New methods for electron tomography

Electron tomography is a method for obtaining three-dimensional structural information from electron micrographs. It can be applied to a wide range of samples that can be prepared for transmission electron microscopy (TEM)may they be of biological origin like e.g. cryo or thin plastic sections of cells and tissue or of material science origin like solid catalysts and electronic devices. Electron tomography data acquisition can be regarded as a kind of super-stereology. Instead of images taken from only two different vantage points, a tomographic data set might consist of more than 151 images taken over an angular range of 150?. A strong advantage of electron tomography is that it does not depend on averaging over unit cells or particles or on the assumption and exploitation of symmetry in samples, as is the case for methods like angular reconstitution, nuclear magnetic resonance spectroscopy, electron and X-ray crystallography. Electron tomography can thus be applied to truly unique structures. In this publication we discuss three new methods, that make electron tomography data acquisition faster, more accurate and opens it for novel acquisition modes. In chapter 2 we describe a method for automated electron tomography data collection that reduces acquisition time by a factor of five, enabling data collection in less than an hour. The method includes a pre-calibration stepmeasurement and correction of image shift and defocus change at low magnification to detect displacements that might be as large as several µmbefore image acquisition. This step avoids switching back and forth between high and low magnification during image acquisition, which is a very time consuming step regarding the electron magnetic lens stabilizations needed. The method is based on the fact that these dislocations occur due to a displacement of the feature of interest from the eucentric height, a displacement of the optical axis from the tilt axis and some movements intrinsic to the sample holder/stage combination and are thus almost predictable. In chapter 3 we discuss a method for the correction of autofocusing errors due to specimen tilt, as this would disable accurate defocus prediction in e.g. pre-calibration tomography. Defocus determination using the beam-tilt method can be hampered for low magnifications and high-tilt angles due to a defocus ramp in the images. The method for the correct cross-correlation of two images of a tilted sample acquired under tilted-beam conditions is a modification of the cosine stretch used in the alignment of images acquired under different tilt angles. Electron tomography is not restricted to the use of transmission electron microscopy. Image acquisition in high angular annular dark-field scanning transmission electron microscopy (HAADF-STEM) mode makes the method readily available for e.g. crystalline samples, which cannot be acquired in TEM mode as they show angle dependent diffraction contrast. As it was shown that nm-sized inclusions in crystalline material can be detected by this method we have applied the approach to the detection of ultrasmall immuno-gold labels absorbed to heavy-metal stained plastic sections of biological material. Our initial experiments discussed in chapter 4 provide good evidence that HAADF-STEM electron tomography could be a useful tool for the accurate three-dimensional immuno-localization of proteins