target. The target usually is positioned so that the microfocused electron beam glances the side of the target. X-rays from the microfocal spot form a cone which diverges through a thin window. Focusing is achieved by placing a grid with bar widths of 6 and 12 µm into the beam and adjusting the current in the condenser and objective lenses. The image is captured on a TV monitor. Finally, test radiographs are recorded on x-ray sensitive film. The x-ray radiograph is a shadowgraph and, where the sample is thick, the overlaying shadows will be complicated to interpret. Two methods can be used to help interpret the image. One method is to take stereo pairs of the sample and view these images with a suitable stereo viewer which will show a threedimensional image. The other method is tomography.

In the projection x-ray microscope, the entire specimen is in focus at once with the same resolution, determined by the spot size, although different depths within the specimen are magnified by different amounts. Three-dimensional images can be made by moving the specimen either by tilting or translating it between exposures in the divergent x-ray beam. These images are then viewed with a suitable stereo viewer which will show a three-dimensional image of the structure removing some of the confusing overlay of detail to be found. We have also obtained images from pieces of tissue subject to tensile elongation.

In the design of our x-ray microscope there is considerable distance separating the x-ray film from the specimen, and this space allows the apparatus for tomography to be set up. Tomography is a form of imaging whereby a section or "tome" can be taken through a specimen by using x-rays, and details above and below the sectional area of interest will be blurred out, giving a fairly clear image of this area. This technique is nondestructive to the sample. The method is similar to that de- scribed by Lindegaard-Andersen and Thuesen, 1 originally proposed by Watson. 2 In their method, the x-ray source is stationary and the subject and recording film are rotated synchronously. The x-ray beam strikes the film at near glancing incidence producing transaxial summation images. However, the associated point spread function has a 1/R radial dependency, which means that smearing of detail will have occurred. The results are sufficiently encouraging, nevertheless, to stimulate the search for improved contrast and resolution, with provision for three-dimensional mapping of internal structures.

The images shown here are of food products containing large air cells, giving maximal contrast in the x-ray beam. We have obtained comparable success with water-filled cellular structures such as vegetable and fruit tissue. In the future, we anticipate further enhancement of contrast and resolution by the use of improved x-ray flux, image capture, and image processing.

Centre for Food and Animal Research, Agriculture and Agri-Food Canada, Ottawa, Ontario, Canada Over the past 20 years, electron microscopy (EM) has proven to be a useful tool in assessing the effects of various physical and chemical treatments on the microstructure of egg components (related to the production of egg products, baking, behaviour in various food systems, nutrient bioavailability, etc.) Much of this information has been acquired using transmission electron microscopy (TEM) to view yolk samples which have been biochemically extracted after removal from the egg.

The author studied microstructural changes in yolk during its in vivo digestion by the epidermal cells of the chick embryo yolk sac. Yolk is digested within these cells, as signified by its presence in varying degrees of microstructural alteration, compartmentalized within cell organelles. A temporal sequence can be inferred with the degree of microstructural change observed. It was possible to follow the fate of yolk granules and low density lipoprotein (LDL) particles during digestion, because yolk granules retain their integrity during preparation for EM. LDL particles were made visible by crosslinking action of the fixatives and by the use of a specific enhancement technique.

Initial work focused on the microstructure of native, undiluted yolk which had been fixed using conventional and novel fixatives. Thin sections were examined by TEM. These experiments allowed us to optimize fixation conditions and confirmed earlier results by other workers that yolk is composed of granules which have a subgranular structure and lack a boundary membrane. The granules sit in a suspension of closely packed LDL particles. Yolk granules from incubated eggs were identical to those from unincubated eggs and showed no microstructural changes to indicate that digestion had taken place outside of the cell.

After clarifying that yolk is digested intracellularly, the next phase of the study was the microstructural description of intracellular yolk and yolk sac epithelial cells during incubation, prepared using half-strength Karnovsky's fixative and the imidazole-buffered osmium tetroxide protocol (IBO) of Angermuller and Fahimi (1982) which enhanced lipid staining. Epithelial cells were examined systematically using EM, from apex to base, to study the three processes associated with digestion: uptake, breakdown, and transport. Using TEM, granules and LDL particles were observed to enter the cells in uptake structures, the microplicae and coated pits, respectively. When samples were viewed using SEM (Fig. 1) , these uptake structures appeared as fossae of different sizes, which have not previously been described in the literature. Yolk granules and LDL particles were observed by TEM, within the cells, in phagosomes, endosomes, and acid phosphatase-containing vacuoles (secondary lysosomes), components of an active intracellular digestion system as it is presently known. The secondary lysosomes contained yolk granules exhibiting various degrees of microstructural alteration (Fig. 2) . The microstructure of these intermediates in yolk digestion appears to be very similar to that appearing in the literature describing known biochemical manipulations of yolk ex ovo, and indicates that a re- IV-70 Scanning Vol. 16, Supplement IV (1994) FIG. 1 SEM of apices of yolk sac epithelial cells from an embryo incubated for 8 days. Yolk granules have entered the cell, and a fossa (f) is observed. mp = microplicae.

lationship may exist among structural, functional, and biochemical information.

The results of these individual experiments were used to propose a scheme of yolk digestion based on progressive microstructural changes of intracellular yolk. In addition, a transport system for yolk lipid and its digestion products to the embryo was microstructurally demonstrated using IBO. This transport system shared ultrastructural characteristics with that reported for lipid transport in intestinal cells, especially with respect to the formation and transport of chylomicronlike particles. Laser scanning confocal microscopy and digital image processing were used to visualize the pattern of de novo blood vessel formation in the quail embryo. Stage 9 quail embryos were fixed on the yolk, excised, then immunolabeled with QH1, an antibody marker for angioblasts and vascular endothelial cells . After incubation in fluorescein-conjugated secondary antibodies, the embryos were mounted in PBS/glycerol (1:9 ratio) under a #1 glass coverslip. Specimens were examined on a BioRad MRC1000 confocal laser scanning device equipped with Zeiss optics. A 10× objective lens was used to image the entire caudal half of each embryo during the laser scanning step. Ten 7 µM optical sections were acquired in a plane parallel to the embryonic axis (en face). The ten optical sections were then collapsed into one composite image using BioRad's proprietary software (Fig. 1) . The image was imported into Photoshop ™ software on a Macintosh Quadra 950. Using this software, the single composite fluorescence image was processed with the "emboss" tool (Fig. 2) . Other image processing routines such as edge tracing, sharpening, and contrast enhancement can also be applied with ease (not shown).

In contrast to conventional microscopy, this procedure yields a map of the entire endothelial network of the quail embryo in sharp focus, and with a highly favorable signal-to-noise ratio (Fig. 1) . While some geometric distortion occurs when ten sections are collapsed into one imaginary plane, the flat- (1) and LDL particles (2) are found in separate organelles, and are also found together within the same organelle (3) . Myelin bodies (m) and lipid drops (l), products of intracellular digestion of yolk, are also observed.

FIG . 1 This immunofluorescence image depicts a map of all the vasculature elements in the caudal half of a developing quail embryo (stage 9). The wide vessels on each side of the embryonic axis are the dorsal aortae; lateral to the aortae are two fields of rapidly forming vascular networks.

tened nature of the early avian embryo minimizes this problem. Also, since the QH1 epitope (a carbohydrate) is present on multiple cell surface molecules, the image gives a reasonable approximation of vessel morphology and the protrusions of the vascular endothelial cells. The rendering of the data into a pseudo three-dimensional relief provides the observer with a more familiar visual format (Fig. 2 ). The type of image shown in Figure 2 facilitates comparison of endothelial sprouts and anastomotic foci during formation of the first vasculature in the quail embryo. Based on images from embryos at progressively older developmental stages, we suggest that morphogenesis of the lateral anastomotic network occurs by mechanisms that involve angioblastic tractional structuring of the extracellular matrix (Madri et al. 1983 , Montesano et al. 1983 , Stopack and Harris 1982 . This hypothetical mechanism more recently has been elaborated upon by Vernon and colleagues (1992) . According to the latter workers, cellular responses to morphogenic cues within the highly planar extracellular matrix underlie early vascular patterning.

In addition, the digital imaging approach shown here offers an improved method of comparing normal vasculogenesis with experimentally produced malformations (Drake et al. 1992) . (5.6-5.9 ). Chemical composition and pH did not change during extended heating. The SHMP sample was the firmest and the SC sample was initially the softest. The firmness of the SC and TSP samples increased with extended heating whereas the firmness of the DSP sample did not change. Meltability decreased with extended heating in all samples, except the SHMP sample which had the lowest meltability of all samples throughout the experiment.

There were marked differences in the microstructure of the process cheese protein matrices after heating. Osmiophilic areas gradually developed during heating and their incidence was related to the melting salt used. After 5 h of heating, they were numerous but considerably smaller in the DSP sample ( Fig. 1 ) than in the TSP sample (Fig. 2) . The highest and the lowest incidence was in the SHMP and the SC samples, respectively, which were the samples with the lowest and highest meltability. New York, USA Mozzarella cheese contains parallel protein fibres created by stretching the curd in hot water during manufacture. The protein fibres are oriented parallel to the direction of extrusion and are separated by milkfat or whey (Kaláb 1977 , Taneya et al. 1992 . By varying manufacturing parameters and storage time, cheeses with different sensory/functional properties, as well as with different microstructural characteristics, can be produced (Kiely et al. 1992 ). In the present study, the impact of varying the stretching temperature (54°C, 60°C, and 66°C) and storage time (3 days, 4 weeks, and 7 weeks) at 4°C

on the microstructure of Mozzarella cheese was investigated using light microscopy (LM) and scanning electron microscopy (SEM).

For both LM and SEM, samples were fixed in 4% glutaraldehyde in 30 mM PIPES containing 1 mM CaCl 2 , before cryosectioning (10 µm sections). Sections were cut both parallel to the protein fibres (longitudinal) and across the fibres. For LM, sections were stained with ice-cold oil red O and methyl- ene blue. Because the fat is not well fixed by aldehyde fixation, slides were held on a bed of ice during staining to prevent fat migration. For SEM, the sections were stained with uranyl acetate in methanol, washed in methanol, and stained with lead citrate, dried, mounted on aluminum stubs, and gold-coated. This procedure removed most of the free fat and water/whey, allowing the association of the protein fibres to be clearly visualized. Increased stretching temperature increased the size of the fat globules, which was particularly noticeable in cross sections. With 54°C stretching, the fat globules were relatively small and uniformly distributed across the section, while with 60°C and 66°C stretching there were almost two populations of globules: some small and some very large. In longitudinal sections, the protein strands formed during stretching at 54°C and 60°C were slender and smooth, while those formed during stretching at 66°C were thicker and less regular. Little difference was seen with storage time in either longitudinal or cross sections for the three different stretching temperatures using LM. With SEM, however, after removal of the fat, the ability of the fibres to hold together decreased with storage time. This is best seen in longitudinal sections, where the protein fibres are closely associated in cheese stored for 3 days (Fig. 1) , and are able to withstand the stress of drying, while in cheese stored for 7 weeks (Fig. 2) , the fibres have been weakened (presumably by proteolytic activity from coagulant and culture enzymes) and pulled apart as the section dried.

