The technique of photo-electron emission microscopy (PEEM) is based on imaging of photo excited electrons from a surface. Typically ultra violet (UV) light above the work function of a metal will cause electrons to be emitted from a surface. Since photo excited electrons originate very near to the surface, they essentially reflect the electronic structure of the surface. These electrons may be accelerated and imaged, and the image will reflect the properties of the surface.

While the PEEM technique has been understood in a basic sense for many years, it has been limited by the lack of high intensity UV light sources. The most crucial part of the electron imaging system for PEEM, the objective lens, is essentially the same as for the sister technique of low energy electron microscopy (LEEM), and advances in electron optics capabilities have been exploited both for LEEM and for PEEM.

To obtain a full and detailed understanding of the spatiotemporal dynamics of surface processes such as epitaxial growth, strain relaxation, phase transformations and phase transitions, chemisorption and etching, in situ real-time observations have proven to be invaluable. The development of two experimental techniques, i.e. Low Energy Electron Microscopy (LEEM) typically operating at electron energies below 10 eV, and Ultra-High-Vacuum Transmission Electron Microscopy (UHV-TEM) at several 100 keV, has made such in situ studies routinely possible. In many cases, the videodata obtained from such experiments are amenable to detailed, quantitative analysis, yielding statistical, kinetic and thermodynamic information that cannot be obtained in any other way.

I will discuss recent experimental developments, including the design and construction of a new and improved LEEM instrument. Figure 1 shows a schematic diagram of this new machine. There are several features that distinguishes this design from most other LEEMs. One is the use of a 90 degree deflection magnetic prism array,

Due to the submicron size of crystallographic defects, characterization of dislocations has been done. mostly by electron microscopy techniques. Transmission electron microscopy has generated invaluable structural information at the atomic scale. However, the influence of these electrically active defects on carrier transport can only be learned from lower resolution (∼ 1 μm) techniques such as electron beam induced current (EBIC) and photocurrent measurements. Near-field scanning optical microscopy (NSOM) is a novel optical technique that circumvents the diffraction limit. In this talk, we will present the application of NSOM to perform near-field photocurrent (NPC) measurements on strain-relaxed GeSi films on Si substrates to study the electrical activity of individual threading dislocations. Photoexcited carriers are generated locally by NSOM light and are collected by the builtin p-n junctions in the sample resulting in an external photocurrent. As the tip moves across the sample, topographic and NPC images were acquired simultaneously.

There is increasing interest in the quantitative characterization of nanometer-scale semiconductor alloy layers, such as those of SiGe or AlGaAs. Electron holography, which has the ability to detect small changes in mean inner potential Vo, can in principle provide information about local changes in composition. However, the measurement of absolute values of Vo is not always straightforward and values for only a few materials are known with confidence. In this study, we measure Vo for Ge using electron holography, and we compare our experimental data with calculations. The holographic approach that we use to measure Vo is described in the paper of Gajdardziska-Josifovska et al. The Ge samples were nominally 90° cleaved wedges from a <100> wafer. Holograms were recorded on a Philips CM200 FEG equipped with a Gatan 794 CCD camera, and data were obtained for several crystal tilts (determined from diffraction patterns) to assess dynamical contributions to the phases.

It is challenging to access the microscopic distribution of strain in subsurface systems with a lateral resolution better than 1 nm and an accuracy of about 10-3 because such measurements require the detection of displacements by 1 pm at near atomic resolution. Nevertheless, such measurements are commonly desirable, in particular, for the investigations of thin heteroepitaxial films. A particular method developed in the past extracted geometrical information from unit cells in lattice images to perform this task in thin layers of cubic quantum well structures. Strain relaxation in thin TEM samples may falsify such measurements and needs to be controlled. It is the purpose of this paper to describe a quantitative method to measure strain with the required accuracy and resolution in the more complex hexagonal systems. A AxGa1-xN/GaN heterostructure grown on a sapphire substrate was investigated. Local strain comes from compositional variations which alter the size of the unit cells that are shown in Figure 1.

Introduction: The “real” semiconductor processing world consists of the universe of industrial and government semiconductor laboratories and manufacturing facilities plus a small number of university departments with heavy funding from industry—or a consortium of the above. We are bombarded with reports of how the drive toward faster, denser, lower power consuming and more reliable semiconductor products will accelerate with time. The smallest features of today's devices have already shrunk from view in visible light microscopes! Electron microscopy studies of semiconductor device materials, interconnects, insulators, and device packaging materials play a vital role in these extraordinarily “high tech” and financially important arenas.

Shifting semiconductor analytical paradigms: For the most part, the analytical paradigm has shifted: from visible light microscopy to SEM analysis, at low magnification and high accelerating voltage… to much more sophisticated SEM analysis, at high resolution and/or low accelerating voltages… to the many variations of TEM available today… and to new types of scanning probe microscopy.

