The Mean Inner Potential of Hematite α-Fe2O3 Across the Morin Transition.

We measure the mean inner potential (MIP) of hematite, α-Fe2O3, using electron holography and transmission electron microscopy. Since the MIP is sensitive to valence electrons, we propose its use as a chemical bonding parameter for solids. Hematite can test the sensitivity of the MIP as a bonding parameter because of the Morin magnetic phase transition. Across this transition temperature, no change in the corundum crystal structure can be distinguished, while a change in hybridized Fe-3d and O-2p states was reported, affecting ionic bonding. For a given crystallographic phase, the change in the MIP with temperature is expected to be minor due to thermal expansion. Indeed, we measure the temperature dependence in corundum α-Al2O3(112¯0) between 95 and 295 K showing a constant MIP value of ∼16.8 V within the measurement accuracy of 0.45 V. Thus, our objectives are as follows: measure the MIP of hematite as a function of temperature and examine the sensitivity of the MIP as a bonding parameter for crystals. Measured MIPs of α-Fe2O3(112¯0) above the Morin transition are equal, 17.85 ± 0.50 V, 17.93 ± 0.50 V, at 295 K, 230 K, respectively. Below the Morin transition, at 95 K, a significant reduction of ∼1.3 V is measured to 16.56 ± 0.46 V. We show that this reduction follows charge redistribution resulting in increased ionic bonding.

Evaluating direct detection detectors for short-range order characterization of amorphous materials by electron scattering.

The introduction of direct electron detectors (DEDs) to transmission electron microscopy has set off the 'resolution revolution', especially for cryoTEM low-dose imaging of soft matter. In comparison to traditional indirect electron detectors such as Charged-Coupled Devices (CCD), DEDs show an improved modulation transfer function (MTF) and detective quantum efficiency (DQE) across all spatial frequencies, as well as faster frame rates which enable single electron counting. The benefits of such characteristics for imaging, spectroscopy and electron holography have been demonstrated previously. However, studies are lacking on the application of DEDs for localized characterization of short- to medium- range-order (SRO, MRO) in amorphous materials using electron scattering. Therefore, we evaluate the performance of a Monolithic Active Pixel Sensor DED for the characterization of SRO and MRO in nanoscale volumes of amorphous materials, using SiO2 and Ta2O5 thin films as test cases. The performance of the detector is compared systematically to electron scattering measurements recorded on an indirect detector (CCD) using 200 keV electrons and electron doses starting at approximately 500e-Å2 . In addition, the effects of sample cooling and energy-filtering on the measured SRO of the oxides were investigated. We demonstrate that the performance of the DED resulted in improved SRO characterization in comparison to that obtained from the CCD measurements. The DED enabled to achieve a larger measured maximal scattering vector, ∼16.51Å compared to ∼151Å, for the CCD. Furthermore, an improved signal-to-noise ratio of approximately two-fold was observed across all spatial frequencies for both 200 keV and 80 keV electrons. These improvements are shown to result from the superior DQE of the DED. Consequently, the DED measurements enabled to determine the coordination numbers of atomic bonds more accurately. We expect that further benefits of the DED for S/MRO characterization will be highlighted for ultra- sensitive materials that cannot withstand electron doses above several e-Å2 .

Elastic and inelastic mean free paths for scattering of fast electrons in thin-film oxides.

Quantitative transmission electron microscopy (TEM) often requires accurate knowledge of sample thickness for determining defect density, structure factors, sample dimensions, electron beam and X-ray photons signal broadening. The most common thickness measurement is by Electron Energy Loss Spectroscopy which can be applied effectively to crystalline and amorphous materials. The drawback is that sample thickness is measured in units of Inelastic Mean Free Path (MFP) which depends on the material, the electron energy and the collection angle of the spectrometer. Furthermore, the Elastic MFP is an essential parameter for selecting optimal sample thickness to reduce dynamical scatterings, such as for short-range-order characterization of amorphous materials. Finally, the Inelastic to Elastic MFP ratio can predict the dominant mechanism for radiation damage due to the electron beam. We implement a fast and precise method for the extraction of inelastic and elastic MFP values in technologically important oxide thin films. The method relies on the crystalline Si substrate for calibration. The Inelastic MFP of Si was measured as a function of collection semi-angle (ß) by combining Energy-Filtered TEM thickness maps followed by perpendicular cross-sectioning of the sample by Focused-Ion-Beam. For example, we measured a total Inelastic MFP (ß∼157 mrad) in Si of 145 ± 10 nm for 200 keV electrons. The MFP of the thin oxide films is determined by their ratio at their interface with Si or SiO2. The validity of this method was verified by direct TEM observation of cross-to-cross sectioning of TEM samples. The high precision of this method was enabled mainly by implementing a wedge preparation technique, which provides large sampling areas with uniform thickness. We measured the Elastic and Inelastic Mean Free Paths for 200 keV and 80 keV electrons as a function of collection angle for: SiO2 (Thermal, CVD), low-κ SiOCH, Al2O3, TiO2, ZnO, Ta2O5 and HfO2. The measured MFP values were compared to calculations based on models of Wenzel, Malis and Iakoubovskii. These models deviate from measurements by up to 30%, especially for 80 keV electrons. Hence, we propose functional relations for the Elastic MFP and Inelastic MFP in oxides with respect to the mass density and effective atomic number, which reduce deviations by a factor of 2-3. In addition, the effects of sample cooling on the measurements and sample stability are examined.