Introduction. By volume of material produced, optical disc manufacturing is the world's large nanotechnology industry, far surpassing semiconductors and magnetic disks. The development of various optical disc media has introduced a variety of challenges, which necessitate viewing the geometrical structure of the bumps, pits, or grooves on the surface of the discs. The Atomic Force Microscope (AFM) has the resolution and sensitivity needed for device analysis, but in the past has been limited by magnification errors, image distortion, data handling, and calibration. Furthermore, the common measurement procedure for AFM images consists of manual dimensional measurements. This introduces error and limits the number of measurements taken at one time. We have developed an automated process, which overcomes these limitations and provides a comprehensive approach to feature measurement, data analysis, calibration, and reporting.

The electrochemical formation of gradients in self assembled monolayers has been demonstrated recently [1]. The capacity to image these gradients provides useful information on the physical chemistry of electrochemical striping.

Imaging chemical gradients requires the ability to sense the chemical moiety on the top of the self-assembled monolayer. This has been accomplished by derivatizing an atomic force microscope (AFM) tip with molecules selected to have specific interactions with the sample in a technique known as chemical force microscopy [2]. Typical tapping mode AFM is then used to image the sample; the tip is oscillated vertically above the sample and the tip-sample interaction modulates the amplitude of the tip.

The sample adhesion, sample stiffness, and sample topography all influence the oscillation amplitude of the tip. Pulsed Force Mode (PFM) [3] is an extension for atomic force microscopes. The PFM electronics introduces a sinusoidal modulation to the z-piezo of the AFM with an amplitude between 10 to 500 nm at a user selectable frequency between 100 Hz and 2 kHz.

Silicon Drift Detectors (SDD) are being developed for analytical x-ray spectrometry having large active area, high-energy resolution and capability of operating at high counting rates. The development derives from the charged coupled device (CCD) for light-signal imaging, utilizing the extremely low capacitance of the detector and readout electronics and subsequent developments of silicon drift detectors for high-energy physics applications and more recently, x-ray spectroscopy applications. The now well-known advantage of the drift detector design is that, unlike traditional planar detectors, it allows for relatively large active area while still maintaining a very low anode capacitance (60 fF). This low value of detector capacitance results in a lowering of the series-noise component and hence the overall inherent electronic noise. Additionally, the reduction of the series noise leads to faster optimal shaping time, and as a consequence this provides for extremely high count rates.

Samples to be studied using EELS techniques are more or less sensitive to the effects of the radiation exposed to them by the beam. Quite often the sample is damaged due to the high radiation doses in the areas where the beam has been focused. Secondly, the build up of carbon contamination on the surface of the sample caused by the beam affects the results of the analysis if the sample itself contains carbon. The solution to these problems is to minimize the exposure times needed to record the spectrum, which will lower the radiation-doses on the sample. In addition, problems related to beam instabilities would be reduced due to the shortened acquisition times. However, in order still to achieve an acceptable signal-to-noise ratio the need for a detector with high detective quantum efficiency is essential. We have investigated the possibilities using silicon detectors such as photodiode arrays (PDA) and charge coupled devices (CCD) in direct electron detection mode using 100 keV incident electrons.

Excellent linearity and high sensitivity have made SSCs the ideal image detector for almost every TEM application. Their ability to make high quality digital images available within fraction of seconds for further evaluation and processing in a PC, have made them a non-dispensable accessory for any modern TEM. However, despite their excellent characteristics, SSCs provide a restricted number of individual image points in respect to a negative, what is considered to be the main disadvantage of this detector. To compensate for this, CCDs with 2048x2048 pixel are available since some time. SSCs using these 2kx2k CCD arrays not only provide 4 times the pixel number but also offer a lot more options people have waiting for: e. g. highly resolved low-dose or ESI images with significantly improved signal to noise ratio, or higher resolved images for diffraction analysis and holographic reconstruction.