Indian Dairy Association, Sector-IV, R.K. Puram, New Delhi, India

Milk is a canvas of seemingly silent molecular structure. There is a large variety of "resident molecules" with molecular harmony in milk which are of mammary gland origin. In general, the "structural matrix" of milk comprises three classes of biomolecules: molecules in suspension (casein micelles), molecules in solution (lactose, proteins, vitamins, and salts) , and molecules in emulsion (fat globules). Because of its excellent nutritive composition, milk is ubiquitous and the most popular "ready-to-consume" food from time immemorial. Today, in IV-74 Scanning Vol. 16, Supplement IV (1994) FIG . different parts of the world, milk from various species of animals is used for food. In the U.S., however, the cow furnishes virtually all of the available market milk, whereas in my country more than 55% of the total milk produced is of Buffalo origin. Hence, structural studies particularly on milk and milkfood products made from various animal species are more fascinating and challenging from a global angle. It is the local consumer habits that determine the final structural orientation of the product in the country of origin, especially in a country such as India where 50 to 55% of milk produced is converted into a variety (30 plus) of traditional milk products, using processes such as coagulation (heat and/or acid), desiccation, and formulation. Hence, a presentation on the "structural style" of milkfoods which are developed on the basis of "environment-friendly green technology" in the tropical countries might generate a new dimensional need for future studies on milkfood products. A comprehensive literature scan of published papers in the areas of "Food" and "Food Structure" during the last 30 years reveals the following publication status: During the period 1968 to 1991:

1. Global: Of the total number of 38,900 food articles published, 10,712 are related to "Food Structure." The number on "Food Microstructure" is 658. When I think of their contributions, I claim no merit for my presentation in this conference today. These data provide a "food-for-thought" why food structural studies do not receive the desired attention. We may ponder this issue.

While preparing this presentation, the excellent review article by Kalab 1 on Food Structure and Milk Products was extensively consulted. Keeping in view the existing and reported studies on the subject, the situation in the tropical countries, particularly regarding Indian milk such as buffalo milk, deserves special attention.

According to Kalab 1 the structure of milk and milkfoods determines their properties such as firmness, spreadability, elasticity, viscosity, and susceptibility to syneresis, which are globally recognized as texture. The understanding of the processes both conventional and traditional and their relationship between "structural style" and "textural mood" 2 is important in product development. In the tropical countries, the preprocessing exposure of milk to the environment, that is, temperature, microbes, and humidity has to receive extra attention. The dynamic status and kinetics of molecular interaction in different species of milk should be a vital area of structural studies in Asian countries.

The structural studies on milk and milkfood are generally conducted by use of microscopy. The methods used are optical microscopy such as fluorescence microscopy and confocal scanning laser microscopy, and electron microscopy consisting of two major types: scanning electron microscopy (SEM) and transmission electron microscopy (TEM). These techniques meet the particular requirements of the study. The application and findings derived using SEM and TEM in various milk food systems have been extensively covered recently by Kalab 1 . Repeating and reproducing the published data may not be necessary for the participants in this conference. However, a brief microcosm might be refreshing, and is hence documented below:

1. Milk casein micelles (100-300 nm), being small, cannot be seen using a light microscope, but TEM shows them as globules that are apparently composed of submicelles. 2. The casein micelles, if deprived of their stabilizing factor, aggregate and form a gel of a regular structural "crowd." 3. In the case of Acid-gel, the micelles start disintegration as a result of the solubilisation of their structural partners such as calcium phosphate. 4. There is an interstructural and intrastructural relationship between the suspended casein micelles and soluble Blactoglobulins. Heat disturbs their stability equilibrium and their molecular relationship. 5. Among the milkfoods, cultural milk products like yoghurt are highly influenced by heat in relation to their "set-style" and "stirred-style." The Indian counterpart, "DAHI," offers its advantage when made from buffalo milk because of its high curd tension due to high calcium level. This area deserves a close look using SEM technique.

6. As far as cheese is concerned, because of its inherent multinatural processing steps using chymosin on one hand and starter culture on the other, it is an evergreen exploratory platform using either SEM or TEM. 7. Dried milkfoods as spray-dried milk particles are globular of larger diameter fat-globules, however, they undergo some changes as revealed by SEM studies. Lactose plays a commanding role in a so-called instantisation process. While making a presentation before a "highly-structured" qualified scientist using SEM and TEM, it is tempting for me to present our own work of limited nature. In our laboratory, during the last decade, we carried out a SEM study on Buffalo milk and its casein. 3 The micelles are of larger size and have higher-bound calcium unlike cow casein micelles. Regarding the casein fractions, an elaborate study on the B-casein in relation to structure and function as affected by heat and enzymic cleavage were carried out. Work relating to microstructure development on another Indian milk product, namely "PANEER," from buffalo and cow milk has been reported. 4 More recently, while working on casein micelles in CSIRO Dairy Research Laboratory in Australia, the author made an original contribution when ascertaining the micellar integrity of casein micelles in milk. 5 He used a very simple technology not dependent on SEM or TEM. It was on the "colour communication" property of molecules as an integrated system or in isolation. Using this new technique, it is possible to declare the miscellar integrity and to identify milk from different animal species, based on the different miscellar structure-dependent light-reflecting phenomena.

The time has come to develop a "global strategy" on milk structural studies with a view first to understand the local product and then to ensure marketing of "rightly-structured milkfoods" of the right type to meet the countries' marketing needs.

Starch is a major component of many foods and is commonly used to modify the texture of a system. Starch gels are composites of swollen gelatinized granules embedded in a continuous amylose network. The rheological characteristics of the starch pastes or gels depend on the shape and swelling power of the granules, amount of amylose and amylopectin leached outside the granule, network entanglement, and interaction between the paste components.

Many food systems also include in their formulations a lipid phase, thus, starch-lipid interaction becomes a major concern in starch paste rheology. The interactions of triglycerides might be different from those of monoglycerides due to their amphiphilic character, dispersion states, and steric hindrances. Triglycerides owe their characteristics to their fatty acid composition (chain length and insaturation) and distribution (Davis et al. 1986) .

The objectives of this work were to analyze the effect of different lipid phases on the swelling power of the starch granules, as well as to analyze the rheological behavior of the starch pastes (apparent viscosity and viscoelasticity).

Commercial corn starch (CS) (Refinerías de Maíz, Argentina) was used as thickener. The lipid phases used were: (1) shortening (SH) (Molinos Río de La Plata S.A., Argentina) containing 71.6% of mono-and 1.5% of polyunsaturated and 24.0% of saturated fatty acids; (2) sunflower oil (SO) (Molinos Río de La Plata S.A., Argentina) containing 28.6% of monoand 60% of polyunsaturated and 8% of saturated fatty acids, and (3) commercial glycerol monostearate (GMS) Mivaplex 600 (Eastman Kodak Co., U.S.A.), which contained 90% of monoglycerides and 5% of diglycerides.

Seven percent starch suspensions with and without lipid phase (5% of SH or SO or 1% of GMS) were prepared using a modification of the method of Eliasson (1985) . Swelling power was calculated as the weight of sedimented gel divided by the original dry weight of starch.

Samples were placed on slides and micrographed in a Leitz Ortholux II microscope with a photographic camera Leitz Vario Orthomat (Leitz, Germany).

The suspensions were gelatinized in a thermostatic bath at 90 ± 0.2°C under standardized conditions. A rotational viscometer Haake Rotovisko RV2 (Germany) with a sensor MVIP of concentric cylinders was used. The measurements were performed at 60°C and the transient shear stress curves (τ vs. time) were obtained at different constant shear rates (D) from 16 to 024 s −1 . Apparent viscosities were calculated as τ/D ratio at D= 512 s −1 . IV-76 Scanning Vol. 16, Supplement IV (1994) Table I shows the swelling power obtained from the starch suspensions with and without lipid phase. When either of both triglycerides, SH or SO, was added, the swelling power values were higher than the value of the control (CS). The SO, with the highest insaturation content, presented the higher swelling power. The addition of the monoglyceride (GMS) led to the lowest value. The relative sizes of the swollen granules shown on the micrographs of Figure 1 confirm these observations. GMS forms a helical inclusion complex with amylose, which might delay the transport of water into the granule and consequently decreases the swelling power (Eliasson 1985) . The formation of inclusion complexes between the triglycerides and the amylose seems to be difficult because of the steric characteristics of these lipids.

Apparent viscosities correlated well with the swelling power (Table I) ; the larger the swelling power the higher the viscosity obtained from the rheological curves at long shear times.

Bird-Leider model (Dickie and Kokini 1982) was used to analyze the viscoelastic behavior and the shear time dependence of the different pastes:

where n and m are the power law parameters and b and c are adjustable parameters related to the viscoelasticity and to the structure breakdown of the samples. A satisfactory goodness of fit was obtained (R 2 min= 0.975) (Fig. 2) . When the swelling power increases, b parameter decreases (Table I) .

Since viscoelasticity (b) depends mainly on the network entanglement (leached amylose) and the volume of the swollen granules, both factors should be considered to explain this behavior. Because of the health risk and other environmental factors that airborne grain dust presents to the working population, our laboratory has initiated studies on the isolation, identification, and characterization of bacteria that possess multiple resistance to a series of antibiotics and other compounds (e.g., insecticides and pesticides) at the molecular level.

Samples of grain dust were collected from various grain elevators in the Duluth-Superior regions of the U.S. during the diverse growing seasons. Each sample collected consisted of a heterogenous population of constituents that vary with encountered geographic, climatic, and handling differences. In addition, the geographical growth regions and the mechanism of storage of grains appear to be directly associated with the microbial flora and occurrence of toxic substances.

Scanning electron micrographs of the concentrated grain dusts were morphologically consistent with the observation of previous investigators. The dust from various plants consisted of a distinct assortment of particles; small husk fragments or pericarp (seed coat in case of flax) and "trichrome-like particles" were also present. Numerous bacteria spores were seen at high magnification, particularly durum wheat and barley. Light photomicrographs showed a heterogenous population of both gram negative and gram positive bacteria. The bacteria consisted of different shapes such as short rods, long rods, and cocci. Transmission electron photomicrographs revealed the isolated strains to consist of one or more flagella attached to the membrane surfaces. At least three of the bacterial strains isolated were encapsulated. We exposed, selectively, twelve of the isolated bacterial strains to a variety of antibodies, pesticides, and insecticides. As a result, 5 of the 12 bacterial strains tested showed resistance to both ampicillin and bacitracin (50 µl/mg). Of the 12 strains, 7 showed resistance to insecticides (SEVIN) and the pesticides (ENFORCER) as high as 100 µl/mg. Three of the five strains that were resistant to both ampicillin and bacitracin were also resistant to the insecticide SEVIN at high concentrations (100 µl/mg). These data suggest that bacteria found in grain dust may be directly or indirectly related to occupational health disorders. Chlordiazepoxide (librium) is a commercial antianxiety agent, but its use has not been associated with any striking health improvements. Nevertheless, millions of chlordiazepoxide (CDZ) prescriptions have been written and the drug is widely employed clinically as a muscle relaxant, anticonvulsant, anxiolytic, and hypnotic. Adverse effects vary from skin rash, nausea, headache, and impaired sexual functions to vertigo and lightheadedness, as well as complications from the drug's administration during pregnancy. The ciliated protozoan, Tetrahymena pyriformis, has been of significant importance to research since its successful axenic culture 70 years ago. The drug's tranquilizing effects on humans may possibly be elucidated by investigating CDZ-induced impairment of the protozoan's growth and motility through modifying microtubular-directed ciliary function (Bell et al. 1994) . Here, uptake, recovery, and attempted localization of CDZ, as well as its alterations of cellular ultrastructure are reported. Thin layer chromatography of CDZ (Hoffman LaRoche, Nutley, N.J.) in three different mobile phases revealed optimal CDZ stability when dissolved and stored in isoamyl alcohol; benzene. After 3 days of liquid culture, T. pyriformis removed 15% of the radioactivity from [14C] -CDZ in the growth medium. Liquid scintillation counting of cell washes during the 5-day time course suggested surface, nonspecifically-bound radioactivity. Following 5 days of culture, [ 14 C] -CDZ together with its metabolites and/or breakdown products were recovered from homogenates of Tetrahymena. To detect the intracellular sites of CDZ localization and possible action, both transmission electron microscopy and immunoelectron microscopy were performed. The former revealed that CDZ disrupted cytoplas-buffer or cacodylate-buffered DOPA revealed DOPA/PPO reaction product but failed to reveal a definitive, substrate localization of the enzyme. Instead, cytoplasmic morphologic distortions of buffered DOPA-treated hyphae were apparent, mandating modification of the employed higher-plant, cytochemical procedure for localizing fungal PPO. Thus, attempts to establish the route of PPO secretion to the growth medium utilizing liquid cultured hyphae may be of limited value as the sheath appears to be sloughed during shake culture. In this connection, C. versicolor grows upon a solid substratum when decaying timber in situ. Support: DOE-BCTR Program. The evaluation of surface topography is conventionally done with the scanning electron microscope (SEM) by visually reconstructing stereo images in a stereo viewer. The monocular clues in an SEM image, shown for example in the image in Figure 1 , can also be used to map the topography of a sample surface. The first attempt to identify the surface gradient from a secondary electron detector in this way was reported by Suganuma (1985) who found that (1) and described the inclination of the surface to the specimen plane, where A and B are the signal outputs from the two detectors and A n and B n are the signal outputs from the same material when it is perfectly horizontal.

