Nanometer-scale measurements of Fe (super 3+) /Sigma Fe by electron energy-loss spectroscopy; a cautionary note

The effects of electron-beam damage on the Fe (super 3+) /Sigma Fe (total iron) ratio were measured by electron energy-loss spectroscopy (EELS) with a transmission electron microscope (TEM). Spectra were acquired from crushed and ion-beam-thinned cronstedtite. For fluences below 1X10 (super 4) e/Aa (super 2) , the Fe (super 3+) /Sigma Fe values from crushed grains range between 0.43 and 0.49, consistent with undamaged material. These measurements were acquired from flakes 180 to 1000 Aa thick. With increase in fluence, samples <400 Aa thick become damaged and exhibit Fe (super 3+) /Sigma Fe values >0.5. The critical fluence for radiation damage by 100 kV electrons as defined by Fe (super 3+) /Sigma Fe <0.5 for cronstedtite at 300 K, is 1X10 (super 4) e/Aa (super 2) . The absorbed dose to the specimen during acquisition of a typical EELS spectrum is large, with values around 2.2X10 (super 10) Gy (J/kg), equivalent to the deposition of 620 eV/Aa (super 3) . Cooling to liquid N (sub 2) temperature did not significantly slow the damage process. Ion-beam thinning produces an amorphous layer on crystal surfaces. Spectra from the thinnest regions, which are amorphous, exhibit Fe (super 3+) /Sigma Fe <0.7. With increase in sample thickness, the Fe (super 3+) /Sigma Fe values decrease to a minimum, consistent with data from the undamaged material. The increase of Fe (super 3+) /Sigma Fe with respect to electron-beam irradiation is likely caused by loss of H. At low fluences, the loss of H is negligible, thus allowing consistent Fe (super 3+) /Sigma Fe values to be measured. The cronstedtite study illustrates the care required when using EELS to measure Fe (super 3+) /Sigma Fe values. Similar damage effects occur for a range of high-valence and mixed-oxidation state metals in minerals. EELS is the only spectroscopic method that can be used routinely to determine mixed-valence ratios at the nanometer scale, but care is required when measuring these data. Consideration needs to be given to the incident beam current, fluence, fluence rate, and sample thickness.