Extensive long range transport of copper has been observed in polycrystalline aluminum-copper (Al-Cu) interconnects subjected to accelerated electromigration testing. Cu depletion is believed to be the rate limiting factor for electromigration damage accumulation. It is not clear if this is also the case for Al-Cu interconnects with bamboo/near-bamboo microstructures. In order to understand this phenomenon, Cu transport under electromigration conditions in bamboo microstructures has to be characterized quantitatively. This paper describes the procedure used to determine Cu distributions on a series of bamboo test structures subjected to accelerated electromigration (EM) testing. Copper composition profiles were determined for these structures through the use of electron probe microanalysis (EPMA) techniques and customized corrections for non-standard interaction volumes.

Aluminum (0.5 wt.% Cu) interconnect test line structures were fabricated with four metal layers and surrounding silicon dioxide dielectric. Following EM testing, the samples were deprocessed to expose the test lines at metal layer three, as shown in Figure 1.

Perovskite manganites, which have the the general formula R1-xAxMnO3 (R = La, Pr, or Nd and A = Ca, Sr, Ba, or Pb), have generated much recent interest because they exhibit “colossal magnetoresistance” (CMR), i.e. a small change in an applied magnetic field dramatically changes the electrical resistance of the material. Materials that exhibit this effect are being developed for various field-sensing applications but currently, the mechanism by which CMR occurs is not known. Conduction in these materials is explained by the “double exchange” mechanism, where the conductivity is attributed to electrons hopping back and forth between neighboring manganes ions. Such hopping is a maximum when the magnetic moments of the magnese ions are aligned parallel and a minimum when they are aligned antiparallel. In short, these materials show metallic conductivity when they are ferromagnetic and insulating behavior when they are antiferromagnetic. Moreover, they change from metallic to insulating behavior as a function of temperature.

Tin dioxide (SnO2) with rutile type structure exhibits unique electronic and optical properties. In applications of this material as gas sensors, a film-type of SnO2 provides high ratio of surface area to volume and lead to high sensitivity and fast responses. It has been found that substrate material, deposition conditions, and annealing procedure may directly control the microstructure of thin films, hence control gas-sensing properties. In this paper, we present transmission electron microscopy (TEM) studies of the microstructure and crystal defects of tin oxide thin films on sapphire substrate with subsequent annealing at high temperatures.

Tin oxide thin films were deposited on the surface of sapphire by e-beam evaporation of high purity SnO2 (99.999%) at 350°C followed by annealing in air at 600°C - 700°C. Microstructures of the films were characterized by X-ray diffraction (XRD) and TEM techniques.

GaSe, a member of the III-VI compound semiconductors, and its related compounds have recently gained an considerable attention because of their high non-linear optical coefficients in the infrared ranges, making them candidates for second harmonic generation (SHG) materials[l,2]. While the optical properties of those materials in bulk form are quite promising, poor thermal and mechanical properties preclude their easy applications. In thin film devices, the thermal and mechanical properties are dominated by those of the substrate, therefore, heteroepitaxially grown thin films of GaSe and related materials on substrates such as GaAs, Si and A12O3 should enable their application in device structures. Development of such new generation of materials, however, require fundamental knowledge about the surface and interface structure that play decisive roles in the thin film crystallinity and materials properties.

The growth of high quality GaAs layers on single crystal Ge substrates has attracted a great deal of interest in recent years. Currently under development at NREL is the GaAs/GalnP-on-Ge high efficiency multi-junction solar cell. Unlike other heteroepitaxial systems such as GaAs-on-Si, physical properties such as the lattice parameter and thermal expansion coefficient of the Ge and GaAs are much closer. Further, Ge has a bandgap of 0.7 eV that makes it a suitable bottom cell in a multi-junction stack. However, the growth of GaAs on Ge is not a straightforward process, and little is known about the Ge surface and the nucleation of GaAs on Ge in the MOCVD environment. In this study, TEM results show that a number microstructural defects are associated with the Ge/GaAs interface and the initial stages of growth.

III-V nitrides grown on (0001) sapphire substrates often contain a very high density of defects; this high density is usually attributed to the poor lattice and thermal match between the nitride and the sapphire [1]. There is thus a need to understand the defect structures in these materials. In the present study, a microstructural characterization of MBE-grown GaN films on (0001) sapphire using various TEM techniques was performed to evaluate defect structures in this material. This paper will outline some of the different types of defects found in these films and suggest some possible mechanisms by which they were formed.

Cobalt disilicide (CoSi2) is an attractive contact material for Si1-xGex devices due to its low resistivity, low Schottky barrier, and good thermal and chemical stability. Previous studies have shown that blanket Co films on Si0.80Ge0.20 react at annealing temperatures of 700 °C to form CoSi, CoSi2 and a Ge-rich Si1-xGex alloy; no Co-Ge phases were found. The Co preferentially reacts with the Si, which can lead to Ge segregation and island formation. To gain a better understanding of cobalt silicide formation, an electron microscopy investigation of Co/Si1-xGex/Si was performed.

Si1-xGex films (x ≤ 0.3) 64 nm to 260-nm thick were grown by molecular beam epitaxy at 550 °C on phosphorus-doped Si <001> substrates. Co films ranging in thickness from 5 nm to 25 nm were electron-beam evaporated (evaporator base pressure < 6.67 x 10-8 Pa) at room temperature on the Si1-xGex films.

