Computed tomography of cryogenic biological specimens based on X-ray microscopic images

Soft X-ray microscopy employs the photoelectric absorption contrast between water and protein in the 2.34-4.38 nm wavelength region to visualize protein structures down to 30 nm size without any staining methods. Due to the large depth of focus of the Fresnel zone plates used as X-ray objectives, computed tomography based on the X-ray microscopic images can be used to reconstruct the local linear absorption coefficient inside the three-dimensional specimen volume. High-resolution X-ray images require a high specimen radiation dose, and a series of images taken at different viewing angles is needed for computed tomography. Therefore, cryo microscopy is necessary to preserve the structural integrity of hydrated biological specimens during image acquisition. The cryo transmission X-ray microscope at the electron storage ring BESSY I (Berlin) was used to obtain a tilt series of images of the frozen-hydrated green alga Chlamydomonas reinhardtii. The living specimens were inserted into borosilicate glass capillaries and, in this first experiment, rapidly cooled by plunging into liquid nitrogen. The capillary specimen holders allow image acquisition over the full angular range of 180 degrees. The reconstruction shows for the first time details down to 60 nm size inside a frozen-hydrated biological specimen and conveys a clear impression of the internal structures. This technique is expected to be applicable to a wide range of biological specimens, such as the cell nucleus. It offers the possibility of imaging the three-dimensional structure of hydrated biological specimens close to their natural living state.     

Quantification and thickness correction of EFTEM phosphorus maps

We describe a method for correcting plural inelastic scattering effects in elemental maps that are acquired in the energy filtering transmission electron microscope (EFTEM) using just two energy windows, one above and one below a core edge in the electron energy loss spectrum (EELS). The technique is demonstrated for mapping low concentrations of phosphorus in biological samples. First, the single-scattering EELS distributions are obtained from specimens of pure carbon and plastic embedding material. Then, spectra are calculated for different specimen thicknesses t, expressed in units of the inelastic mean free path lambda. In this way, standard curves are generated for the ratio k0 of post-edge to pre-edge intensities at the phosphorus L2,3 excitation energy, as a function of relative specimen thickness t/lambda. Thickness effects in a two-window phosphorus map are corrected by successive acquisition of zero-loss and unfiltered images, from which it is possible to determine a t/lambda image and hence a background k0-ratio image. Knowledge of the thickness-dependent k0-ratio at each pixel thus enables a more accurate determination of the phosphorus distribution in the specimen. Systematic and statistical errors are calculated as a function of specimen thickness, and elemental maps are quantified in terms of the number of phosphorus atoms per pixel. Further analysis of the k0-curve shows that the EFTEM can be used to obtain reliable two-window phosphorus maps from specimens that are considerably thicker than previously possible.     

Quantitative electron holographic tomography for the 3D characterisation of semiconductor device structures

Electron tomography and electron holography experiments have been combined to investigate the 3D electrostatic potential distribution in semiconductor devices. The experimental procedure for the acquisition and data reconstruction of holographic tilt series of silicon p-n junction specimens is described. A quantitative analysis of the experimental results from specimens of two different thicknesses is presented, revealing the 3D electrostatic potential variations arising from the presence of surfaces and damage generated by focused ion beam (FIB) sample preparation. Close to bulk-like properties are measured in the centre of the tomographic reconstruction of the specimen, revealing higher electrically active dopant concentrations compared to the measurements obtained at the specimen surfaces. A comparison of the experimental results from the different thickness specimens has revealed a 'critical' thickness for this specimen preparation method of 350nm that is required for this device structure to retain 'bulk'-like properties in the centre of the membrane.     

Elemental imaging and resolution in energy-filtered conventional electron microscopy

Energy-filtered transmission electron microscopy (EFEM) was used to image the distributions of uranium and carbon in uranyl acetate stained catalase crystals. The spatial resolution obtained from inelastic C K-edge and U O4,5-edge images, determined from the highest-order reflection in the computed diffraction pattern, was 3.4 nm for both carbon and uranium. The resolution limit imposed by the delocalization of inelastic scattering was estimated from cross-section measurements to be 0.6 nm for U and 0.2 nm for C. Considering both delocalization and the effects of microscope aberrations, for an objective lens chromatic aberration coefficient of 2.8 mm and 10 eV energy window, the calculated resolutions are 2.0 nm for C and 1.2 nm for U. The effects of plural inelastic and elastic-inelastic scattering were sufficiently large to show crystalline structure in unprocessed pre-edge inelastic images. Previously suggested methods for eliminating these artifacts were applied to obtain the compositional information in the catalase EFEM images.     