Although it is possible to add additional detectors and to modify their signal outputs in a conventional electron microscope, the technique can also be implemented in a straightforward manner by using the differences in signal intensity from the two halves of a split backscattered detector. Thus if A and B are the signal outputs from the two different halves of the detector, the backscattered signal, which is usually collected as topographic information (A − B) or composition information (A + B), can be multiplied together (A 2 − B 2 ) to produce the basis of the empirical relationship described by Suganuma IV-80 Scanning Vol. 16, Supplement IV (1994) FIG . (1986) to fit the contrast variations. Indeed, since the denominator in this equation is also a constant and there is invariably a flat spot somewhere in the image, it is easier to simply use the expression

to try and describe the surface gradient. This function can then be scaled to an absolute magnitude of about 60 to take maximum advantage of the contrast variations that exist.

The simplest way to evaluate the surface topography without actually modifying the microscope is to process a digital image in a computer. Such an image can be collected directly or a Polaroid micrograph can simply be scanned into the machine. In either event the processing is fairly straightforward to accomplish the use of software such as the NIH Image program.

The surface profile is obtained by integrating the surface gradient over the scan distance, and in Figure 2 we show, by way of an example, a faceted surface displayed as a conventional secondary image, as a topographic profile, and as a grey scale image indicating the faceted surfaces with the same orientation to the electron beam direction.

Suganuma T: Measurement of surface topography using SEM with two secondary electron detectors. J Electron Microsc 34 (4), 328-337 (1986) Corn kernels of an unknown Italian cultivar were surfacesterilized in full strength bleach (5.25% sodium hypochlorite), rinsed in sterile distilled water, and germinated on Difco potato dextrose agar (PDA) under fluorescent light at room temperature. Isolation of bacteria was made either directly from the rhizosphere of seedlings or from colonies produced on agar. Bacteria were identified by their fatty acid profiles (Microbial Identification System, Newark, Del.) and diagnostic biochemical test (Micro-ID, Durham, N.C.) . Fusarium moniliforme used to test for bacterial antagonism were isolated from corn.

Experiments designed to infect seedling roots with the isolated bacterium were performed on corn kernels that were subjected to a double sterilization process in which the kernels were surface-sterilized with bleach and then subjected to a mild heat treatment to remove internal bacteria and fungi. This process produced sterile seedlings which remained so at least up to the 3 to 4 leaf stages of growth. The corn cultivars used were Trucker's Favorite, Silver Queen, Reid's yellow dent, and an inbred cultivar, PR.

Samples were fixed in 4% glutaraldehyde in sodium cacodylate buffer and postfixed in 2% osmium tetroxide in buffer. The tissues were dehydrated in an ethanol series, critical-point dried and coated with gold-palladium or embedded in Spurr's medium and sectioned. Samples were examined using a Philips 505 scanning or JEOL 100CX transmission electron microscope.

The bacteria isolated from roots of the Italian corn cultivar were gram-negative rods and were further characterized as being positive for Voges-Proskauer reaction, nitrate reductase, and ornithine decarboxylase and B-galactosidase. The isolate fermented arabinose and malonate. It was negative for phenylalanine deaminase, lysine decarboxylase, urease, H 2 S, and indole production. It lacked the ability to utilize adonitol, inositol, and sorbitol, and was negative for esculin hydrolysis. These biochemical characteristics served to place this bacterium in the family Enterobacteriaceae, the tribe Klebsielleae, and it was identified as E. cloacae according to Ewing (1986) .

Microscopic studies established that the bacterium was endophytically associated with the roots of the corn cultivar. Scanning and transmission electron microscopy demonstrated that the bacterium was distributed uniformly over the corn root epidermis but was randomly distributed intercellularly in the root cortex and outer margin of the pericycle, usually adjacent to phloem cells. Although there was a proliferation of bacterial cells in the intercellular spaces of roots (Figs. 1, 2) , there was no evidence of damage to host cells, decline in seed germination, nor seedling growth from E. cloacae infection during the 3week observation period. The presence of the bacterium on kernels of several corn cultivars enhanced the growth of corn seedlings and inhibited growth of F. moniliforme when the bacteria-infected kernels were germinated in fungal-amended soil.

This indicated that E. cloacae is biologically associated with corn plants.

The bacterium exhibited strong antagonism to several F. moniliforme isolates when grown on nutrient agar and PDA plates. The fungus growth was inhibited and restricted to aerial growth on agar plates. The endophytic nature of E. cloacae in corn roots and its antagonism against isolates of F. moniliforme indicate that this bacterium has potential for the biocontrol of F. moniliforme, a corn pathogen. Tracheal organ cultures have been widely used for the demonstration of the cell-and organ-destroying capacities of bacteria and viruses colonizing the respiratory tract. However, the assessment of the cell-damaging abilities of fish parasites, the gill organ culture appeared to be a suitable tool (Stadtländer and Kirchhoff 1989) .

Piscine gill epithelium represents the relevant target tissue for cytotoxicity studies of fish parasites, but this tissue is diffi- IV-82 Scanning Vol. 16, Supplement IV (1994) FIG cult to cultivate in vitro. Excessive mucus production complicates the investigation with the light and electron microscope. This is even more difficult when infected specimens are investigated. Here, cell debris due to released cells and cell fragments require special attention of the investigator to obtain interpretable electron micrographs. A fine balance is required between preservation of the status quo of the in vitro situation and the necessary procedures (multi-step preparation) for scanning electron microscopy (SEM).

In this abstract, we describe the detailed and improved preparation of piscine gill epithelium for SEM. This will allow other investigators using piscine tissue for research to obtain scanning electron micrographs with high resolution at high magnification.

Gill filaments were obtained from rainbow trouts (Salmo gairdneri, Richardson) and were infected in vitro with Mycoplasma mobile 163 K, a wall-less prokaryote causing severe damage in gill organ culture (Stadtländer and Kirchhoff 1989) . Each gill filament (noninfected and infected) was rinsed in several changes of 0.1 M cacodylate-trihydrate [cacodylic acid buffer (CB)] at pH 7.3 to free the tissue surface from mucilage, medium, and unattached mycoplasmas. The washing step turned out to be critical for obtaining satisfactory results. Excessive washing led to complete removal of cell debris and did not reveal the same status quo of infection as seen with the light microscope during the in vitro cultivation of gill filaments. On the other hand, insufficient washing complicated the interpretation of electron micrographs, especially the identification of M. mobile on gill epithelium, due to excessive mucus and cell debris released from infected tissue.

The pre-fixation was performed with 1.5 % glutaraldehyde (v/v) (in CB) for 24 h at 4˚C. The aldehyde was removed by careful rinsing in CB (three times for 15 min). Samples were not postfixed, but dehydrated in acetone covering a graded series of solutions (acetone/water) from 30 through 100% (v/v). All specimens were critical-point dried (CPD) using carbon dioxide as the transitional fluid. CPD was performed in a Critical Point Drying Apparatus E 3000 (Polaron Equipment Ltd., England) by filling the chamber with 2/3 of liquid carbon dioxide. The heating process was conducted in 1˚C per min steps until the critical point was reached (32 o C and 72 bar). Samples were immediately placed in an exsiccator containing anhydrous calcium chloride to avoid rehydration of samples by air contact. After being attached to aluminum mounting stubs using double-sided sticky tape, each specimen was coated with a layer of 20 to 40 nm of gold in a Hummer V Sputter Coater (Technics Inc., Alexandria, Va., USA). Samples were examined in an Etec Auto Scan B1 (Etec, Calif., USA) operating at an accelerating voltage of 3.1-50 kV. Results were documented on a Polaroid 4×5 Land Film (type 55, positive-negative).

The method described above allows the detailed investigation of noninfected and infected piscine gill epithelium at high magnification with an SEM (Fig. 1, 2) . Good structural preservation of noninfected, apparently intact secondary lamellae (Fig. 1) and destroyed lamellae after infection with M. mobile (Fig. 2 ) document the usefulness of the described method for the study of the infectious process with this mycoplasma on fish tissue and will certainly also give convincing results in investigations with other fish pathogens. The arrows in Figure 2 indicate the presence of the flask-shaped mycoplasma on damaged piscine gill epithelium. Bars represent Figure 1 :10 µm, Figure 2 :20 µm.

For this study, the SEM was superior to light microscopy (LM) and transmission electron microscopy (TEM). LM does not demonstrate details in the damaged tissue and TEM is much more laborious and time consuming. The mammalian bladder functions in urine storage and expulsion and is thus subject to alternating distension and contraction. Blood flow in the bladder wall is compromised by distension (Dunn 1974 , Levin et al. 1983 , and acute and chronic overdistension result in ischemia and necrosis, respectively. The latter often leads to loss of mucosal integrity. Yet, the vasculature of the bladder wall has rarely been studied (Inoue and Gabella 1991) except at the gross level. In this study we describe several unique features of the bladder microvascular anatomy using light microscopy (LM), transmission and scanning electron microscopy (TEM and SEM), vascular corrosion casting, and alkali digestion (Takahashi-Iwanaga 1986).

Twenty four male New Zealand white rabbits were anticoagulated, anesthetized, and cannulated via the abdominal aorta. For routine LM, TEM, and SEM, the bladder vasculature was flushed with buffered saline at physiologic temperature and pressure, then perfuse-fixed with buffered glutaraldehyde; for corrosion casting, the vasculature was filled with resin (Mercox/methylmethacrylate monomer, 4/1); and for alkali digestion, tissue was treated with 6N NaOH at 55˚C for 20 min. Thin sections were stained with lead and uranium, and casts and digested tissues were mounted on SEM stubs with silver paste or colloidal carbon and viewed at 10-20 kV.

The bladder is supplied by left and right vesicular arteries, branches of the internal (60%) or external (40%) iliac arteries. Within the adventitia, the vesicular arteries send coiled branches dorsally and ventrally over the bladder surface. Secondary arteries penetrate the muscularis to supply a rich capillary plexus (Fig. 1) closely apposed to and located in grooves within the base of the transitional epithelium (Fig. 2) . Capillaries measure 4-10 µm in diameter and are often fenestrated and invested with pericytes. Veins exhibit abundant valves primarily in the basal half of the bladder. The lamina propria of the bladder consists of very loose, flexible, collagenous connective tissue. Unusual capillary "glomeruli," associated with accessory vessels paralleling the primary bladder vessels, are present in the adventitia on either side of the bladder wall. These glomeruli consist of one to four contiguous capillary spheres.

