The phosphorus L23 edge at 132 eV energy loss is very weak but provides an element-specific signal. Direct imaging of phosphorus using electrons which have lost this amount of energy would require a dose much higher than that known to damage molecules. However low dose imaging is possible in the STEM using the simultaneously recorded dark field annular detector signal to locate particles for image averaging. Summing the weak phosphorus signal from several thousand aligned particles should be adequate to obtain a phosphorus map of essentially undamaged molecules. Several data sets collected on STEM I with ribosomal subunits indicate the feasibility of many parts of the project. In that case the difference of large angle and small angle scattering gavea significant signal when averaged over 1,000 aligned particle images, whereas the TMV (tobacco mosaic virus) control subtracted to background. This may be indicative of the RNA distribution in the subunits, but the arguments are indirect. The elemental map should be much more straightforward to interpret. A key unknown is the degree of localization possible with the 132 eV loss electrons. This is estimated to be better than 0.5 nm, but needs to be tested. Control specimens will be TMV and filamentous viruses which are ideal for alignment and have known or strongly predicted phosphorus distributions. Once the protocols for data collection are established and control spectra match theory, phosphorus mapping will be attempted on 30S and 50S ribosomal subunits. In addition to distinguishing nucleic acid from protein, this technique should be useful for mapping phosphorylated proteins and phospholipids. The same technique can be applied to elemental mapping of B, C, N, O, F, Fe, Mg, Ca and other elements in our specimens using the appropriate energy loss and control signals. With completion of STEM3 in Oct. '95, we have begun EELS work. The first step is to verify the properties of the microscope, the spectrometer, the detectors and the control software. This is nearly complete and we are starting to measure signal to noise (S/N) ratios on various test specimens.
Showing the most recent 10 out of 101 publications
