IDENTIFYING ELEMENTS VIA XRF> X-RAY FLUORESCENCE

(10000)
  • Next Previous
    • Geo>K0FF
      Dec 18, 2012
      IDENTIFYING ELEMENTS VIA XRF> X-RAY FLUORESCENCE
      The natural characteristic X-Rays unique to each element on the Periodic Table of the Elements!
       
      For a long time XRD or X-Ray Fluorescence has been used to identify elements by exciting them with an external form of energy (X-Rays) and reading the resulting "Characteristic X-Rays" subsequently given off by those elements when they relax back to stability. Commercially, detection is done with extremely expensive, usually cry-cooled, detectors.

      Similarly, the same process occurs in a Scanning Electron Microscope using beams of electrons as the exciting energy. Micro beams allow very precise areas of a sample to be analyzed, one by one.

      Since we don't have a handy X-Ray micro beam or Scanning Electron Microscope here in the Home Lab, much less an HPGe probe cooled by liquid nitrogen, we use what we have: inexpensive, small sealed exempt quantity radioactive test sources, mostly by Spectrum Techniques - and a scintillation type detector fed into a student grade MCA- in this case we use a UCS series MCA by Spectrum Techniques. My favorite excitation source has been Kr-85 but most will work, even Sr-90 or Cs-137.

      Exciting the elements has never been a problem. Measuring the resultant K shell X-Rays has been a challenge however, due to the rather low energy range, 2 to 75 keV of elements between aluminum and lead.

      Lead (Pb) has always been the easiest to identify and we have done so in lead crystal glass and many other objects using our amateur methods. Most scintillation probes will give a decent spectrum at 75 keV, even as low as 20-30 keV depending on the housing material. Tin / Lead solder gives a nice double peak, one from tin, the other from lead. Peak amplitude (corrected for detector efficiency) is an indication of alloy percentages.

      Thick section crystals as we all know by now detect higher energy gammas and X-Rays rather well, and even respond well to low energies if the housing material allows them to get inside to the actual crystal. A disadvantage to using a thick crystal, even one with a thin beryllium window is of course the over response to higher energy stray radiation.

      A solution is of course to use a thin crystal with a thin window. Higher energies will pass right through while lower ones will be efficiently absorbed. Indeed this is the design of all LEG or Low Energy Gamma probes, such as the Ludlum 44-3 or the FIDLER. Such probes are automatically physically tuned to the 14-17 keV X-Ray energies of Plutonium, seeking which is their prime usage in industry and military. Almost always such probes are used in SCA / Windowed metering circuits, thereby further peaked electronically at 10-20 keV.
      Such a thin crystal, typically only 1 mm thick, is sensitive enough but does not provide the required response to give a really nice looking spectrum on an MCA.

      All available LEG's have been tried (NaI(Tl), CaF2, CsI etc.), all work but none do the job with sufficient precision using our amateur methods.

      Home Lab experiments over the years have indicated that a sodium iodide, thallium activated crystal of about 4mm thickness, inside a beryllium windowed housing SHOULD give ideal results under our circumstances.

      Scionix made one to my specs and the results are amazing. Scionix Model GEO 4mm! FWHM about 9.3% @ 122 keV Co-57.
       
      FIG.1> Setup showing excitation source, detector and sample arrangement:
       

      FIG 2.> Resulting energy spectrum obtained, specific for cobalt alone:
       
       
      FIG.3> Setup using small piece of indium solder as the sample under test (note in corner is the cadmium filter mentioned next)
       
       
      FIG. 4> Results of the scan for indium element, note position of unique peak at the energy of indium K-Lines:
       
       
      Historically, one of the nifty lab items we use for selective energy filtering is a metal disc, 2 3/4" diameter and 0.85mm thick, removed from an old Ludlum made uranium enrichment analyzer. The material it was made from was never specified, nor could Ludlum identify it after so many years. I assumed it was either tin or cadmium, since its qualities were similar to either. Weight/density and hardness testing leaned more towards cadmium. Since cadmium is a somewhat toxic metal and this filter never had a warning label or anything else to alert an operator, tin would be a more likely choice. Anyhow to be safe I labeled it as Cd and always handled it accordingly. Still it has bugged me not knowing for sure.

      Today I know for sure. It is cadmium, nearly pure the best we can tell. We have metallographic samples of 0.9999 pure elements, one of which is cadmium. By shining the rays and particles from the Kr-85 source onto the various metals, a signature is developed using the Scionix probe into an MCA, for each element. The filter- "Device Under Test" or DUT matches perfectly the 0.9999 Cd metal sample.
       
      FIG.5> Results of .999 pure cadmium element test showing unique energy lines of cadmium:
       
       


      FIG.6> Results of unknown metal filter test, peaks coincide exactly with pure cadmium element:
       
       
       
       
      FIG. 7> Setup using old US dime coin to test for silver:
       
       
      FIG. 8> Results of a test of pure tin element (Sn) showing unique energy lines for Sn:
       
      Observe the specially made periodic charts of the elements we had made (Thanks AtomicDave!). Instead of protons and neutrons these charts give the characteristic XRF energies of the two K shell and two L shell electrons.
      Our method of beta excitement works mostly on the K shell, so the energy read with our probe is roughly an average of both the K shell energies together. Each element can therefore be identified very easily by comparing it to a known sample. Furthermore, some alloys, like Bronze can be identified by a double peak, one from copper the other from tin. Two elements that are right next to one another on the chart will merge usually, but either in pure form can easily be identified.
      XRF CHARTS LINKS:

      8.5" X 11" Size: http://www.qsl.net/k/k0ff/XRF%20Periodic%20Table/PeriodicTable3.pdf

      11" X 17" Size: http://www.qsl.net/k/k0ff/XRF%20Periodic%20Table/PeriodicTable11x17.pdf

      24" X 26" Wall Chart Size: http://www.qsl.net/k/k0ff/XRF%20Periodic%20Table/PeriodicTable24x36.pdf


      Using betas will identify only the surface elements, not what is beneath. An example is a modern USA penny. Once made from mostly all copper, today they are made of zinc, with a very thin copper flashing. If you remove the copper from one side and analyze each side separately, the copper and zinc are easily discerned even though they reside each other on the chart. If one were using X-Rays to excite, they would penetrate further into the clad coin, and would give a merged spectrum.

      This series of experiments were designed as proof-of-concept, not as a superior invention of a working apparatus. No one would build such a machine when better ones are available, (at a somewhat elevated cost today of about $30,000 USD! see Niton XRF)
       
      More advanced experiments following this same basic technique will be shown soon, with which we have identified all the elements from Calcium through Bismuth.

      George Dowell