Porous silicon has potential applications in the microelectronics industry. It has been investigated as an electroluminescent source1, as a sensing device in chemical sensors and as an antireflective coating for solar cells. The low efficiency of the solar cells is enhanced by the antireflective coating that porous silicon provides, while the cost of fabrication of the cells is decreased. Porous silicon is normally processed by electrochemical anodization of silicon wafers. Since parameters such as current density, anodization time, and surface conditions can affect the microstructure of the films obtained, we varied the current density to study the effect of the microstructure on the resultant reflectance for the purpose of improving it. It has been found that higher current densities result in higher reflectance films than the lower current densities.

A transmission electron microscopy (TEM) analysis on the regrowth of GaAs by a two-stage reaction between a metal layer (M) and a GaAs substrate is presented. The first stage of the regrowth process is the consumption of GaAs in a low temperature reaction with the metal layer, producing an intermediate phase of (MxGaAs). A second solid-phase reaction, induced by the deposition of Ga or As, results in the decomposition of the intermediate phase and the epitaxial regrowth of a layer of GaAs. The sample growth and reactions were performed in-situ in a molecular beam epitaxy system, using Ni for the metal and As deposition for the second reaction. TEM data confirm the formation of the ternary phase, NixGaAs, and its subsequent decomposition into NiAs and GaAs by reacting with the deposited As. A layer of AlGaAs, 100 nm thick, was grown in all samples as a marker.

High-resolution transmission electron microscopy (HRTEM), multislice image simulation and multiple dark field TEM imaging techniques have been used to investigate the structure of extended defects in AlN and GaN thin films grown on (0001) α-Al2O3 by metal-organic chemical vapor deposition (MOCVD). AlN layers were grown directly on the (0001) sapphire. In the case of GaN thin films, 300-500 Å AlN buffer was deposited first.

Cross-sectional TEM revealed the presence of domain boundaries in these Ill-nitride films. In this study we investigated these defects by multiple dark field imaging technique and proved some of them to be IDBs lying in planes. The multiple dark field images of several adjacent domains of AlN film are shown in Fig. 1 (a, b). The images were obtained exactly in [110] zone of AlN using (0002) and (000) reflections.

The properties of aluminum nitride (A1N) make it an interesting material for applications in the microelectronics industry. These properties include: a large bandgap, good thermal conductivity, high-temperature stability and chemical inertness. Additionally, thin A1N films are used as buffer layers for the growth of GaN on alumina. In the present study, thin-films of A1N have been grown on (000l)-oriented α-Al2O3 through the use of an electric field. At elevated temperatures an electric field applied across a normally insulating oxide material will typically induce a flux of ions that carry current. In the case of Al2O3, since the more mobile species are the Al3+ cations, by applying an electric-field across the material, with an appropriate electrode material, a flux of 3+ cations can be induced toward the cathode surface (negative electrode).

Scanning Capacitance Microscopy (SCM) was first developed in 1985 as a method for sensing tip-to-sample spacing for surface topography profiling in connection with the RCA VideoDisc. Williams and coworkers were the first to use an SCM for obtaining dC/dV doping profiles in semiconductors, albeit with a rather modest resolution of 200 nm. More recently, it has been developed as a 50-nmresolution tool for microscopic doping analysis of semiconductors by measuring the tip-to-sample rf capacitance in an AFM controlled by other means, e.g. by laser beam deflection of a cantilever tip. In this paper we report on the application of SCM to study the 2D doping profiles of InP-based devices, such as multi-quantum well lasers.

It is particularly convenient to prepare cross sections of III-V devices, since the material readily cleaves on [110] planes, as compared to silicon where cross sections must be obtained by painstaking polishing.

TEM has not traditionally contributed to characterization of semiconductor junctions, except for some intriguing results by Perovic, Hull and Alvis where highly doped layers gave contrast under very specific sample preparation and imaging conditions. Since electron holography is sensitive to the phase of the electron wavefront that has passed through the sample, it has the potential to provide voltage contrast. In principle, given its inherent two-dimensional and quantitative character, electron holography is a strong candidate for analysis of semiconductor device junctions.

The sample for the work reported here was one of a group of test specimens fabricated at IBM for a dopant metrology round-robin comparison to evaluate various profiling methods. The substrate was <100> p-type silicon, boron-doped at 11-25 Ω-cm-1. The test structure was fabricated using low- temperature epitaxial growth and consisted of an abrupt p-n junction, formed by abutting 1020cm-3 doped regions of boron and phosphorus

It has been demonstrated that doped regions in semiconductors yield contrast in secondary electron images taken under appropriate imaging conditions. Perovic et al. attributed this "electronic" contrast to surface potentials arising from the bending of the energy bands at the surface of the semiconductor. This band-bending results from the presence of surface states which pin the Fermi level. These effects have found application to an important problem in the electronics industry -- the profiling of dopant distributions in two-dimensions. However, such dopant distributions are rarely uniform, i.e., the dopant concentration may change over many orders of magnitude within distances of only a few tens to hundreds of nanometers. Therefore, it is necessary to understand how the observed contrast from the doped regions varies for non-uniform dopant distributions to apply the technique to real electronic devices.