Imaging with Surface Spectroscopies

  • PDF / 4,056,421 Bytes
  • 4 Pages / 604.8 x 806.4 pts Page_size
  • 80 Downloads / 230 Views

DOWNLOAD

REPORT


f the information. What the materials researcher gets from surface analysis are spectra, data tables, or the like. They need to be integrated in a way that tells, for instance, what structure the process is making. Hère is where imaging brings value: it provides part of the intégration process. Superimposing the concentration of (for example) chromium on the non-etched portions of a copper printed wiring board does more than just display what could be learned from a stack of spectra taken at various points. The pattern of contamination might—and in this example did—point to malfunction in an earlier process step. The image organized the information, preparing it for rapid interprétation. How It Works It is useful to first recall just what surface analysis sees (the first three références give more détail). In gênerai, we bombard the surface with a species that will drive an atomic-scale process that places the sought-for information on an outgoing species that we will detect. Electron spectroscopy for chemical analysis (ESCÀ), the oldest and probably most familiar example, uses soft x-rays (Mg-K, Al-K) to eject électrons from core levels with kinetic énergies from near that of the x-ray photon down to zéro. The mean free path in solids of such électrons is a few nanometers, so that any we detect must hâve corne from the outer few layers, making ESCA inherently surface sensitive. The électron energy distribution gives elemental composition and chemical state information. Also called XPS (x-ray photoelectron spectroscopy), it still accounts for the majority of surface analysis. Auger électron spectroscopy (AES) provides substantialîy comparable information using électron bombardment. Ion bombardment ejects the material composing the surface itself, which is examined by mass spectroscopy: secondary

ion mass spectroscopy (SIMS) and related approaches. The discussion could continue; there are more than 100 technique acronyms. Most surface analysis images are "maps." We collect the necessary part of the spectrum at a séries of locations laid out to cover the région of interest. We could then acquire at each a complète spectrum having signal-to-noise quality adéquate for quantitation at a 10% statistical uncertainty level. The image could then be made from any peak, however we might la ter décide. Considering that such a spectrum takes several minutes to over an hour to acquire, it is easy to see that a typical 256 x 256 pixel image requires a différent approach! Instead, in our laboratory we acquire the spectrum of a very large area, then sélect the peak of interest and adjacent background région. The data quality of the spectrum also tells us how long to count for each. Then at each analysis location, we acquire just the peak and the background, not the whole spectrum. Subtracting gives the concentration. Even counting just two small slices of spectrum, the collection time for a full image can still be unacceptable for low concentrations. More generally, to construct the image, we assign to the chosen spectral property (