Current non-destructive elemental characterization methods, such as scanning electron microscopy-based energy dispersive spectroscopy (SEM-EDS) and micro X-ray fluorescence spectroscopy (MXRF), are limited to either elemental identification at the surface (SEM-EDS) or suffer from an inability to discriminate between surface or depth information (MXRF). Thus, a non-destructive elemental characterization of individual embedded particles beneath the surface is impossible with either of these techniques. This limitation can be overcome by using laboratory-based 3D confocal micro X-ray fluorescence spectroscopy (confocal MXRF). This technique utilizes focusing optics on the X-ray source and detector which allows for spatial discrimination in all three dimensions. However, the voxel-by-voxel serial acquisition of a 3D

No abstract

No abstract

We propose a novel and highly efficient waveguide polarizer based on the phenomenon of resonant tunneling. Based on this principle it is possible to fabricate both TE and TM pass polarizers. We show that by proper choice of materials it is possible to obtain TM mode loss greater than 80 dB/mm with a TE mode loss of only 0.5 dB/mm for the TE pass polarizer and TE mode loss up to 24 dB/mm with a TM mode loss less than 0.8 dB/mm for the TM pass polarizer.

Electron or x-ray excited X-ray fluorescence analysis is a widely deployed technique in various fields such as materials science, semiconductor manufacturing and defect review for thin film characterization. Layer thicknesses with sub nanometer accuracy and elemental composition to below a weight percent can be measured. With electron beam instruments such as SEMs, spatially resolved fluorescence maps are obtained by raster scanning a finely focused electron beam using energy or wavelength dispersive spectrometers (EDS/WDS) to collect elemental distribution and concentration information. The spatial resolution for these maps is fundamentally limited for thick (>1um) samples by the electron interaction volume to the order of one micrometer. This restriction can be overcome by the use of high-resolution Fresnel zone plate lenses as x-ray imaging optics. Based on this concept we have designed an x-ray fluorescence imaging spectrometer which combines the elemental identification capabilities of a spectrometer with the high spatial resolution (sub-80nm) of zone plate imaging optics. Such a spectroscopic imaging system can potentially be employed advantageously in many semiconductor applications. As a first application we demonstrate sub-surface imaging of copper interconnects and identification of manufacturing problems and failures.The system described in this paper is operated on a scanning electron microscope and is optimized to image copper structures in ICs. It is equipped with a zone plate manufactured by Xradia Inc. to image the 930eV x-ray fluorescence line of copper (CuLα 1 line) at an imaging resolution of 50nm. The system can also be tuned to other x-ray fluorescence wavelengths to image elements different from copper. The description of the instrument and its capabilities are described already in detail elsewhere [1]. To demonstrate the spatial resolution of the imager, we imaged a copper resolution target with 70nm thickness and a copper grating structure and these are shown in fig 1. The copper test target was first imaged with a field emission SEM (shown in fig 1A) to show the structure of the target and then with the fluorescence imager ( fig 1B). The narrowest separation between the lines is 50nm. Fig 1C shows the x-ray fluorescence image of a copper grating structure with a period of 140nm. From figs 1B and 1C, it was concluded that the spatial resolution of the imager is limited currently to approximately 65nm. Furthermore, the xray fluorescence image in fig 1B shows an inhomogeneous distribution of copper which are not apparent in the SEM image. This is a direct consequence of the different contrast mechanism. Fig 2A shows the image of a three layer copper chip [2]. All three layers are visible and in focus at the same time. The width of the thinnest metal line is about 250nm. The bright dots are vias connecting various metal layers. Fig 2B is obtained by convolution of fig 2A with a point spread function of 1um and assuming a pixel size of 0.5um; typical parameters normally encountered in EDS or ...