There are many advantages of IVEMs over more conventional microscopes, all of which make it possible to obtain more accurate data at higher resolution. One significant drawback of higher accelerating voltages is degraded performance of the electron detector, a problem that is particularly serious with CCD cameras. At any voltage, one faces a tradeoff with the CCD performance between sensitivity and resolution. Electron scattering within the scintillator and its support broaden the point spread function, while the decreased scattering cross section at higher voltages decreases the light output from the scintillator. At voltages much above 100 kV these effects have been found to seriously limit CCD performance. With the trend towards electronic data collection and microscope automation, it is essential that a high performance detector be developed for the IVEM. One approach to optimizing the performance tradeoff employs lens coupling to improve resolution, at the expense of sensitivity.

In order to increase analytical capabilities of SEM we have developed a multifunctional detector unit, which allows the simultaneous and highly efficient registration of the information from the chosen area of the specimen using backscattered electrons (BSE) and cathodoluminescent (CL) radiation. At the same time secondary electrons (SE) could be collected by the regular Everhart - Thornley detector. Thus, the unit allows to perform microstructural investigations in three modes of SEM operation simultaneously - BSE, CL and SE. A schematic diagram of the unit is presented in Fig. 1.

In the plate made of scintillating material, and installed between the pole piece of SEM and the specimen, the bottom side is a part of spherical surface. The surface covers completely the whole top surface of the specimen and collects BSE and CL radiation in the solid angle ∼2 π sr.

The commercial introduction of high-count-rate, near-room-temperature silicon drift detectors (presently available) and high-energy-resolution cryogenic microcalorimeters (forthcoming) is an exciting development in x-ray microanalysis, in which detector choices and capabilities have been essentially stable for many years. Both of these new energy-dispersive detectors promise improved capabilities for specific applications, e.g., faster EDS mapping (silicon drift detectors) and nanoscale particle analysis (microcalorimeters). In this paper, we briefly examine some of the important benefits and drawbacks of microcalorimeter EDS (μcal EDS) for x-ray microanalysis.

The primary benefit of μcal EDS over conventional semiconductor EDS is the factor of ∼ 20 improvement in energy resolution (∼ 4 eV, real-time analog signal processing), as shown in Figure 1.

Based on excellent energy resolution and single quantum detection sensitivity, cryodetectors are offering a variety of new, analytical solutions for the analysis of elementary surface compositions, especially for the analysis of light elements and very small sized structures. Cryodetectors operate typically at temperatures between 30 and 200mK and require vibration free and fully automated cooling systems in order to qualify for industrial applications. Cryodetectors are low temperature superconductors where the two most prominent types are based on microcalorimeter and tunnel diode principles. Cryodetectors are mainly employed for surface analysis applications as energy dispersive X-ray spectrometers with energy resolutions of less than 15eV, but may also be used as highly sensitive UV, VIS or even mass spectrometers in the future.

Conventional EDX detectors are semiconductors. An impinging X-ray quantum creates a number of electronhole pairs dependent on the energy of the triggering event thus allowing energy dispersive measurements. The performance limit of semiconductor detectors has almost been reached and is determined by the excitation energy necessary to create electron-hole pairs.

While the performance of single-pixel microcalorimeter x-ray detectors has been shown to be very useful for microanalysis1, arrays of microcalorimeters will have a further impact on energy dispersive spectroscopy (EDS) by providing higher total count rates and larger effective areas. We are developing a system with a 32x32 pixel array with a total area of 40 mm2 and with a total count rate between 105 and 106 s-1. One of the challenges in constructing an array-based system will be the readout electronics.

In a typical microcalorimeter EDS setup, shown in Fig. 1, the microcalorimeter is readout using a cryogenic amplifier (SQUID system) and analog feedback electronics to control the SQUID. The signal from the room-temperature SQUID electronics is either attached to a spectroscopy-grade shaping amplifier and MCA for real-time analog signal processing or attached to a digitizer for optimal pulse processing.