These various methods provide a clear, three-dimensional view of the microvasculature of the rabbit bladder, reveal the very close association between the urothelium and the underlying capillary plexus, and describe the fine structure of the mucosal capillaries. Several unique features of the bladder vasculature including capillary glomeruli require further char- IV-84 Scanning Vol. 16, Supplement IV (1994) acterization. The latter may be associated with sensory ganglia related to pressure sensation, but their function has not been determined. These results form the basis for comparison of normal bladder vasculature with that of experimentally compromised vasculature. The epithelial cell layer lining the nasal turbinates of humans is functionally and histologically consistent with that of the lower airways. This anatomic site is easily sampled using an inexpensive, disposable curette, and the epithelium obtained can be evaluated for both clinical and experimental objectives. The original rationale for ultrastructural evaluation of clinical specimens of nasal epithelium in this facility was to document index ciliary lesions consistent with the diagnosis of primary ciliary dyskinesia (PCD) (Figs. 1, 2 ) among individuals considered at risk. Patients referred for study encompassed both adult and pediatric populations and had lifelong histories of chronic sinusitis, bronchiectasis, and/or otitis media, but with normal immunoglobulin levels. Several patients presenting with situs inversus, polysplenia, or infertility problems-clinical findings considered risk factors for PCD-also were evaluated. Approximately 25 nasal biopsies are submitted annually for evaluation, of which approximately 8% demonstrate impaired ciliary motion and ultrastructural abnormalities that could be the basis for altered ciliary motion and a diagnosis of PCD. This finding is consistent with the suspected prevalence of PCD in the general population although the rate appears higher among individuals of Scandinavian descent, Pacific islanders, and inbred populations. The diagnosis of PCD is confirmed by the electron microscopic documentation of any of three ultrastructural level lesions of airway cilia. These lesions are: (1) dysmorphology of dynein arms, (2) absent radial spokes, and (3) microtubular transpositions. In our experience, dysmorphology of the dynein arms represents the major form of ciliary abnormality associated with PCD. Missing radial spokes and microtubular transpositions have not been documented among any of our patients. These clinical studies also have provided a unique opportunity for defining a spectrum of ciliary abnormalities which distinguish the heritable primary ciliary abnormalities associated with PCD from acquired cil- iary abnormalities associated with chronic disease or acute injury due to infectious processes or exposure to irritant gases in the ambient air. In addition, other histopathologic features such as mucus cell hyperplasia have been encountered occasionally among children presenting with chronic respiratory disease and referred for evaluation. In summary, our experience shows that ultrastructural evaluation of nasal epithelium can provide a clinically significant perspective on respiratory health. This work was supported by SCOR grants HL19171, HL34322, and HL42384 from the National Heart, Lung, and Blood Institute and by U. S. Environmental Protection Agency Cooperative Agreement CR817643 to Philip A. Bromberg. This is an abstract of a proposed presentation and does not necessarily reflect EPA policy. Using conventional light microscopy, silicone has been described as a translucent, clear, mucoid, refractile material which is often difficult to visualize. It is not birefringent, as is silica, by polarized light microscopy. Silicone gel tends to form homogenous, rounded "globules," "vacuoles," or "droplets" unlike the angulated, sharp spicules observed with silica. In paraffin-embedded tissue sections, silicone gel often appears to be slightly out of the plane of focus. In addition, silicone gel "globules" occasionally drop out of standard 4 µm histologic sections during tissue processing, leaving partially or totally vacant holes or "cysts." Because silicone extravasation and deposition into surrounding fibrous breast capsules is difficult to visualize by standard light microscopy techniques, 1 the electron microscopist must often "blindly" examine multiple fields/grids in a labor-intensive fashion. To decrease the time generally required for the positive identification of silicon by electron probe microanalysis (EPMA), alternative light microscopy techniques for preliminary screening purposes were investigated.

Six periprosthetic capsulectomy specimens, 2 synovium from a previously reported silicone breast augmentation patient with arthritic pain 3 and a silicone granuloma from another patient with a ruptured prosthesis were utilized in this study. Sections were cut at 4, 10, 20, and 30 µm and mounted on glass slides. In addition to standard Permount mounting media, Aqua-Mount was stained with ink preparations: Black Stamp Pad Ink, India Ink and Aniline Artist Dyes-black, blue, brown in a 1:1 mixture for coverslip application. Each series was stained with a battery of 33 common histochemical stains. The sections were then viewed with a Zeiss Axioplan microscope utilizing conventional incidental light, polarized, non-Koehler, phase contrast, and darkfield microscopy. A commercial silicone gel and silicone gel extracted from a previously implanted silicone breast prosthesis were smeared and examined unstained or stained with Papanicolaou and "Diff Quik." Silicone was noted to be refractile, nonpolarizable, and nonstainable.

The results confirm the refractile, nonpolarizable, and nonstainable properties of silicone in histologic and cytologic preparations (Fig. 1 , scale bar = 4 µm). Decolorizing techniques (Toluidine blue O, etc.) that were not completely differentiated, especially with thicker sections, occasionally demonstrated nonspecific dye "trapping" on the larger silicone globules. The relative ease of silicone localization was greatly increased in histologic specimens with non-Koehler, phase contrast, and darkfield microscopy when compared with conventional light microscopy. Although standard H&E staining was adequate for silicone localization in tissue sections, uniform dark staining with Toluidine blue O increased the contrast between the stained tissue and the unstained refractile silicone. The contrast was also enhanced by Aqua Mount Black Stamp Pad Ink mounting media, with the silicone appearing milky white. In thicker sections, negative staining was slightly accentuated because of the increased concentration of stain as well as thicker, more refractile silicone globules, especially with darkfield microscopy. By utilizing these sensitive screening techniques, we were then able to sample paraffin-embedded tissue selectively, which dramatically decreased the time and effort for correlating confirmatory identification of silicon by EPMA (Fig.2 ) IV-86 Scanning Vol. 16, Supplement IV (1994) The majority of biological samples are high in water content. Preparing tissues for scanning electron microscopy (SEM) requires removal of this water and often produces severe structural distortions due to surface tension forces during phase transitions. This can result in bulk shrinkage artifact, surface cracking, curled cell borders, clumping or flattening of cilia, and collapse of surface vesicles. Failure to postfix biological material with OsO 4 can lead to extraction of surface membrane lipids during solvent dehydration. It is likely that many shrinkage artifacts are related to incomplete or improper fixation and drying.

Critical-point drying (CPD) has become the standard procedure for most biological materials. Although it produces a relatively intact end product, Wollweber et al. have reported shrinkage of as much as 45% after CPD of glutaraldehyde and osmium-fixed macrophages and lymphocytes. The use of mordant techniques to complement or enhance OsO 4 and uranyl binding may aid in preserving fine structural details regardless of the drying method. Numerous substitution/transition fluids have been introduced for both critical-point and direct-evaporative drying, some with more success than others. An advantage in using solvent drying techniques as alternatives to CPD is the ability to process large numbers of samples simultaneously.

Early attempts at direct drying of nonosmicated samples from ethanol and other solvents produced a generalized cell shrinkage and collapse. Freon 113 evaporative drying was shown to be a useful rapid drying technique by Liepins and de Harven. Gamliel further refined the Freon 113 direct-drying technique to include the use of guanidine HCl as a bifunctional mordant, prior to osmification, to minimize shrinkage of leukocytes. Direct drying from hexamethyldisilazane(HMDS) was introduced by Nation to dry insect tissue and has subsequently been used for drying various other tissues (Adams et al. 1987) . Peldri II has since been shown to be an effective solvent drying medium. It is a solid at room temperature. A comparative study by Bray et al. of Peldri II, HMDS, and CPD, using both plant and animal tissues, has produced identical results with animal tissues but not plant tissues.

The purpose of this study is to introduce acetonitrile as a potential solvent for direct evaporative drying. Acetonitrile has been used as a less toxic propylene oxide replacement transition fluid for transmission electron microscopy (TEM) dehydration and infiltration (Edwards et al. 1992) . To date, no studies have been published indicating the of use of acetonitrile as both a dehydration and intermediate transition fluid for direct solvent-drying of tissue-cultured cells for SEM. This study compares solvent drying of KB31(Hela) cells from Freon 113, Peldri II, HMDS, acetonitrile, and ethanol to critical pointdried cells grown on Thermanox coverslips. The mordant (GTGO) technique of Gamliel is included to illustrate its usefulness in shrinkage control. Solvent-plasticware compatibility should always be tested before preparing cultured cells for SEM; whenever possible, the use of glass is best.

Cells grown on Thermanox coverslips were fixed in 2.5% glutaraldehyde, 0.1 M cacodylate, 0.1 M sucrose, 1.0 mM CaCl2, Ph 7.2. The GTGO fixation protocol of Gamliel was found to be the most useful one in preserving cell architecture (Figs. 1, 2) . In Figure 1 , cells were evaporative-dried under a mild aspirator vacuum after GTGO fixation and dehydration in acetonitrile (25%, 50%, 75%, 100%-3×, 5 min each). As a comparison, cells in Figure 2 were identically fixed and criticalpoint dried after ethanol dehydration. The overall appearance of the cells from both treatments is similar at low magnification (Figs. 1a, 2a) , with some shrinkage evident in either case. Under these conditions, it appears that the CPD cells retain more fine structural details than those evaporative dried from AN (Figs. 1b, 2b) . The results of this study are encouraging and may lead to future studies to develop a simple fixation protocol which enables evaporative drying from solvents directly miscible with water such as acetonitrile. Snow, which may occasionally cover up to 23% of the earth's land, supplies about one-third of the water that is used for irrigation and the growth of crops (Gray 1981) . For this reason, estimating the amount of water in winter snow pack is an extremely important forecast activity that attempts to predict the amount of water that may be available for the following growing season. Unfortunately these estimates can be easily IV-88 Scanning Vol. 16, Supplement IV (1994) FIG . confounded by the sizes and shapes of the snow crystals that comprise the snowpack. A snow crystal is a single frozen ice grain that generally results from a process known as nucleation in which atmospheric water vapor condenses on a solid particle or nucleus at temperatures below 0°C. When nucleation occurs, the water molecules form a hexagonal crystal lattice resulting from the specific orientation and binding that occurs between the oxygen and hydrogen atoms. Depending on the temperature and moisture that prevails during formation and descent of snow crystals, nucleation may result in plates, stellar crystals, columns, needles, or dendrites-all of which are based on the hexagonal lattice structure. An individual snow crystal may range in size from 50 µm to 5 mm (Gray 1981 ); aggregations of two or more of these crystals form a snowflake.

The shapes of snow crystals have been extensively studied and photographed with the light microscope (Bentley and Humphreys 1931, Nakaya 1954) . Although these studies have resulted in a classification system that currently recognizes 9 distinct classes and over 30 subclasses of snow crystals (Hobbs 1974), detailed examinations have been hampered by the difficulty of working with a frozen specimen, which is susceptible to sublimation and melting, and by the limiting resolution of the light microscope. For these reasons, an attempt was made to determine whether snow crystals could be collected/stored and prepared for observation and recording in the low-temperature SEM.

Attempts to allow snowflakes merely to settle on a precooled specimen holder were unsuccessful; the snowflakes tended to "bounce" off the holder and those that did alight did not remain attached during subsequent handling. A successful procedure consisted of placing a thin layer of methyl cellulose solution on a holder and precooling it to the prevailing outside temperature during a snow fall. Next, snowflakes were allowed to settle on the surface of the methyl cellulose solution. After a few minutes, the holder was plunged into liquid nitrogen and transferred to the laboratory where it was retrieved from liquid nitrogen, mounted on the transfer rod of an Oxford CT-1500 HR Cryosystem, moved into the prechamber for sputter coating with Au/Pd, and then inserted into a Hitachi S-4100 field emission scanning electron microscope (SEM) equipped with a cold stage that was maintained at −185°C.

These procedures allowed us to observe several forms of the individual snow crystals as well as their nucleation centers. At low magnification, the specimens, which did not appear to be altered by the sputter coating, resembled those that had been previously photographed with the light microscope. The snow crystals were stable in the beam, did not sublime, and could be observed at magnifications of 20,000× or more to reveal microcrystalline water deposits or rime on the surface of some the snow crystals. This procedure, which was used to collect specimens during several snow falls in Beltsville, Maryland during the 1993-1994 winter season, was also capable of preserving sleet, graupel, and hail. Furthermore, storage holders were devised that allowed capture of the snowflakes and their storage in liquid nitrogen until the specimens could be processed for examination in the SEM. Finally, the specimen stage of the SEM allowed specimen tilt so that stereo images of the snow crystals could be recorded (Fig. 1) .

