TEM-EDS microanalysis: Comparison between different electron sources, accelerating voltages and detection systems

Two TEM-EDS quantification methods based on standards of known compositions, namely the Cliff and Lorimer approximation and the absorption correction method based on electroneutrality are employed and the results obtained with three different TEMs and EDS systems, compared. The three TEM instruments differ in source type (field emission vs. thermionic), accelerating voltage (200 vs. 300 kV) and EDS system type (4 in-column silicon drift detector (SDD) vs. single SDD). We found that EDS calibration appears to be "strictly instrument specific", i.e., no universally valid k-factors can exist, but only k-factor sets for a specific combination of microscope and EDS system. As expected, 4-in column SDD systems, because of their larger sensitive areas compared to classical single SDD, are more efficient in data collection and, therefore, have lower detection limits. However, other sources of error may influence the final output, sometimes subverting the expectations. EDS analyses performed with FEG-TEMs exhibit lower radiation-induced migration of weakly bounded elements than TEMs equipped with a conventional source and lower beam current. This result may be explained by the smaller spot size used with the conventional TEM that in total led to a higher electron dose per sample atom. In addition, this work confirms that the absorption correction method is to be preferred whenever dealing with thick and/or dense samples, whereas the Cliff and Lorimer approximation, because simpler and faster, in all the other cases. Finally, we renew the necessity to determine two distinct k_O/Si factors, one for lighter and one for denser compounds.