Although the benefits of operating the scanning electron microscope at low beam energies have been evident since the earliest days of the instrument, the successful utilization of the SEM under these conditions has required the development of high brightness field emission electron source, advanced lenses, and clean vacuums. As these technologies became available the level at which imaging became regarded as “low energy” has fallen from 10 keV, first to 5 keV, and more recently to 1 keV. At this energy state of the art, instruments can now provide an excellent balance between resolution - which becomes worse with decreasing energy - and desirable goals such as the minimization of sample charging and the reduction of macroscopic radiation damage - which tend to become more challenging as the energy is increased.

Most of the applications in microscopic image analysis are still salved by a heuristic method that focuses on the specificity of each problem. This is particularly true for applications involving color images. Because of this, the design of robust algorithms has remained time consuming and is usually performed by an image analysis expert with many years of experience. How then can the knowledge of the expert be transferred to the less specialized microscopist? How can the specificity of a typical application be related to general color image analysis tools?

Only a general approach can work towards decreasing the heuristic aspect of the design phase and help the common user perform image analysis without the assistance of an expert.

The recent note by J.D. Watson in Science (1993) is another of the periodical advice columns in which someone of renown gives his or her perspective of how to succeed in this most difficult of fields. And a very good article it is. I find no fault with any of the four rules presented which I quote as follows: 1) “To succeed in science you have to avoid dumb people.” 2) “To make a huge success, a scientist has to be prepared to get into deep trouble.” 3) “Be sure you always have someone up your sleeve who will save you when you find yourself in deep s—.” 4) “Never do anything that bores you.” Unfortunately, these four rules are directed to a level of scientist that clearly bypasses most of us irrespective of how many good intentions we may have. Therefore I feel obliged to submit a revised list of thoughts for us average beings who plow through the days in an average research environment.

Growing your own microscopy client/server image database requires addressing many issues. If you are going to grow your own, the following ideas will get you “out of the box.”

The ideal microscopy image database environment might appear like this: The core of the image database would be a very powerful, universally available, extensible, cost effective, flexible and serviceable database engine. The core engine would allow the user to perform fast and seamless SQL searches across a large database without reducing the available network bandwidth. It would be able to take advantage of current and future technologies (i.e., 64 bit processing, 1 GB networks, digital, video, multi media, etc.) The same core would be able to share information with other databases, applications and platforms. What an odd concept.

A precision sectioning method is described in which the diamond blade used is specifically mounted at high tension, with the sample being fed into the blade at its inner annulus. The sawing action which results is characterized by low kerf losses, the possibility for thin slices, good surface finish, high straightness, high reproducibility and low damage. Applications which benefit from the method are mentioned, namely in the sample preparation of semiconductor, geological, composite, hard tissue and polymeric samples, as well as most non-biological samples for many areas of microscopy.

My Inter/Micro-84 presentation and subsequent article in The Microscope dealt with optimizing diffraction limited depth of field in photomacrography and is referenced by Brian Bracegirdle in his “Scientific Photomacrography. My analysis was based upon the classical solution for the diffraction pattern image of a point light source, using the Rayleigh criterion of resolution. H. Lou Gibson's method of treating combined diffraction and geometric blurring away from the object focal plane was used. The calculations and the experimental results indicated that a final print resolution, for the object focal plane, of 7 lines/mm (.29 mm Airy disk) gives an optimum balance of depth of field and resolution. The 7 lines/mm criterion is equivalent to an Abbe criterion of total magnification equal to 440 times the numerical aperture. The 500 times numerical aperture Abbe criterion recommended for microscopists with very acute vision corresponds to a maximum resolution of 6 lines/mm in the final image.

For some time now I have been studying Antony Van Leeuwenhoek's microscopes and building prototypes, and have finally settled on a design that I would like to offer to the public.

They are made of thin beaten brass, about 47 mm tall by 25 mm wide, with the lens captured between two such sheets. The subject is placed on a metal pin, adjusted with screws, to change the focus and position. One of the adjusting screws protrudes down from the body and also serves as a handle. These microscopes look very much like the one in the museum at Utrecht (circa ∼ 1690). They have a lens made of a 2 mm diameter glass ball and have a power of about 180X, with surprisingly good resolution.

The first in a series of articles looking at developments in X-ray detectors and Field effect transistors. In recent years we have seen both Si(Li) and Intrinsic Silicon Detectors with resolutions below 130 eV. These same detectors are capable of measuring the soft X-rays like Be K, B K, C K etc. with good resolution and with peak positions reasonably close to the exact position for these lines. The Carbon peak is the one most affected by the L absorption edges in Silicon and in some circumstances can be shifted as much as 30 eV down in energy. Shifts can also result from the settings of amplifier threshold discriminators and ADC designs. At higher energies (e.g. Mo K) the Si(Li) detector, if not properly drifted can also have a down shift by as much as 200 eV.

Over the past 20 years great improvements have been made in the techniques available for doing 3D fluorescent microscopy on living cells. The first approach, generally referred to as image deconvolution, treats the stack of 2D widefield (WF) image data as merely the sum of a number of discrete point-spread functions (PSF) and uses the computer to find the array of emitters'that, when blurred by the PSF, best fits the stored data. If the PSF is known, only presence of statistical and electronic noise in the data, prevents this best-fit set of emitters from being a perfect image of the dye distribution in the specimen. The crucial role played by noise can be appreciated by comparing images from the Hubbell space telescope in its original condition, even after deconvolution, with images taken after the optics had been repaired.

There has been a recent uproar about falsified data in biomedical research. This occurs with some regularity, and even Congress has been aroused recently with the justification that the perpetrators and the universities where they worked had received public grant monies. The incidents may have been relatively few in numbers, but the fires have been fanned by the media. Less publicized, but perhaps more disturbing is the apparent conflict of interest spread by many new biotechnology companies. Researchers who develop and evaluate their products are frequently given substantial chunks of the companies whose products they are testing.

If you want to collect X-rays on your SEM or TEM you have the choice of using an energy dispersive spectrometer (EDS) or a wavelength dispersive spectrometer (WDS). Which is better? That depends on what you want to do with it.

WDS is much more sensitive and has much higher resolution than EDS. Sensitivity means that WDS has a much lower minimum detection limit for trace elements. WDS can be up to several orders of magnitude more sensitive. This does not mean it is more accurate. Accuracy of analysis can be just as good for EDS as with WDS if the analyst takes the same care in sample prep, data gathering, and standardizing. Sensitivity simply means that WDS can detect much smaller amounts of a particular element than EDS. Why? Because the superior resolution provides much better signal to noise ratio than with EDS. Noise, in this case, is mostly contributed by background radiation from the sample. Because of its superior resolution, WDS can also resolve peak overlaps.

In the November issue of this publication, we described how focused ion beam (FIB) microsurgery is used to successfully cross-section and prepare material-specific samples for SEM and TEM analysis. Material specific samples have two or more components possessing different characteristics, such as hardness and chemical etch or sputtering rates. Traditional sample preparation techniques (mechanical grinding and polishing, broad ion beam etching, and chemical etching) alter, delaminate, or destroy these samples.

FIB handles the preparation of these difficult samples well because of its milling geometry and the high current density of the small beam.