In conclusion, low-temperature SEM is a viable technique for examining snow crystals at magnifications that far exceed the resolution of the light microscope. Furthermore, the ability to collect and store samples enables investigators to accumulate samples from numerous locations or different time intervals so that detailed observations and comparisons can be done in a convenient and orderly manner. These were identified and returned into the corresponding reconstruction sections in an automatic tracing procedure, programmed and executed on VIDAS 2.5 (ZEISS/KONTRON Germany). In addition it is possible to obtain a plastic, 3-D-like impression by applying the RCM with oblique illumination. This is achieved by decentralization of Stach's slide (central diaphragm). 2. Two main problems in 3-D reconstruction of histologic specimens are the horizontal distortion during the preparation of serial thin-tissue slides and the following vertical readjustment (alignment). We have recently shown that optical sections can be obtained by RCM within thick tissue-slides. 3 Its confocal-like principle provides an alternative to mechanical slices. The preserved integrity of the examined object allows the precise movement within the optical axis in one (vertical) dimension, thus avoiding manual alignment. Applying the RCM on histochemical and immunohistochemical stains, reflections can be observed within the tissue. 4 The depth of penetration of the light beam amounts to approx. 30 µm, equivalent to about 30 optical tomolevels (Fig. 2) .

Using the example of neurons of the supraoptic nucleus of the rat, we demonstrate this new technique on chrome-alum haematoxylin stained neurosecretion. The reflections of the dye particles associated with neurosecretory granules allow the precise localization of these subcellular structures. The visualization of the neurosecretion and its distribution is more distinct and of sharper contrast than in bright-field microscopy. Reflected light is like a binary signal and therefore generates a suitable prerequisite for automatic discrimination in greyscale image analysis. 5 Thus identified black and white negatives were reconstructed with the module REC3D on VIDAS 2.5.

This paper introduces two extended applications of the RCM for 3-D reconstruction. The generated optical slices and isohypses allow qualitative and quantitative investigations of intracellular structures and surfaces. Quantitative chemotaxis is of great interest in a broad field of cell research (e.g., receptor-ligand interaction, ionic channels, cytoskeletal and metabolic processes, embryogenesis) as well as in clinical studies (e.g., immune reaction, wound healing, infection, tumor invasion).

The most accepted assay to measure chemotactic behaviour of cells is the Boyden filter assay, in which cells move against a gradient of chemical or biological substances. The chemotactic response is analyzed by measuring either the distance travelled by the leading front of cells or by quantifying the number of cells on the lower surface of the filter. This method is laborious and limited by the observer's subjective errors.

We have developed a computer-based image analysis system to estimate the three-dimensional distribution of the migrated cells inside the filter. Using one-micron optical sectioning, we can determine the position, size, and shape of many individual cells. Within 10 min, the position of thousands of cells can be recorded and the migration profile in the filter can be determined. It is possible to distinguish between different cell populations by selecting particular cell parameters.

The system consists of a Nikon LABOPHOT-2A microscope with video microscopy and programmable focus control, used in connection with a HaSoTec-image processor with fast data acquisition and user-friendly software (MSwindows). The high throughput, the consistent accuracy, and the simple operation of the system optimize all aspects of routine chemotaxis analysis in basic research and clinical studies. The Atomic Force Microscope (AFM) is being used increasingly in the life-sciences field. With this increase in usage, a concomitant increase in the need for both better developed specimen preparation techniques and better defined operational parameters for the AFM instrument has occurred. Lifesciences AFM methodology can be divided into three main areas: (1) Choosing the appropriate support substrate for the specimen, (2) choosing the most appropriate immobilization techniques for attaching the specimen to the support substrate, and (3) selection of the optimum instrument scan parameters to ensure reliable transfer of data from the specimen to the image.

Of central importance to life-science AFM is the nature of the substrate to be used. The most reliable substrates are those with a well-documented surface, whose features are at least an order of magnitude smaller than the specimens of interest. Furthermore, the material should be transparent so that other forms of microscopy can be used (i.e., light, fluorescence, near-field scanning optical, etc.) and the surface chemistry should be both well documented and susceptible to chemical derivitization. The two most common substrates to fit the above descriptions are glass and freshly cleaved mica. Both of these surfaces can easily be manipulated to produce a wealth of reactive primary amines, either by coating the substrate with poly-L-lysine or by treatment with 3-aminopropyltriethoxy silane (APTES). An analysis of the modified and unmodified substrates show that treatment with either APTES or poly-L-lysine resulted in an increase in surface roughness (Ra). Based on the roughness information, we recommend modified mica as being suitable for biological material ranging from whole cells to DNA, whereas modified glass is unsuitable for samples with heights < 2.5 nm. The red blood cell (RBC) cytoskeleton was examined by atomic force microscopy (AFM). Samples were placed on either glass or mica and imaged in air. No fixative or stain was used; this allowed modification of the samples between images so that molecular components could be identified. The meshlike structure, which is observed when the RBCs are lysed, is identified as a complex of the cytoskeletal integral proteinsspectrin, actin, and band 4.1. The identification was accomplished by imaging the intact cytoskeleton, then treating the sample to selectively remove these proteins and re-imaging the sample. To support the identification of the cytoskeletal proteins further, images were obtained of samples which had been treated with detergent to remove the lipid membrane and leave the cytoskeleton behind. This is the first study of the intact RBC cytoskeleton which identifies specific proteins. More generally, it shows that AFM is a useful tool for examining biological systems in their native state since sample preparation is simple and, once attached to the substrate, the sample can be treated in a variety of ways. We have reported some features about the structural and morphologic changes in nylon 3 transformed from β-alanine single crystal and nylon 6 polymerized from ε-amino-ncaproic acid one, which are members of the ω-amino acid family, through the solid-state polycondensation procedure. It was also found that p-aminohippuric acid [N-(4-aminobenzoyl)glycine] single crystal could be converted into an aromatic polyamide crystal by heat treatment below its melting point. In this report, crystalline and morphologic structures of polyglycine produced from glycine (the simplest amino acid) single crystal were examined mainly by means of scanning electron microscopy (SEM) and x-ray diffraction technique.

The monomer solution was prepared by dissolving 200 g of commercial glycine powder into 500 ml of distilled water at 80°C. The monomer single crystal was precipitated at 30°C from the solution, which was transparent and prism-shaped. Definitive cleavage planes were observed in the monomer crystal, which is found to be parallel to the a-c plane in the αform crystal of glycine. It has been reported that such cleavage is caused by the characteristic structure of α-form glycine crystal. The monomer single crystal was used as the original specimen for the polycondensation reaction, where the original specimen was annealed in decaline at 130 and 190°C up to 1000 h. Morphology of the polymerized material was observed by using a Hitachi S-410 SEM with accelerating voltage of 5 kV after being coated with gold. X-ray photographs were taken by a flat camera mounted on an x-ray generator with Ni-filtered Cu-Kα radiation, where the crystal took three orientational positions so that a-c (cleavage plane), a-b, and b-c planes made a right angle with the incident beam, respectively. Figure 1 shows two types of x-ray diffraction patterns from a specimen polymerized at 130°C for 800 h with the a-c plane perpendicular (Fig. la) and parallel (Fig. lb) to the incident beam, where patterns reflected from the polymerized material were observed as well to have spots from the original monomer single crystal. In Figure la , two strong arcs are observed, which are indexed as 100 (inner reflection) and 001 (outer reflection) from type-I modification of polyglycine crystal. A more complicated diffraction pattern is shown in Figure lb , which seems to be a fiber diagram. Almost all molecular chains in polyglycine-I crystal are considered to be normal to the a-c plane and parallel to the hydrogen bond in the original monomer single crystal. Figure 2a shows a SEM photograph for the surface of the original monomer crystal. Lamination layers of lamellae are observed on the cleavage plane, edges of which are parallel to the b-axis of the monomer crystal. Voids were observed in the grain boundary or crack region. A SEM picture of the specimen polymerized at 130°C for 300 h is shown in Figure 2b , where the laminal materials are seen to overlap each other crosswise. Such intersecting laminal structure seems to be responsible for the biaxial crystalline orientation observed in the x-ray diffraction pattern shown in Figure la. In the case of the specimen polymerized at higher temperatures, the fibrillar structure was observed over the surface. Figure 2c shows the results for the specimen annealed at 190°C for 10 h.

HIROSHI TOYODA, TAKASHI ITOH, HIROSHI SAKABE, TAKASHI KONISHI Department of Polymer Science, Kyoto Institute of Technology, Kyoto, Japan It is well known that polytetrafluoroethylene (PTFE) is produced by emulsification polymerization in the form of powder or aqueous dispersion to be processed industrially. The dispersion-type material is mainly composed of fibrillar and/or lamellar crystallites, where the spherulite structure generally is unseen because of stiffness of the molecular chain. Such morphologic feature is considered to be responsible for the remarkable repellent property of the polymer to liquid. The authors prepared thin films of PTFE from the aqueous dispersion to observe change in the fibrillar structure through heat treatment, considering that from such a viewpoint the fibrillar morphology in PTFE is the aspect that is most different from other polymeric materials.

The polymerized PTFE used in this study was produced by Dupont Ltd. A glass plate was dipped into the solution and then withdrawn vertically. The thin film was prepared by drying the solution on the plate at room temperature. The heat treatment was performed at 300, 350, 400, 450, and 500°C for 1 h in an air-drying oven. The samples were immediately quenched in ice water (cooling rate: 110,000°C/min) or cooled in air at controlled rates (90, 10, l, 0.5, and 0.1°C/min). Any serious oxidation and/or degradation effects were not observed even during annealing at 500°C. A scanning electron microscope (SEM)(JEOL, JSM-5200LV) was mainly used to observe the Au-coated surface structure of the sample, when an accelerating voltage was set at 25 kV to make the resolution of the SEM high.

Composite materials were produced by coating PTFE on textile of glass-fiber and that of carbon fiber. After the coating and subsequent annealing at 100°C for 3 min, the material was cooled to room temperature. The surface was found to be covered with balls of a diameter of about 300 nm, composed of several hundred nodules of a diameter of about 50 nm as shown in Figure 1 . In this sample, the degree of crystallinity of PTFE was about 30% and no fibrillar structure was observed. By increasing the annealing temperature to > 300°C, the fiber bun- IV-94 Scanning Vol. 16, Supplement IV (1994) dles grew up. Such phenomena may be explained by stiffness of the molecular chain including F instead of H, that is, the chain does not fold even if the molecular motion is stimulated at elevated temperatures. Any sheaf-like or microspherulitic structure (a precursor of the spherulite), which is often observed in thermoplastic polymers such as polyethylene, did not appear in PTFE. When decreasing the cooling rate, the fibrils grow laterally and interfibrillar space tends to become larger. It is unknown how such space affects the physical properties of the material. Thinner fibrils appeared, which connected the original thick ones, in the case of 0.1°C/min cooling as shown in Figure 2 . Such morphologic structure characteristic of PTFE seems to be responsible for the processing efficiency.

Schott Glaswerke, Department of Instrumental Analysis and Mineralogy, Mainz, Germany

Fused-cast refractories of the ZAC-type (zirconia-alumina cast) are of great importance in the manufacturing of glass. Especially glass-tanks for the melting of specialty glasses are built with these materials. ZAC refractories are produced by melting the raw materials at very high temperatures and subsequently casting them into suitable molds. On cooling, part of the material crystallizes, forming Al2O3 (corundum) and ZrO2 (zirconia, Baddeleyite), where the zirconia crystals are also growing within or into the corundum crystals. Besides the crystalline phases there also exists a glassy phase, composed mainly of SiO2, Al2O3, and ZrO2, and small amounts of alkali (Na2O, K2O), TiO2, and Fe2O3. This microstructure of the material must have an influence on the corrosion behavior when subjected to the aggressive melt in a glass tank.

The aim of this work was to establish a method to characterize quantitatively the microstructure of these refractories.

The inhomogeneous nature of the material can already be discerned in the light microscope; however, the resolution is not sufficient to measure the sometimes very small (< 1 µm) crystal sizes of ZrO2, not to mention the determination of a form factor! Therefore, the electron microprobe (EPMA) was used to perform this work.

The instrumentation used was a JEOL JSM-840 scanning electron microscope, equipped with an optical microscope for reflected and transmitted light, motorized stage (x, y, z), and coupled with a combined EDS/image analyzing system VOY-AGER 2110 from NORAN, which also controls the stage movement.

The following parameters of the microstructure, for example, the geometric and chemical properties of the crystalline and glassy phases, were to be determined: maximum and minimum diameters, area of the individual phases and fractional area within the material, form factors and orientation to a reference plane, and the chemical composition of the glass phase.