Observation of three-dimensional elemental distributions of a Si device using a 360 degrees -tilt FIB and the cold field-emission STEM system

A technique for preparation of a pillar-shaped specimen and its multidirectional observation using a combination of a scanning transmission electron microscope (STEM) and a focused ion beam (FIB) instrument has been developed. The system employs an FIB/STEM compatible holder with a specially designed tilt mechanism, which allows the specimen to be tilted through 360 degrees [T. Yaguchi, M. Konno, T. Kamino, T. Hashimoto, T. Ohnishi, K. Umemura, K. Asayama, Microsc. Microanal. 9 (Suppl. 2) (2003) 118; T. Yaguchi, M. Konno, T. Kamino, T. Hashimoto, T. Ohnishi, M. Watanabe, Microsc. Microanal. 10 (Suppl. 2) (2004) 1030]. This technique was applied to obtain the three-dimensional (3D) elemental distributions around a contact plug of a Si device used in a 90-nm technology. A specimen containing only one contact plug was prepared in the shape of a pillar with a diameter of 200nm and a length of 5mum. Elemental maps were obtained from the pillar specimen using a 200-kV cold-field emission gun (FEG) STEM model HD-2300C equipped with the EDAX genesis X-ray energy-dispersive spectrometry (XEDS) system through a spectrum imaging technique. In this study, elemental distributions of minor elements with weak signals were enhanced by applying principal component analysis (PCA), which is a superior technique to extract weak signals from a large dataset. The distributions of elements, especially the metallization component Ti and minor dopant As in this particular device, were successfully extracted by PCA. Finally, the 3D elemental distributions around the contact plug could be visualized by reconstruction from the tilt series of maps.     

Towards single atom analysis of biological structures.

Mapping single atoms in biological structures is now becoming within the reach of analytical electron microscopy. Electron energy-loss spectroscopy (EELS) in the field-emission scanning transmission electron microscope (STEM) provides a particularly high sensitivity for detecting the biologically important element, phosphorus. Imaging can be performed at low dose with dark-field STEM prior to analysis at high dose, so that structures of macromolecular assemblies can be correlated with the numbers of specific atoms that they contain. Measurements confirm theoretical predictions that single atom detection requires a nanometer-sized probe. Although phosphorus atoms may have moved several nanometers from their original positions by beam-induced structural degradation at the high required dose of approximately 10(9) e/nm2, damaged molecules are nevertheless stable enough to be analyzed at 1 or 2 nm resolution. Such analyses can only be achieved by means of spectrum-imaging with correction for specimen drift. Optimal strategies for mapping small numbers of phosphorus atoms have been investigated using well-characterized specimens of DNA plasmids and tobacco mosaic virus.     

Mechanism of image formation for thick biological specimens: exit wavefront reconstruction and electron energy-loss spectroscopic imaging

With increasing frequency, cellular organelles and nuclear structures are being investigated at high resolution using electron microscopic tomography of thick sections (0·3–1·0 μm). In order to reconstruct the structures in three dimensions accurately from the observed image intensities, it is essential to understand the relationship between the image intensity and the specimen mass density. The imaging of thick specimens is complicated by the large fraction of multiple scattering which gives rise to incoherent and partially coherent image components. Here we investigate the mechanism of image formation for thick biological specimens at 200 and 300 keV in order to resolve the coherent scattering component from the incoherent (multiple scattering) components. Two techniques were used: electron energy-loss spectroscopic imaging (ESI) and exit wavefront reconstruction using a through-focus series. Although it is commonly assumed that image formation of thick specimens is dominated by amplitude (absorption) contrast, we have found that for conventionally stained biological specimens phase contrast contributes significantly, and that at resolutions better than ～10 nm, superposed phase contrast dominates. It is shown that the decrease in coherent scattering with specimen thickness is directly related to the increase in multiple scattering. It is further shown that exit wavefront reconstruction can exclude the microscope aberrations as well as the multiple scattering component from the image formation. Since most of the inelastic scattering with these thick specimens is actually multiple inelastic scattering, it is demonstrated that exit wavefront reconstruction can act as a partial energy filter. By virtue of excluding the multiple scattering, the ‘restored’ images display enhanced contrast and resolution. These findings have direct implications for the three-dimensional reconstruction of thick biological specimens, where a simple direct relationship between image intensity and mass density was assumed, and the aberrations were left uncorrected.     

