Improving the accuracy of temperature measurement on TEM samples using plasmon energy expansion thermometry (PEET): Addressing sample thickness effects.

Advances in analytical scanning transmission electron microscopy (STEM) and in microelectronic mechanical systems (MEMS) based microheaters have enabled in-situ materials' characterization at the nanometer scale at elevated temperature. In addition to resolving the structural information at elevated temperatures, detailed knowledge of the local temperature distribution inside the sample is essential to reveal thermally induced phenomena and processes. Here, we investigate the accuracy of plasmon energy expansion thermometry (PEET) as a method to map the local temperature in a tungsten (W) lamella in a range between room temperature and 700 °C. In particular, we address the influence of sample thickness in the range of a typical electron-transparent TEM sample (from 30 nm to 70 nm) on the temperature-dependent plasmon energy. The shift in plasmon energy, used to determine the local sample temperature, is not only temperature-dependent, but in case of W also seems thickness-dependent in sample thicknesses below approximately 60 nm. It is believed that the underlying reason is the high susceptibility of the regions with thinner sample thickness to strain from residual load induced during FIB deposition, together with increased thermal expansion in these areas due to their higher surface-to-volume ratio. The results highlight the importance of considering sample thickness (and especially thickness variations) when analyzing the local bulk plasmon energy for temperature measurement using PEET. However, in case of W, an increasing beam broadening (FWHM) of the bulk plasmon peak with decreasing sample thickness can be used to improve the accuracy of PEET in TEM lamellae with varying sample thickness.