Furthermore, it was necessary to distinguish between the zirconia within the corundum and the more isolated crystals. Since the material is inhomogeneous, a rather large number of image fields had to be analyzed (at magnifications of 500× and higher the individual image field is very small!); as a consequence, the whole procedure had to be automated as much as possible.

Sample preparation: a flat and polished surface of a section of the material had to be produced. Although the grinding and polishing was done with diamond wheels and paste, a certain amount of relief between the hard corundum and the "soft" glass is unavoidable.

The first step in analyzing the microstructure is to recognize/discriminate the three phases in the material. The signal used for image formation and subsequent phase discrimination is the backscattered electron (BSE) image; however, the abovementioned relief due to differences in hardness made the distinction between glass and corundum impossible, since (a) brightness differences between glass and corundum are not very strong in the first place, and (b) edge effects in the corundum crystals showed brightness values of the glass, and vice versa (discrimination, i.e., tranformation of a particular phase into a binary image, is based on such brightness differences). Contrast enhancement or image filters were not sufficient for clear separation of these two phases. Therefore, for the recognition of the glass phase, an element distribution for Si (x-ray map) was used; combined with the image of ZrO2, which is easily discriminated from other phases because of its high average atomic number/high backscatter signal, it yields the inverse of the corundum image. This sounds quite simple; however, since the resolution of the x-ray signal is considerably less than the BSE signal, quite some effort had to be put into the treatment of the Si-distribution to give correct areas and outlines of the glass phase.

Another rather complicated step was the recognition of the zirconia crystals grown within or into the corundum. Crystals fully enclosed by the corundum presented no problem, but those growing from the edge into the Al2O3 were difficult to discriminate as to belonging to this particular structure. The problem here is to tell the machine what the eye and judgement of the operator consider to be part of this feature. An additional treatment of the glass phase (binary image filters) was necessary to solve this task. The binary image of the glass phase is used as a template for the determination of the chemical data of the glass; it provides the control for stepping the electron beam only over the desired areas while acquiring data with the energy dispersive spectrometer (EDS).

All image analysis procedures, including the storage of images and EDS spectra together with the evaluation set-ups (selection of properties to be determined), were combined within a schedule. The stage control and automation software then connects the stage movement to predetermined points (or a number of points along a line) with the execution of the schedule at each of these points on the sample.

After completion of the analysis run, which needs about 10 hours for 40 image fields and is therefore done unattended overnight, the large amount of stored data are statistically evaluated (for the zirconia particles alone, there are easily more than 20,000 data sets, consisting of seven measured or derived properties each!). This task is performed using the Lotus 1-2-3 calculation program. The chemical data are extracted from the stored spectra with the ZAF-correction procedure to yield wt-% oxide data, and also statistically evaluated with the Lotus program.

Although the aim of this work was only to set up an analysis procedure, the test runs on various samples have already shown differences in the microstructure as determined by the geometric and chemical properties of the individual phases. The application of this procedure to "real" samples will then establish correlations between the microstructure and the properties of the material with respect to their use in glass melting. Approximately 53,000 tons of electric furnace flue dust accumulated in an industrial area in Tifton, Georgia. Vehicles transporting the flue dust, classified as K061 hazardous waste, initially dumped the material in a warehouse. Once the warehouse was full, the flue dust was dumped in an uncovered pile. Run-off from the pile and wind-driven particles contaminated nearby industries, residential buildings, and soils over a period of many years.

Scanning electron microscopy-energy dispersive x-ray spectrometry (SEM-EDS) was used to compare the morphology and chemical composition of fly ash dust from the suspect pile ( Fig. 1 ) with samples collected from the surrounding buildings and soil. Post-it Notes (Millette et al. 1991) , modified with a strip of conductive carbon tape, were used to collect dust that had accumulated in buildings surrounding the fly ash dump site. Suspect dust particles were analyzed by SEM-EDS to compare with known dust particles from the fly ash pile. Soil samples were sieved, with "fines" from the dry soil analyzed by SEM-EDS and compared with samples from the fly ash pile. Particles similar in chemical composition and morphology were identified in most of the buildings sampled that surround the fly ash dump site. Soil samples from areas surrounding the dump site were also found to contain fly ash IV-96 Scanning Vol. 16, Supplement IV (1994) FIG. 1 Backscattered electron image of fly ash spheres from the flue dust pile. Scale bar = 10 µm. particles similar in morphology and chemical composition to fly ash from the suspect pile.

In conclusion, soil and dust samples taken from homes and outdoor areas surrounding the fly ash pile were found to contain particles similar in morphology and chemical composition to particles from the fly ash pile.

Scanning electron microscopes (SEMs) are widely used in detection and quantification of material microstructure and imperfections. Results obtain with SEMs need a high expertise to be fully exploited. Specialized programs such as a single-scattering Monte Carlo simulation can effectively predict the electron beam interaction with solids and thus help the quantification. One of the most powerful advantages of Monte Carlo simulation to help microscopists is to generate images. With high-speed computers we can now simulate images in a reasonable amount of time. In this paper we present images of spherical inclusion of MnS in a Fe matrix. It is shown that there exists a difference between the image dimension and the real dimension. Also, it is shown that geometric effects can alter the resulting image.

The Monte Carlo program used for low-energy simulation is described elsewhere . Figures 1-4 show simulated image of MnS inclusions (the dark center of the figures) in a Fe matrix (lighter part). To build this image we need to simulate 50,000 primary electron trajectories and then calculate the associate backscattering coefficient for each pixel of the screen. The image will then need 10,000 simulations for a total time of approximately 200,000 CPU min on a RISC 6000 workstation. Because of the symmetry of the image we can use an better method. Starting from the center of the inclusion we simulate 50 points moving on a radius to the border. Then we rotate this line on 360°and add random noise for the Monte Carlo simulation to obtain the full image. This image is coded in windows BITMAP format and can then be converted to TIFF or another format.

For our simulation, we used a diameter of 10 nm for the electron beam, at which value the incident beam current would be 1×10 −10 amps in our JEOL 840 SEM. Figure 1 shows a spherical inclusion with a radius and at a depth of 500 nm simulated with a beam energy of 10 keV. Figure 2 shows the same inclusion but simulation at 20 keV. Figures 3 and 4 show the inclusion with a ratio depth/radius of 1.4, simulated at 10 and 20 keV, respectively. In Figure 5 we can observe the discrepancy between the image dimension and the real dimension as a function of the ratio depth/radius. As was expected, the error increased with the depth/radius ratio. It is also interesting to note that increasing the energy of the incident electrons increases the image error.

The geometry of the solids affect the backscattering coefficient. We can obseve a white border in Figures 1-4 A similar problem occurs in the middle of the inclusion which appears lighter. It is interesting to see the same phenomena in Figure 6 for an image of a real inclusion.

The photograph of the embedded particulates in a matrix taken by SEM in backscattering mode should be analysed carefully because this method usually overestimates the dimension of such particulates.

The measurement of strain, using lattice parameter changes, can be determined from channeling patterns in the scanning electron microscope (SEM) using HOLZ lines. The method used by Kuzubowski, 1 for example, in the [001] orientation of silicon, utilizes the change of the height of the triangle from the intersections of the (660) and two {571} HOLZ lines. More recently we have been investigating the channeling patterns for silicon to calibrate accurately the voltage in an SEM and we have utilized a pin wheel pattern formed from the {5 11 1} HOLZ lines in [001] pattern at 26.86 keV. At this voltage, the (5 -1 -1 -1) and (5 11 1) are very close to each other, within 0.005 degrees, and these two lines split as the voltage increases or decreases. The width of the splitting is quite sensitive to any voltage changes as are the {991} intersections with these lines. An experimental electron channeling pattern (ECP) of these lines, along with a simulation are shown in Figure 1 . The voltage calibrated from the simulation is 27.28 keV as shown in Figure 1c . These same HOLZ lines are, of course, also appropriate for strain measurement since the variation in the voltage is equivalent to an isotropic strain field that would uniformly change the lattice spacings. The shifting of the HOLZ lines in the electron channeling patterns due to strain are from both the change in the magnitude of Bragg angle and the change in the orientation of the diffracting planes. The former corresponds to a change in the distance of reciprocal lattice point from the origin(i.e., d spacing) and the latter is due to the rotation of reciprocal lattice point about the origin. Thus the total angular change can be written as where θ B is the Bragg angle, ∆θ P is the rotation of the plane, g is the diffraction vector, λ is the electron wavelength, and | | represents the length of a vector. E is a matrix equal to ε + I where ε is the strain tensor and I a unit matrix. The approximation ~ is due to the small angle approximation for sin(∆θ B )~∆θ B and 1 − ∆θ B 2~ 1. The HOLZ lines toward the center of the channeling patterns are more sensitive to lattice parameter variations because they have a larger value of g and the change of the Bragg angle is in proportion to this magnitude. The angular IV-98 Scanning Vol. 16, Supplement IV (1994) ∆θ = ∆θ B + ∆θ P~E g − g widths of these channeling lines are also smaller and so they are well defined. 2 When the strain is not isotropic, the sensitivity is also determined by the orientation relationship between the strain ellipsoid and the g vector. Thus the appropriate choice of the g vector can be used to maximize the second term in the equation. In Figure 2a , for example, we show the simulation of a channeling pattern for a strain along only one direction (e.g., [100] ) which breaks the four-fold symmetry of the [001] pattern. It can be seen that the strain effects on the {991} are all the same since all the planes in this family are symmetrical to the strain direction. However, the sensitivities to the strain in the {5 11 1} are not all equal, (11 5 1) being more sensitive than (5 11 1). The comparison of the sensitivity can be seen from the simulation and from Figure 2b , in which the angles of splitting are plotted as a function of the strain. For years LaB 6 has been the industry standard for thermionic emission cathode material. In 1991, FEI introduced CeB 6 as an improved alternative to LaB 6 . CeB 6 directly replaces LaB 6 and has certain distinct benefits. CeB 6 has a lower volatility than LaB 6 , increasing the lifetime of the cathode. This reduces the frequency of cathode purchases and replacements. In addition, greater beam stability and faster startup are achievable from CeB 6 's greater resistance to contamination. However, to gain the benefits of CeB 6 , it must be operated correctly. Studies have shown that the operating characteristics of CeB 6 , such as total emission current, are different from LaB 6 . Proper operation of CeB 6 comes from understanding the operating characteristics of CeB 6 and how they interact with the system in which it is operating. To compare the operating characteristics of CeB 6 and LaB 6 , we performed parametric studies in a JEOL 6400 scanning electron microscope. The JEOL 6400 is a self-biased system where the bias voltage (Vb) is dependent upon the total emission current as Vb = Ie*Rb, where Rb is the bias resistor and Ie is the total emission current.

The results show that CeB 6 operates at a lower total emission current than that of LaB 6 (Fig. 1) . Therefore, for the same bias resistor settings, the bias voltage on CeB 6 is less than that on LaB 6 (Fig. 2) . Please note that the lower emission current of CeB 6 does not imply a lower probe current. CeB 6 has high transmission and has been shown to provide probe currents similar to LaB 6 .

To optimize the operation of CeB 6 in a self-biased system, the operating conditions must be optimized for a low emission current source. These operating conditions include the bias resistor value and Wehnelt-to-tip spacing. We have found that increasing the bias resistor improves the performance and lifetime of CeB 6 . Other experiments show that adjustments to the Wehnelt-to-crystal tip distance further improve the performance of CeB 6 . The effects of adjusting the Wehnelt-to-tip distance vary for different electron microscope systems.

For independently biased systems, we recommend setting the filament temperature to 1800K (the corresponding filament current is provided with the cathode) and subsequently adjusting the bias until the desired emission image is obtained. This is possible because the emission image of CeB 6 is the same as the LaB 6 emission image. The Wehnelt-to-tip distance should be set to the distance recommended for LaB 6 .

In transmission electron microscopes (TEMs), the operating point is set by viewing the source image. Because the emission image of CeB 6 is similar to LaB 6 , the procedures for obtaining the LaB 6 operating point in TEMs still apply to CeB 6 . However, the total emission current at the operating point will be lower than with LaB 6 .

