We use an FEI Tecnai F20 operated at 200 kV, with an FEG electron source and a 2 k  ×  2 k CCD camera behind an energy filter. Additionally, the electron microscope is equipped with a goniometer that is suitable for tomography, allowing smooth rotation of the stage to angles up to 70°.

Data collection is best performed on a medium- or high-voltage electron microscope operating at an acceleration voltage of 200 or 300 kV. With these higher electron voltages thicker samples (up to ∼400 nm) can be used. For whole cell imaging with transmission the sample thickness is one of the most stringent limitations that require a sufficiently high acceleration voltage. As an electron source it is advantageous to use an FEG. An FEG has the advantages over other sources such as LaB₆ and Tungsten in that the electron beam is more coherent and more suitable for imaging thicker samples at a relatively large defocus, which is necessary for the generation of phase contrast.

Zero loss imaging of the electrons using an energy filter is very advantageous (see Note 2). By using an energy filter in the zero-loss imaging mode (e.g., with an energy slit of 20 eV) the inelastically scattered electrons do not contribute to the final image. Consequently, the signal-to-noise level of images acquired from relatively thick specimen is significantly better compared to unfiltered images. Zero-loss imaging becomes increasingly important at high specimen tilts, since the specimen becomes thicker to the electrons as they traverse it to form a projection image.

Imaging should be performed with a digital image detector, which is indispensible for automated image acquisition. In cryo electron tomography, the number of electrons is in general very low due to the necessary dose fractionation: the acceptable total electron dose has to be distributed over as many images as possible. Consequently, the detector should be highly sensitive. To image the specimen in a large field of view with reasonable resolution, the imaging detector size should be at least 2 k  ×  2 k pixels in size but preferable larger (4 k  ×  4 k or even 8 k  ×  8 k). This allows for either a large field of view or binning of pixels to allow increased effective sensitivity. Most cameras that are currently available for electron microscopes are cooled slow-scan CCD cameras that record light that is generated by the interactions of the electrons with a scintillator positioned on top of the detector. Currently, alternative (direct) electron detectors with superior characteristics in terms of image detection quantum efficiency (DQE) have become commercially available.

Eucentric height. After insertion of the cryo holder into the microscope, it must be set to the eucentric height. This means that the height of the holder in the goniometer must match the rotation axis of the goniometer. Many microscope alignments are related to this height as it defines the objective lens focus plane. When the stage is not at eucentric height upon applying specimen tilt, the sample shifts orthogonally to the tilt axis. The eucentric height can be set manually or automatically. For the automatic version to function robustly, it is advisable to first roughly set the height manually. A rough manual alignment of the eucentric height can be performed by centering a recognizable feature at low magnification, tilting the stage and then recentering the feature using the z-height controls. On our Tecnai the automatic eucentric height function is available in the tomography software suite.

Centering of apertures. The condenser and objective apertures have to be aligned properly to the optical axis to prevent beam movements during the tomographic data acquisition. A quick way to check if a condenser aperture is aligned is to center a focused beam and observe whether the beam stays centered during spreading of the intensity. If the area does not stay centered, the condenser aperture is not well aligned. The condenser aperture can be centered by moving the aperture using the two knobs on the aperture mounting. Repeat adjustments until the illuminated area stays centered and only changes size. The objective aperture can be centered by switching the microscope to diffraction mode (where, at the appropriate camera length, the objective aperture is visible as a circular shape) and to move the aperture such that the central diffracted beam is centered within the circular shape.

CCD calibration files. Most of the calibrations and alignments in the tomography software make use of image shift measurements using cross-correlations. Therefore, the quality of the CCD images is an important factor. Unfortunately, in practice, CCD images can suffer from a number of image deteriorating artifacts that have to be corrected for. In most cases, two types of calibration files, often referred to as gain and dark reference images, are needed for the calibration of CCD detectors. Gain references are recorded to correct for the inhomogeneous sensitivity (i.e., gain) of the individual pixels of the detector arising from various sources, e.g., hardware variations of the CCD chip, variations of the scintillator, dust, scratches, and more (13). Dark references are recorded to correct for intensity variations related to the readout and electronics of the data. Using the Gatan CCD detectors on our systems, the dark references are recorded automatically prior to the recording of the illuminated image. Therefore, only the gain references have to be recorded as a separate step. Gain references are recorded for an even (homogeneous) illumination of the CCD detector. One way to verify whether the gain reference files are fine is to record an image at even illumination (preferably at the same image intensity as is anticipated to record the images during the tilt series) and to compute an auto-cross-correlation of the image. The auto-cross-correlation image should show a sharp intensity peak in the center of the image, while the rest should be noisy without any visible pattern. If a pattern of peaks, or bright lines, appears a new gain reference image should be recorded.

Beam tilt pivot points. This alignment makes sure that, upon tilting the beam, the illuminated area on the specimen stays centered. When the pivot point of the beam tilt is not correct (i.e., does not correspond to the specimen plane), this can be observed by the movement of the illuminated area when the illumination is tilted. Beam tilting is used as a way to measure the defocus during autofocusing since the amount of shift is directly related to the amount of defocus. The beam tilt pivot points can be corrected in the direct alignment steps of the microscope. Note that for the alignment and focusing it is important that the sample is at eucentric height. When this is not the case, the illuminated area will shift upon beam tilt. In particular, this can induce problems when the illuminated area is relatively small and is only slightly smaller than the CCD.