On the applicability of electron diffraction to precisely measure temperature in TEM

Abstract

With the overwhelming success of in situ electron microscopy, fueled by the recent advances in instrumentation, an everlasting question in electron microscopy comes into focus again: the precise determination of the temperature of the sample, especially in in situ heating experiments [1]. So far temperature is determined by sensors (thermocouples, resistance thermometers) placed close to the sample. In this work we elaborate on the applicability of parallel beam electron diffraction in TEM to precisely measure temperature directly on the sample and on a local scale. Well defined metallic particles are applied on the sample surface using a dewetting process after thin film deposition. SAED is used to determine the lattice spacing from ring patterns averaging over a large number of particles. The determined lattice spacing can be translated into temperature using reference data on thermal expansion. To test the performance of this approach, a MEMS‐based sample heating system from DENSsolutions is used inside of a double corrected FEI Titan Themis³ 300 . Various metals (Ag, Au, Pt) are used to cover different temperature ranges. Fig. 1 shows image and diffraction pattern of a typical setup. To allow for the highest precision, a proper alignment of the microscope is crucial. Therefore a method to assure perfect beam parallelity has been developed in this project, based on the change in diffraction pattern upon changing the sample position relative to the eucentric height. The accuracy of the evaluation based on peak fitting to the azimuthally averaged radial profile critically depends on the determination of the center and correction of astigmatism in the ring pattern. An algorithm has been employed allowing for correction of the astigmatism to the fourth order in the ring pattern (exemplarily shown in Fig. 2). The error due to the evaluation method can be estimated from the results of the evaluation of 100 subsequent diffraction patterns at a constant temperature shown in Fig. 3. The position of the (220) ring of Pt is determined to be 7.27183 ± 0.00023 nm −1 (corresponding to a temperature accuracy of approximately ± 3.5 K). The temperature accuracy can be tailored to the particular application by selecting a different metal. Low melting metals show a higher thermal expansion increasing the sensitivity of the temperature measurement, but, on the other hand, limiting the maximum usage temperature because of melting or sublimation. Fig. 4 demonstrates the applicability of the presented approach. Ag particles have been used to measure the temperature during a dynamic in situ experiment, showing a very good agreement with the temperature reading from the holder.