By following these guidelines, the operation of CeB 6 can be optimized and with this optimized operation, the benefits from additional lifetime and increased stability over LaB 6 can be achieved.

TAKESHI HATSUZAWA, YOSHIHISA TANIMURA, KOUJI TOYODA, MAKOTO NARA*, SYUUJI TOYONAGA*, SHIN-YA HARA*, HIROTAKA IWASAKI*, KAZUHIKO KONDOU* National Research Laboratory of Metrology, MITI Tsukuba; *Nikon Corporation, Tokyo, Japan A compact laser interferometer with a piezo-driven scanner has been developed for metrological micro-linewidth measurement in regular scanning electron microscopes (SEMs). So far, special SEMs combined with various scanners and interferometers (Postek 1989 , Hatsuzawa et al. 1990 ) are necessary to perform absolute and precise measurements; however, this device solved the problem by miniaturizing a one-dimensional mechanical scanner and a multi-optical path interferometer.

The arrangements of optical components are illustrated in Figure 1 . A mechanical scanner is constructed in the center of a 130 mm diameter base disk. The square part, slit by using a electric discharging machine, is suspended by thin elastic suspensions at each corner. The table is driven by a piezo-electric actuator of a traveling length of 10 micrometers. A He-Ne laser beam is introduced on the disk by a single mode fiber through a collimator lens, and it is split and deflected by beam splitters and lenses so that two beams are facing each other and form a differential interferometer. At both ends of the table, right-angle prisms are facing each other so that the laser beam goes back and forth five times. By using the differential arrangement and optical-path multiplication technique, the resolution of the interferometer is improved ten times from the original Michelson's arrangement. The reflected beams are surperimposed by a half-mirror to generate an interferometric signal, and it is detected by four photo diodes (PD0-PD270) after changing its phase by 90˚ through a half-and a quarter-wave plate arrangement. This detection method and an electrically operational processing enhance the resolution 40 IV-100 Scanning Vol. 16, Supplement IV (1994) times from its original. Eventually, by using physical and electrical methods, the resolution of the compact interferometer is 0.8 nm (âλ/800).

To evaluate the performance of the compact interferometer, it was installed in the vacuum chamber of a regular SEM (JOEL JSM-840A) as shown in Figure 2 . The fiber and electric wires are introduced through two flanges next to the chamber. A software servosystem is constructed by using the interferometer and a piezo driver. According to the positional information read through the interferometer counter, a D/A converter commands the piezo driver to change table position precisely, allowing simultaneous sampling of the secondary electron intensity distribution. In the system, the resolution is improved to 0.4 nm (âλ/1600) by using the counter function. Thus, a precise measurement system is realized for the absolute measurement of micro-linewidth in a regular SEM. A maximum drift of the interferometer counter of 5 nm/h was observed by fluctuations in room temperature, however, the influence of the drift can be neglected in actual measurements since a line-scan is finished within a dozen seconds.

A comparison of measurements between the metrological SEM (Hatsuzawa et al. 1990 ) and the compact system was made by using silicon micro-line artifacts, ranging from 0.5 to 0.8 micrometer. The measured linewidths agree within a couple of nm in both measurements, although the measurement conditions are different in acceleration voltage, etc. The results show that the measurement system using the compact interferometer has the same performance as the metrological SEM. This means that absolute and accurate measurements can be obtained everywhere by using the compact laser interferometer and a regular SEM. This device can be applied to various types of scanning probing microscopes as well as to optical microscopes by improving the table mechanism and the arrangement of optics.

Institute of Ecology and Department of Botany, University of Georgia, Athens, Georgia Recent studies of Lake Lanier, Georgia, revealed a fungal epidemic on the planktonic alga, Synedra acus. Clonal isolates of the fungus were identified as Zygorhizidium planktonicum (Chytridiomycetes), an obligate parasite of freshwater diatoms. Although frequently present in lakes and reservoirs of Western Europe, the occurrence of Z. planktonicum in North America has not been previously confirmed. Earlier studies have described the morphology of Z. planktonicum on Asterionella formosa; however, little is known of the fine structure and infection process on S. acus. As a basis for further investigations, morphology and mechanism of infection were characterized by scanning and transmission electron microscopy. These observations provided exceptional accounts of germinating spores and developing thalli. Moreover, conjugation was characterized as the fusion of heterogametangia by means of an extended smooth-walled tube, emanated from the smaller "male" gametangium. Upon attachment, the club-shaped conjugation tube adhered to a small region of the adjoining thallus. This point of contact became continuous with the maturing "female" gametangium which appeared smooth-walled and often highly vacuolate. Ultrastructural examinations also illustrated a shared cytoplasm between conjugating gametangia and apparent migration of organelles; however, fusion of nuclei was not observed. The mechanism of infection on S. acus appeared identical to earlier descriptions of Z. planktonicum on A. formosa (Beakes et al. 1992) . Following encystment, spores typically produced a single germ tube which grew over the frustule valve in an unwavering, linear fashion. Penetration occurred between an overlapping region of the outer and inner frustule. At the point of intrusion, the rhizoid appeared slightly swollen, often further displacing the outer diatom wall.

A program for Monte Carlo simulation of electron energy loss in nanostructures

A Monte Carlo calculation of the backscattering coefficient for a multilayer sample

Monte Carlo program for minicomputers using Mott cross sections

An improved method of measuring biological submicron motion and displacement using laser amplified motion detection and analysis

Friedlander SK: Smoke, Dust and Haze, Fundamentals of Aerosol Behavior

Subcommittee on Airborne Particles: Airborne Particles

Micromanipulators and Micromanipulation

Moor H: Recent progress in the freeze-etching technique

Optimization and application of jet-freezing

Platinum-iridium/carbon: A high-resolution shadowing material for TEM, STM, and SEM of biological macromolecular structures

Freeze-fracturing for conventional and field emission low-temperature scanning electron microscopy: The scanning cryo unit SCU 020

Cryo-preparation and planar magnetron sputtering for low temperature scanning electron microscopy

Imaging of intramembranous particles in frozen hydrated cells (Saccharomyces cerevisiae)

Vapor pressure data for some common gases. R.C.A. Review

An improved cryo-jet freezing method

In vitro spinal cord trauma

Early post trauma changes in rat spinal cord: Electron probe microanalysis

Surface studies by STM

Fabrication technique for tips with controlled geometry for scanning tunnelling microscopy

Scanning probe metrology

Low temperature thermal oxidation sharpening of microcast tips

Microfabrication of AFM tips using focused ion and electron beam techniques

Dimensional metrology with scanning probe microscopes

A rocking beam electrostatic balance for the measurement of small forces

A scanning tunneling microscope with a capacitance-based position monitor

Scanning probe tips formed by focused ion beams

Envelope reconstruction of probe microscope images

Surface recovery in scanning probe microscopy

Probe characterization for scanning probe metrology

Comparison of diffraction techniques for the SEM. Scan Electr Microsc

Electron channelling in the SEM

A review of excimer laser projection lithography

M: Direct electron-beam patterning for nanolithography

Direct STEM fabrication and characterization of selfsupporting carbon structures for nanoelectronics

Die Entstehung einer Vielzahl von Kontaminationsfäden unter der Electronen-Mikrosonde

Transition from chemical etching to chemical polishing studied by the SEM

Chemical preparation of dielectrics for studying their microtopography by the SEM

Microprobe analysis in human pathology

Shelburne JD: Preparation of biological tissue sections for correlative ion, electron and light microscopy

Negative Staining: Applications and Methods

Detection and identification of viruses by electron microscopy

Diagnosis of viral infection by electron microscopy

Electron Microscopy in Diagnostic Virology

Electron Microscopy in Viral Diagnosis

Genetic identification of a hantavirus associated with an outbreak of acute respiratory illness

Isolation of Muerto Canyon virus, causative agent of hantavirus pulmonary syndrome

Distinction between Bunyaviridae genera by surface structure and comparison with Hantaan virus using negative stain electron microscopy

Anatomic complications of abdominal surgery with special reference to the ureter

Urological complications of renal transplantation can be prevented or controlled

The microvasculature of the guinea pig ureter. A scanning electron microscopic investigation

Application of an NaOH maceration method to a scanning electron microscopic observation of Ito cells in the rat liver

SEM blood vessel cast-analysis

Microangioarchitecture of the islets of Langerhans in the snakes, Naja Naja, Vipera russelli and Echis carinatus

Histochemical methods for acid phosphatase using hexazonium pararosanilin as coupler

Acid phosphatase activity in the inner ear

The development of the stria vascularis in the mouse

Kimura RS: Distribution, structure and function of dark cells in the vestibular labyrinth

Secretory epithelial linings in the ampullae of the guinea pig labyrinth

La fosfatasi acida del labirinto membranoso dell'embrione di pollo durante lo sviluppo

The development of human placental villous tree

Scanning electron microscopic observations on the surfaces of chorionic villi of young and mature placentas

Ultrastructure of the epithelium of the chorionic villi of the human placenta

Some new findings about Hofbauer cells in the chorionic villi of the human placenta

The fine structure of human placental villus as revealed by scanning electron microscopy

Monte Carlo Simulation with EBIC

A Monte Carlo calculation backscattering coefficients

Calculations of Mott scattering cross section

Simulation of SEM screen image by a Monte Carlo method

Quantitative x-ray microanalysis of spherical inclusions embedded in a matrix using a SEM and Monte Carlo simulations

A standard procedure for the modeling of the decrease in detection efficiency with time for low-energy EDS spectra

An empirical stopping power relationship for low-energy electrons

Measuring the backscattering coefficient and secondary electron yield inside a scanning electron microscope

Applications of a knock-on process Monte Carlo simulation based on the Mott cross section to quantitative electron microprobe analysis

X-ray production as a function of depth for low electron energies

Theoretical electron-atom elastic scattering cross section

Cross section for K-shell ionization by electron impact

The use of polyacrylamide as an embedding medium for immunohistochemical studies or embryonic tissue

Immunocytochemical studies of cardiac myofibrillogenesis in early chick embryos I: Presence of immunofluorescent titin spots in premyofibril stages

Novel applications of acrylamide for cryosectioning of isolated cells, tissues, and arthropods

Whole-mount analyses of cytoskeletal reorganization and function during oogenesis and early embryogenesis in Xenopus

Confocal microscopy of thick sections from acrylamide gel embedded embryos

Resolution of subcellular detail in thick tissue sections: Immunohistochemical preparation and fluorescence confocal microscopy

The use of polyacrylamide as an embedding medium for immunohistochemical studies of embryonic tissues

Application of acrylamide as an embedding medium in studies of lectin and antibody binding in the vertebrate retina

Developmental angiogenesis: Quail embryonic vasculature

Embryonic vascular development: Immunohistochemical identification of the origin and subsequent morphogenesis of the major vessel primordia

Vasculogenesis and angiogenesis: Two distinct morphogenetic mechanisms establish embryonic vascular pattern

Endothelial cell origin and migration in embryonic heart and cranial blood vessel development

Morphogenetic mechanisms in avian vascular development

Near-field optics: Microscopy, spectroscopy and surface modification beyond the diffraction limit

Mechanical detection of magnetic resonance

Related scanning techniques

Morphology and diameters of crystallites in remineralized enamel

The shape of enamel crystal within human enamel

Densitometric study of polarized light images from carious lesions

Conditions required for detection of specimen specific SE-1 secondary electrons in an analytical SEM

A high resolution SE-1 SEM study of enamel crystal morphology

High resolution topographic imaging of enamel crystal surfaces

Analysis of metal films suitable for high resolution SE-1 microscopy

Antiviral activity of RNAdye combinations

Microspectrophotometry and digestibility of alkali-treated walls in bermudagrass cell types

Simplified highly efficient apparatus for photographic transaxial x-ray tomography

Embryonic vascular development: Immunohistochemical identification of the origin and subsequent morphogenesis of the major vessel primordia in quail embryos

