Electron Probe Microanalysis EPMAKey pointsSpectrometers- OrientationFocussing GeometryTemperature Dependence-1Temperature Dependence-2Temperature Dependence-3Slide 8Spectral ResolutionSpectrometer EfficiencySpectrum = Characteristic + continuumProbe for Windows Peak and Background displaysBackground modelCurved background-1Curved background-2Congested backgroundsBackground OffsetsBackgrounds andAbsorption Edges-the EDS perspectiveBackgrounds and Absorption Edges - the WDS perspectiveAr Absorption EdgeHoles in the Background-1Holes in the Background-2On Peak InterferencesOn Peak Interference-SlightInterference CorrectionPeak Centering-1Peak Centering-2Peak Centering-3PowerPoint PresentationX-ray Peaks = PoissonianSatellitesPeak Shifts-1Peak Shifts-2Slide 34Peak Shifts-4Identifying Lines in SpectraGuidelines for WDS peak ID-1Guidelines for WDS peak ID-2Virtual WDSVirtual WDS - Ti on V interferenceElectron Probe MicroanalysisEPMA WDS II:Spectrometers and SpectraUW- Madison Geology 777Revised 10-5-08Key points• X-rays are dispersed by a crystal with only one wavelength (n) reflected (=diffracted), with that one wavelength (n) passed to the detector.This is a monochrometer. Recall that n can be >1, so other elements can cause interference if their wavelengths are at an integral fraction of the desired wavelength.• The detector is a gas-filled (sealed or flow-through) tube where gas is ionized by X-rays, yielding a massive multiplication factor (‘proportional counter’) • X-ray focusing assumes geometry known as the Rowland Circle• Key features of WDS are high spectral resolution and low detection limitsUW- Madison Geology 777Spectrometers- Orientation The typical electron microprobe has vertically mounted spectrometers (a), i.e. the Rowland circle is vertical. This orientation permits up to 5 spectrometers to be mounted around the column, as well as having room for an EDS detector. In this orientation, X-ray intensities are susceptible to small differences in the Z position of the sample (= X-ray defocusing, not good). We use polished planar samples, and focus the sample Z with reflected light.Reed, 1993, Fig. 6.9, p. 70. There are some applications (e.g., industry) where the probe is utilized in production, and samples are rough surfaces. For this specialized situation, the spectrometer can be inclined (b), which then permits X-ray diffraction from a range of heights, all now on the inclined Rowland circle. However, only 2 inclined spectrometers will then fit around the column.UW- Madison Geology 777Focussing GeometryThe point source of X-rays in the electron microprobe is not optimally diffracted by a flat crystal, where only a small region is “in focus” for the one wavelength of interest. X-ray intensities are improved by adding curvature to the crystals:• Johann geometry: the crystal is bent/curved to a radius 2r (=diameter of Rowland circle); • Johansson geometry: the crystal is curved to 2r, and the surface is ground to r. This is a difficult process, so most crystals have Johann geometry.Reed, 1993, p. 67.UW- Madison Geology 777Temperature Dependence-1We attempt to keep a constant temperature in the probe lab, around 68°F (20°C). This is also the temperature of the chilled water that circulates both through the electronic cabinet as well as the column (and outer jacket of diffusion pump).Several aspects of the instrument are sensitive to temperature changes – the spectrometer crystals, the detector P10 gas.• P10 gas pressure must be constant, as this is critical to having reproducibility in counting rates between the standards and the unknowns, both over short time spans as well as longer (e.g. 24 hour) durations. (Loss of air conditioning on very hot summer days precludes probing.)UW- Madison Geology 777Temperature Dependence-2• Spectrometer crystals have a linear expansion coefficient, and expansion will change the 2d. The effect upon PET is 4x worse than upon LIF, and increases rapidly with increasing sin .For LIF 200, the linear expansion coefficient is 3.4 x 10-5. At  =45° (sin  = .707107), one degree of temperature change causes the sin  to change to .707131, about 2.4 sin  units, not a lot -- but for say 5° C that is 12 sin  units--a major shift.For PET the effect is more serious, as the coefficient is 1.2 x 10 -4.*UW- Madison Geology 777* Typo corrected 9/29/08Temperature Dependence-3There is also a potential thermal effect, in that the motors on the stage and associated electronics put out heat, and can cause the stage height to slowly drift upward over hours. So if points are stored when the stage is cold, the Z will could be 10-15 microns out of focus 8 hours later -- thus autofocus is recommended for overnight automated runs.As mentioned in the first slide, cooling water is constantly circulating. If there is a failure of either the second floor “red water” heat exchanger, or of the master heat exchanger in the sub-basement, the chilled water no longer is chilled and problems can develop. The probe’s electronic cabinet also needs to be kept cool and there are 3 ‘squirrel cage’ fans that draw hot air past the chilled water tubing. If one or more fails, one or more circuit boards overheat and problems develop with operation of the probe.UW- Madison Geology 777Temperature Dependence-3UW- Madison Geology 777Temperature sensitivities of crystal analyzers; correlations between change in 2 value/°C. (X axis is in units of 2, commonly used in XRF; Y axis is the peak shift due to thermal expansion of the crystal) From Jenkins and DeVries, 1967, Fig 2.4 (p. 38)This figure shows how PET (top curve) is much more sensitive to temperature changes than LIF200. No curve for TAP is shown, but it perhaps is similar to ADP.Si Ka on PET is at a sin of .8, which is an angle of 54° so 2 is 108°.Si Ka on TAP is at a sin of .277 or an angle of 16° so 2 is 32°. This is the reason that it is generally preferable to use TAP for Si Ka measurement in EPMA.PETLIFSpectral ResolutionWDS provides roughly an order of magnitude higher spectral resolution (sharper peaks) compared with EDS. Plotted here are resolutions of the 3 commonly used crystals, with the x-axis being the characteristic energy of detectable elements.Note that for elements that are detectable by two spectrometers (e.g., Y L by TAP and PET, V K by PET and LIF), one of the two crystals will have superior