Abstract Nanoscale devices such as carbon nanotubes, fluorescent nanoparticles, and molecular conductors serve as a benchmark for the tools and techniques that will be required in the future to analyze processing defects and determine the cause of electronic device failures. In this paper, the authors compare and contrast the nanoscale capabilities of several imaging techniques, including STEM-in-SEM, forward scattered electron imaging, photoelectron emission microscopy, and atomic force microscopy. Along with the results of the various characterization techniques, the paper also includes a survey of the more common nanoscale devices, showing that many require more than one imaging method to make a complete and accurate assessment.
Reactive ion beam etching of polyimide thin films was investigated using x-ray photoelectron spectroscopy (XPS) and etch rate measurements. The etching mechanism and the near surface damage produced in polyimide by exposure to argon and oxygen ion beams were compared. The etch rate of polyimide by oxygen ions was studied as a function of ion current density and neutral oxygen molecular flux, and the results were found to match a model for the contribution of neutral fluxes to the etch process. Ion beam etching with inert argon ions was found to produce a graphitelike layer on polyimide. Reactive ion beam etching with oxygen ions resulted in much faster etching than for argon ions, and did not produce a graphitized layer.
Abstract We report on a quantitative investigation of doping-induced contrast and topography-induced contrast in photoelectron emission microscopy (PEEM). Calibration samples were fabricated using standard photolithography and focused ion beam writing to test both types of contrast. Using a near-threshold light source, we find that the doping-induced contrast increases monotonically with B concentration over the measured range of 1017 – 2x1020 cm-3. The variation in doping-induced contrast as incident photon energy is varied was also investigated. Optimal doping-induced contrast and PEEM sensitivity is achieved by imaging with photon energy slightly above the highest nominal photothreshold of interest. The photoemission model, based on near-threshold emission, used to describe doping-induce contrast gives good agreement with the measured intensities. Thus, measuring the relative intensity ratio provides a robust technique for determining doping levels. Topography-induced contrast was investigated by imaging Ti samples of various step heights (75, 150, 290, and 550 nm). Image data suggests that edge contrast increases with step height. Numerical simulations show that non-uniform electrostatic fields at step edge are responsible for this contrast. Experimentally, we systematically vary the lateral field strength and show that edge contrast can be controlled. This technique could be useful in failure analysis by identifying breaks in metal lines.
Abstract Blind deconvolution techniques were used to enhance scanning electron microscope (SEM) images in the range of 200,000x to 500,000x magnification. Typical SEM samples were imaged including a gold island reference standard, a plasma delayered integrated circuit, and an integrated circuit cross section. Image resolution improvement up to 40% was observed. However, it was necessary to use 16-bit TIFF images with greater than 120:1 signal to noise ratio, which required 10 minute frame times.
We demonstrate that scanning SQUID (Superconducting Quantum Interference Device) microscopy is a fast and easy method for finding the location of power-to-ground shorts in a series of chip-first MCM (multi-chip module) samples. Previous work has shown that the scanning SQUID microscope is capable of locating shorts in a C4 chip carrier and a ball grid array package, however the physical analysis was not performed to verify the (presumed) fail locations. In the present work, shorted areas in a chips-first MCM were located by SQUID microscopy and confirmed by mechanical cross-sectioning. When a short in a second MCM was found by SQUID microscopy and repaired with a laser, the power-to-ground resistance increased by more than a factor of 1000, bringing the part within specification. Techniques such as emission microscopy were unsuccessful in detecting the shorted current paths in these devices. Thus scanning SQUID microscopy is the only known method for locating and repairing defects in this product.
The thermal stability of epitaxial SrTiO3 thin films grown by molecular-beam epitaxy on Si (001) has been studied using x-ray diffraction (XRD), optical microscopy (OM), scanning electron microscopy (SEM), and thermodynamic calculations. Our studies focus on the stability of the SrTiO3/Si structures under the conditions typically employed in the pulse laser deposition (PLD) growth of complex metal oxide heteroepitaxy on Si. We observe additional Bragg peaks in thermally treated SrTiO3 buffered Si structures, corresponding to possibly TiSi2 and/or SrSiO3, reaction products which are consistent with the reaction schemes we propose. In addition, OM and SEM reveal microstructures that are not readily accounted for solely by the solid state reactions as put forth by previous workers but can be reasonably explained by our proposed reaction schemes. Using our observations and thermodynamic analysis, we argue that reactions involving the gaseous species SiO(g), the reactivity of which has not been previously considered in this system, are important. We attribute the onset of degradation of the SrTiO3 film at high temperatures, to the circular void forming reaction Si(s)+SiO2(s)→2SiO(g) at the interface and suggest that the reactions considered by previous workers involving all solid state reactants occurs only at the conclusion of degradation. Our results points to the need for keeping the PLD temperature as low as possible to minimize the production of reactive SiO(g) in avoiding the deleterious reactions.
The diffusion of deuterated polystyrene (d-PS) in a polystyrene matrix was used to probe the damage to the polymer surface caused by reactive ion beam etching (RIBE). Diffusion was seen to be hindered in a d-PS film treated by RIBE, an immobilization apparently due to crosslinking of the surface monolayer of the polymer sample.
Abstract This article describes two innovative methods that can significantly improve the resolution of SEM imaging: scanning transmission electron microscopy in a scanning electron microscope (STEM-in-SEM) and forward-scattered electron imaging (FSEI). Both methods can be implemented in any SEM using special sample holders. No other modifications are required. Test results presented in the article show that 1 to 2 nm resolution is possible in thin sections, uncoated polysilicon gates, and photoresist.
Abstract Scanning electron microscopy (SEM)/energy dispersive x-ray spectroscopy (EDS) is generally thought of as a bulk analysis technique that is not suited for nano-scale analysis. This paper discusses several options for reducing or eliminating the interaction volume size and obtaining x-ray data with much higher spatial resolution and surface sensitivity than is typically achieved in the SEM. These include collecting data at very low accelerating voltages to minimize beam spread in the sample, tilting the sample to keep the interaction volume near the surface, and analyzing thin sections to reduce or eliminate the problem of beam spread in the sample. Computer software simulations, in conjunction with experimental data are used to illustrate these methods. The paper also discusses issues effecting EDS analysis in the environmental SEM. It has been shown that computer modeling is a useful tool for determining the optimum beam conditions to improve energy dispersive analysis in the SEM.