Antibodies to b 1 -integrins cause alterations of aortic vasculogenesis, in vivo

Capillary endothelial cell cultures: Phenotypic modulation by matrix components

In vitro rapid organization of endothelial cells into capillary-like networks is promoted by collagen matrices

Connective tissue morphogenesis by fibroblast traction. I. Tissue culture observations

Reorganization of basement membrane matrices by cellular traction promotes the formation of cellular networks in vitro

Consumption of various process cheese products has been steadily increasing in the U.S.A. and worldwide. These products are manufactured in various styles depending on composition and physical properties

Sodium citrate (SC), trisodium phosphate (TSP), disodium phosphate (DSP), and sodium hexametaphosphate (SHMP) were used at 2.7% levels as melting salts. After cooking, the samples were held at 80°C for up to 5 h. They were removed after 0, 1, 2.5, and 5 h, cooled at 4°C, and stored for 2 days, after which they were analyzed for moisture, fat, protein, pH, firmness, and meltability. For transmission electron microscopy, small samples (1 mm 3 ) were fixed in 3.5% glutaraldehyde and 2% osmium tetroxide, embedded in Spurr's low-viscosity medium, sectioned (90 nm thick sections), stained with uranyl acetate and lead citrate, and examined at 60 kV accelerating voltage

The data in Figure 1 were converted to a pseudo three-dimensional relief map of the embryonic vessels using digital image processing software. It can be argued that this digital rendering, which contains apparent overhead illumination (and shadows below), provides a more comfortable format for assessing visual information. References Caric M, Kaláb M: Processed cheese products

Textural properties and microstructure of process cheese food rework

Microstructure of processed cheese products

Milk gel structure. VI. Cheese texture and microstructure

Effect of draw pH on the development of curd structure during the manufacture of mozzarella cheese

Structure and rheology of string cheese

Mozzarella cheese: Impact of coagulant type on functional properties

Encyclopedia of Food Science & Technology 2

Food texture and microstructure

Electron microscopic observations on the casein micelles of buffalo milk: A preliminary study

Development of microstructure in raw, fried, and fried cooked paneer made from buffalo, cow and mixed milks

The role of casein micelles in changes in the colour of milk

Microstructural evaluation of model starch systems containing different types of oils

Use of the Bird-Leider equation in food rheology

Starch gelatinization in the presence of emulsifiers. A morphological study of wheat starch

Fixation analysis of Tetrahymena pyriformis ultrastructure

Distribution of polyphenol oxidase in organelles of hyphae of the wood-deteriorating fungus, Coriolus versicolor. Biodeterioration Research

A study of the bladder blood flow during distention in rabbits

A vascular network closely linked to the epithelium of the urinary bladder of the rat

The effects of acute overdistention of the rabbit bladder

Applications of an NaOH maceration method to a scanning electron microscopic observation of Ito cells in the rat liver

Light microscopy techniques for the demonstration of silicone

Synovial metaplasia of a periprosthetic breast capsule

Demonstration of silicon in sites of connective tissue disease in patients with silicone-gel breast implants

Biological specimen preparation for SEM by a method other than critical point drying

Comparison of hexamethyldisilazane (HMDS), Peldri II, and critical point drying methods for scanning electron microscopy of biological specimens

Acetonitrile as a substitute for ethanol/propylene oxide in tissue processing for transmission electron microscopy: Comparison of fine structure and lipid solubility in mouse liver, kidney, and intestine

Optimum conditions may allow air drying of soft biological specimens with minimum cell shrinkage and maximum preservation of surface features

A rapid method for cell drying for scanning electron microscopy

Nation JL: A new method using hexamethyldisilazane for preparation of soft insect tissue for scanning electron microscopy

The use of a simple method to avoid cell shrinkage during SEM preparation

Handbook of Snow: Principles, Processes, Management and Use

Hobbs PV: Ice Physics

Snow Crystals: Natural and Artificial

Interference reflection microscopy in cell biology: Methodology and applications

Rekonstruktion des Oberflächenreliefs von Erythrozyten mit Hilfe der Leitz-Reflexionskontrast-Einrichtung. Leitz-Mitt Wiss Tech VII/7

Reflection contrast microscopy within chrome-alum haematoxylin stained thick tissueslides

Reflexionskontrastmikroskopie in der Immunhistochemie

Lichtmikroskopische Untersuchungen zum Einfluss von atrialem natriuretischem Peptid (ANP) am Nucleus supraopticus

Scanning electron microscopy of Post-it ™ Notes used for environmental sampling

Development of a Monte Carlo program for low energy work

Measurement of small elastic strains in silicon using electron channeling patterns

Electron channeling patterns in the scanning electron microscope

A metrological electron microscope system for microfeatures of very large scale integrated circuits

Scanning electron microscope-based metrological electron microscope system and new prototype of scanning electron microscope magnification standards

Comparative ultrastructural ontogeny of zoosporangia of Zygorhizidium affluens and Z. planktonicum, chytrid parasites of the diatom Asterionella formosa

IV-102 Scanning

Heterogametangia attached by a fully developed conjugation tube (arrowhead)

Infection sites of diatom host showing intruding germ tubes (arrowheads) and developing thalli. Scale bar = 5 µm

The author is grateful to Dr. David Howell for critically reviewing this manuscript and to Ms. Lara Muffley for technical assistance.This research is supported in part by NIH grant NDDK R01 DK47563-01. The authors gratefully acknowledge Dr. Steven Armstrong for his assistance in animal preparation.

mic integrity and induced mitochondrial hypertrophy without altering microtubular ultrastructure (Fig. 1a, b) . To localize exogenously administered CDZ, a polyclonal antibody to the drug conjugated to Keyhole Limpet Hemocyanin (Pierce, Rockford, Ill.) was prepared by immunizing New Zealand rabbits and subsequent bleeds. The antibody titer was both detected and quantified by an indirect ELISA assay (Pierce ELISA Starter Kit). The detected antibody was separated from other serum proteins by immunoaffinity chromatography utilizing Pharmacia's Mab Trap G and then tagged with 5 nm immunogold particles. Transmission immunoelectron microscopy employing tagged antibody and glutaraldehyde/ paraformaldehyde-fixed, DMF dehydrated, and Lowicryl-embedded CDZ-treated and nontreated Tetrahymena revealed no immunogold association with either microtubules or any other cytoplasmic organelle. This suggested that intracellular CDZ was cytosolic and leached out during EM processing. Thus, CDZ appears to impair growth and motility through an effect on general metabolism, for example, protein synthesis and/or respiration, rather than a direct action upon microtubules. However, isolation, purification, and characterization of microtubular proteins from Tetrahymena cultured with and without CDZ are required to substantiate this tentative conclusion. Support: NSF-RIMI Grant No. RII-8912781. Coriolus versicolor, a white-rot, wood-decay basidiomycetous fungus, elaborates extracellular ligno-cellulolytic enzymes which possess marked industrial and agricultural applications. Thus, we have been attempting to overproduce and enhance/regulate the secretion of these enzymes employing polyphenol oxidase (PPO) as a model enzyme. It catalyzes the conversion of o-diphenols (tree-generated resistance factors) to o-diquinones and oligomerizes syringic acid, alignin derivative. Previously, we (Moore et al. 1994 ) employed biochemistry and immunoelectron microscopy to map the route of PPO secretion through intracellular endomembrane and possible wall-associated components for hyphae cultured in defined liquid (biochemistry) and solid (microscopy) media. Here, the ultrastructures of C. versicolor hyphae cultured in Kirk and Kelman's defined liquid or solid media are compared. In addition, ultrastructural cytochemistry to define further the intracellular route of PPO secretion to the growth medium in liquid cultured hyphae is described. Hyphae of various culture ages (5-12 days) were prefixed 30 min in 2.5-3.0% glutaraldehyde buffered with 0.05 M cacodylate/cacodylic acid, pH 7.4 and after washing with buffer postfixed for 1h in buffered 2% 0 s 0 4 . For cytochemistry, prefixed hyphae were washed and treated with either cacodylate buffer or buffered 50 mg ml -1 TLC pure, dihydroxyphenylalanine (DOPA) on ice, followed by 1h at 37°C and then 0 s 0 4 postfixed. The hyphae were dehydrated through a graded acetone series and embedded in Spurr's low epoxy resin. The ultrastructures of hyphae cultured in defined medium containing or lacking agar were similar (Fig. 1A , B) except that hyphae grown upon agar possessed a sheath (HS) external to the cell wall. Comparisons of numerous micrographs of aldehyde-fixed hyphae treated with cacodylate CHRISTIAN H. RICKERT AND TIMM J. FILLER Institute of Anatomy, Westfälische Wilhelms-Universität, Münster, Germany Until today, quantification in cytochemistry has mainly been performed on supracellular level and by subjective estimation which makes it difficult to compare results of different investigations, even for standardized cytochemical procedures. Reasons for this are the lack either or of calibrations or of relative reference points. Greyscale image analysis principally allows the quantification of dye-density, but in practice stain intensities depend on many technical circumstances, that is, slide thickness, density of materials, light conditions, or costaining effects, and do not allow automatic identification because of weak grey-contrast. Thus, greyscale image analysis for cytochemical quantification on subcellular level necessitates four prerequisites: (1) applicability within tissue-slides combined with high resolving power, (2) thin optical tomolevels to avoid superimposition of stained structures, (3) clear distinction of structures (specificity), and (4) high contrast.

The reflection contrast microscope (RCM; LEICA Germany) meets the above mentioned requirements. It combines effective suppression of aspecific reflected light with epi-illumination. Because of its confocal-like principle, the RCM can be applied on thick tissue slides to obtain distinct optical sections. 1, 4 At a magnification of 100×, the depth of these sections amounts to approximately 1-2 µm, circumventing an accumulative effect as seen in transillumination. 2 Some cytochemical stains show specific reflections in the RCM, increasing the detection sensitivity to a level of objects 15 nm in size. 1, 3 Reflections are like a binary signal (all-or-nothing principle) and deliver an intense contrast against a dark background, thus facilitating image processing by substituting the common density measurements with field measurement corresponding to areas of reflections.We designed an analysis program on VIDAS 2.5 (ZEISS/KONTRON Germany) for quantification of RCM images. One of the main parts of this application handles the greyvalue manipulation. Apart from standard routines for image optimization and automatic region selection, the mean greyvalue of the whole image was determined. This was used as a reference parameter for the identification threshold in order to minimize the deviation of the identified area from the real area of reflection caused by inconsistencies of the light intensity.Applying the RCM with consecutive image analysis on Gomori-stained neurons of the supraoptic nucleus (SON), we verified the validity of our measuring routine by employing it on a recognized system. A linear correlation between the process of neurosecretion and nuclear volume of the SON is well known. Switching from RCM to bright field, it is possible to obtain topographically identical images of the chosen tissue region. Thus, we photographed reflecting neurosecretory granules in the former while karyometry was performed on the latter. The nuclei were classified by area and matched with the corresponding nucleus area of the same cell. A linear regression of these parameters was computed within a 95% confidence interval. Our results in general confirm former findings about the measurable relationship between nuclear size and specific cell activity, surpassing earlier methods by increasing the sensitivity and decreasing the duration.This paper introduces the RCM combined with digital processing as a useful tool for quantification in the investigative gap between light and electron microscopy. The reflection contrast microscope (RCM; LEICA Germany) is a light microscopic instrument, making reflections along interfaces visible by means of centrally polarized epi-illumination. These reflections cause interference patterns that are suitable for the analysis of superficies or detection of contact zones 1 . We present two properties of this widely unknown technique allowing three-dimensional (3-D) reconstructions in the following manners:

1. Applying the RCM on the surface of air-dried unstained erythrocytes, we made use of the phenomenon that neighbouring zones of equal altitude are joined by closed interference fringes; these can be interpreted as isohypses. To quantify the angle of declivity within a period of the resulting dark-light pattern, it is necessary to know the mean wavelength. The difference of layer thickness between two interference lines of equal tone is about 226 nm for monochromatic light of λ=546 nm. 2 The advantage of this technique is the simultaneous presentation of the profile of all the isohypses in just one picture (Fig. 1) .