The observation of nano materials or nano phases is generally performed using Transmission Electron Microscopy (TEM) because conventional Scanning Electron Microscopes (SEM) do not have the resolution to image such small phases. Since the last decade, a new generation of microscopes is available on the market. These are the Field Emission Scanning Electron Microscope (FE-SEM) with a virtual secondary electron detector. The FE-SEM have a higher brightness allowing probe diameter smaller than 2.5 nm with incident electron energy, E0, below 5 keV. Furthermore, what gives FE-SEM outstanding resolution is the virtual secondary electron (SE) detector. The virtual SE detector is located above the objective lens and it is also named a through-the-lens (TTL) detector. Therefore, the SE images are mostly made up of all SE of type I and II, because those of type III, which are generated by the backscattered electrons in the chamber, are not collected.

The resolution of a low-voltage electron microscope is limited by the chromatic and spherical aberration of the objective lens, see Fig. 1. The design of state-of-the-art objective lenses is optimised for minimal aberrations. Any significant improvement of the resolution requires an aberration corrector. Recently, correction of both Cc and Cs has been demonstrated in SEM, using a combination of magnetic and electrostatic quadrupoles and octupoles (Zach and Haider, 1995). The present paper presents an alternative design, which is based on a purely electrostatic concept, potentially simplifying the ease-of-use of an aberration corrected microscope.

In 1936 Scherzer showed that the fundamental lens aberrations of round lenses are positive definite, in absence of time-varying fields and/or space charge. Negative lens aberrations, required for the correction of Cc and Cs, can only be obtained using non-round lenses, e.g. quadrupoles and octupoles (Scherzer, 1947).

Aside from the electron column, there are many factors that determine image quality in an ultra high resolution Field Emission Scanning Electron Microscope (FESEM). To obtain quality images, the operator must minimize sample contamination and be able to select the appropriate signal generated from the sample surface. Hitachi has developed a new in-lens FESEM that both minimizes surface contamination and improves signal efficiency.

Sample contamination is reduced by incorporating an oil free vacuum system, consisting of a 300 1/s turbomolccular pump and a 200 1/min scroll pump. For demanding applications, an additional 50 1/s turbomolccular pump can be added to back the main high vacuum pump. Furthermore, sample contamination caused by electron beam irradiation has been reduced by employing a high speed electrostatic beam blanking system. By blanking the beam during scan retrace (flyback), the electron dose accumulated on the sample can be minimized.

It has been estimated that more than 90% of all scanning electron microscope (SEM) images ever published have been obtained using secondary electrons (SE) which are defined as being those electrons emitted with energies between 0 and 50eV. The properties of these secondary electron are therefore of considerable interest and importance. However, although secondary electrons have been intensively studied since their discovery by Starke in 1901 the majority of the work has been aimed at determining the SE yield coefficient and its variation with energy for elements and compounds. The energy spectrum of secondary electrons has received far less attention although it is evident that the form of the spectrum must have an effect on the image contrast observed in the SEM because SE detectors are energy selective devices. The few studies that have been made have mostly concentrated on spectra obtained from clean samples observed under ultra-high vacuum conditions. This is understandable, because it is certain that the presence of a surface layer of contamination will change the SE spectrum to some degree or other, but it is unfortunate because all specimens in real SEMs are dirty and it is information about this situation that is required.

From the best fit of two adjustable parameters, B (escape probability) and ƛ (empirical escape depth of the secondary electrons, s.e.), the constant loss theory accounts fairly well for the s. e. emission from materials: δ (E0) = B [E0/E(e.h)] [ƛ/R] [1 - exp. -(R/ ƛ)] (1) (with E(e.h): energy required for generating a s.e. inside the specimen; R: range of the incident electron related to the primary beam energy, E0, via a power law of the form: R=C E0n). The influence of the angle of incidence, i, may also be estimated by changing R into R cos i (see fig la for the example of diamond with B/ E(e.h) = 22 keV-1; ƛ ≈ 10 nm). For insulators, δ is obtained from short pulse experiments (to prevent charging) and, in the 1-5 keV range, it is often one order of magnitude (or more) larger than that of metals via a larger ƛ value (also more sensitive to the temperature, to the crystalline state and resulting from a random walk involving many elastic and inelastic-phonon- scattering events).

Under high-energy electron bombardment, electrical charge can rapidly build up in non-conducting materials. The electron-induced charging process is very complex and is still not well understood. Injection of energetic electrons into insulators generates electron-hole pairs which can be localized at pre-existing or newly created charge-trap centers. Charging of insulators results from a balance between the emission of secondary electrons and the diffusion and trapping of incident electrons. In the SEM, charging is usually a dynamic process: the degree of specimen charging depends on the energy and the dose of the incident electrons, the image magnification, the pixel dwell time (PDT), and the history of the electron irradiation of the specimen. Although the degree of charging can be reduced by varying the energy of the incident electrons, charging of non-conducting specimens cannot be completely eliminated due to variations in the local SE yield across the specimen surface and the presence of charge-traps inside the specimen.

As the semiconductor industry further shrinks feature sizes, new analytical solutions are required for elemental composition analysis. Conventional EDX analysis is limited by the obtainable energy resolution. In order to analyze extremely small impurity particles, material doping, interdiffusion effects or small chip structures, the diameter of the area hit by an electron beam and particularly the volume of electron-surface interaction has to be reduced accordingly. This can be achieved by reducing the electron beam energy.

X-ray spectroscopy is most frequently used for damage free material testing. Nowadays, semiconductor based energy dispersive X-ray spectrometers (EDS) are employed on a routine basis and are frequently used in combination with scanning electron microscopes (SEMs). But performance is limited by their rather poor energy resolution, which is typically in the 130eV range for a 6keV X-ray line, and their limited capability of resolving X-ray lines within short spectral ranges. Further, overlaps of spectral lines are most critical in the low energy range.

Experimental work in a number of research groups over the past five years have demonstrated, unequivocally, that low voltage i.e. 5kV and below, quantitative x-ray microanalysis is a valid analytical technique which can be applied to a wide range of specimens. The advantages of this approach are now generally well known and include improved image contrast, diminished bulk charging of the sample and for bulk specimens, a reduction in the size of the beam-sample interactive volume from which the x-rays are generated, and as a consequence, a considerable improvement in the microanalytical spatial resolution. In addition, low voltages ensure a sensitive surface analysis procedure which provides true topographic contrast and that in the case of light element analysis, quantitation is simplified because the atomic number (Z), absorption (A) and fluorescence (F) effects are minimised. All these advantages are improved further if one uses a field emission gun as the primary electron source.

Joy and Pawley (1992) defined the limitations of spatial resolution in the SEM as follows. “The spatial resolution of the scanning electron microscope is limited by at least three factors: the diameter of the electron probe, the size and shape of the beam/specimen interaction volume with the solid for the mode of imaging employed and the Poisson statistics of the detected signal.” The interaction volume of the primary beam in the sample increases with increasing accelerating voltage (Vo), because electrons with higher energy penetrate deeper into the sample. The interaction volume affects the spatial resolution, because the electrons that finally form the image do not only originate from the impact point of the primary beam, but from the much larger interaction volume. It is not simple to understand how this affects spatial resolution, since resolution is also heavily influenced by the properties of the sample surface (the surface coating) and it is the issue of ongoing debates.

Surfaces The majority of the SEM secondary electron signal (SE) comes from the first few (<2nm) nanometers of the surface. The quality of the outermost few atomic layers of an exposed and reactive sample can therefore have a profound effect on image contrast. With light elements and low voltages the SE signal is dominated by scattering at the point of incidence of the primary probe, but more generally there is a significant contribution to surface SE generation by primary electrons backscattered from deeper in the sample. With the advent of very low voltage EDS, and the application of more conventional surface science methods, we learn that even without a coating the surface of a sample may be atypical of the bulk (Fig. 1). Contamination of an otherwise clean surface occurs by transport of the sample through the air, leading to oxidation, hydrocarbon accumulation, and occasionally to rather more dramatic pyrophoric effects.