Towards automated electron holographic tomography for 3D mapping of electrostatic potentials

Electron-holographic tomography (EHT), that is, the combination of off-axis electron holography with electron tomography, was successfully applied for the quantitative 3D mapping of electrostatic potentials at the nanoscale. Here we present the first software package (THOMAS) for semi-automated acquisition of holographic tilt series, a prerequisite for efficient data collection. Using THOMAS, the acquisition time for a holographic tilt series, consisting of object and reference holograms, is reduced by a factor of five on average, compared to the previous, completely manual approaches. Moreover, the existing software packages for retrieving amplitude and phase information from electron holograms have been extended, now including a one-step procedure for holographic tilt series reconstruction. Furthermore, a modified SIRT algorithm (WSIRT) was implemented for the quantitative 3D reconstruction of the electrostatic potential from the aligned phase tilt series. Finally, the application of EHT to a polystyrene latex sphere test-specimen and a pn-doped Ge ‘needle’-shaped specimen are presented, illustrating the quantitative character of EHT. For both specimens the mean inner potential (MIP) values were accurately determined from the reconstructed 3D potential. For the Ge specimen, additionally the ‘built-in’ voltage across the pn junction of 0.5 V was obtained.     

Transmission electron microtomography without the "missing wedge" for quantitative structural analysis

A three-dimensional (3D) visualization and structural analysis of a rod-shaped specimen of a zirconia/polymer nanocomposite material were carried out by transmission electron microtomography (TEMT) with particular emphasis on complete rotation of the specimen (tilt angular range: +/-90 degrees ). In order to achieve such an ideal experimental condition for the TEMT, improvements in the specimen as well as the sample holder were made. A rod-shaped specimen was necessary in order to obtain a high transmission of the specimen upon tilting to large angles. The image resolution of the reconstructed tomogram was isotropic, in sharp contrast to the anisotropic image resolution of the conventional TEMT with a limited angular range (the "missing wedge" problem). A volume fraction of zirconia, phi, evaluated from the 3D reconstruction was in quantitative agreement with the known composition of the nanocomposite. A series of 3D reconstructions was made from the tilt series with complete rotation by limiting the maximum tilt angle, alpha, from which a couple of structural parameters, the volume fraction and surface area per unit volume, Sigma, of the zirconia, were evaluated as a function of alpha. It was confirmed from actual experimental data that both phi and Sigma slightly decreased with the increasing alpha and reached constant values at around alpha=80 degrees , suggesting that the specimen may have to be tilted to +/-80 degrees for truly quantitative measurements.     

A new method for the determination of the wave aberration function for high-resolution TEM.; 2. Measurement of the antisymmetric aberrations

A new method is presented for the determination of the antisymmetric coefficients of the wave aberration function from a tableau of tilted illumination images. The approach is based on measurements of the apparent defocus and two-fold astigmatism using a phase correlation function and phase contrast index calculated from a short focus series acquired at each tilt. This method is shown to be suitable for a wide range of specimens and is sufficiently accurate for exit plane wave restoration at 0.1 nm resolution. Experimental examples of this approach are provided and the method is compared to results obtained from measurements of conventional power spectra.     

Limitations of beam damage in electron spectroscopic tomography of embedded cells

Elemental mapping in the energy filtering transmission electron microscope (EFTEM) can be extended into three dimensions (3D) by acquiring a series of two‐dimensional (2D) core‐edge images from a specimen oriented over a range of tilt angles, and then reconstructing the volume using tomographic methods. EFTEM has been applied to imaging the distribution of biological molecules in 2D, e.g. nucleic acid and protein, in sections of plastic‐embedded cells, but no systematic study has been undertaken to assess the extent to which beam damage limits the available information in 3D. To address this question, 2D elemental maps of phosphorus and nitrogen were acquired from unstained sections of plastic‐embedded isolated mouse thymocytes. The variation in elemental composition, residual specimen mass and changes in the specimen morphology were measured as a function of electron dose. Whereas 40% of the total specimen mass was lost at doses above 10⁶ e^?/nm², no significant loss of phosphorus or nitrogen was observed for doses as high as 10⁸ e^?/nm². The oxygen content decreased from 25 ± 2 to 9 ± 2 atomic percent at an electron dose of 10⁴ e^?/nm², which accounted for a major component of the total mass loss. The specimen thickness decreased by 50% after a dose of 10⁸ e^?/nm², and a lateral shrinkage of 9.5 ± 2.0% occurred from 2 × 10⁴ to 10⁸ e^?/nm². At doses above 10⁷ e^?/nm², damage could be observed in the bright field as well in the core edge images, which is attributed to further loss of oxygen and carbon atoms. Despite these artefacts, electron tomograms obtained from high‐pressure frozen and freeze‐substituted sections of C. elegans showed that it is feasible to obtain useful 3D phosphorus and nitrogen maps, and thus to reveal quantitative information about the subcellular distributions of nucleic acids and proteins.     

Improving image quality by zero-loss energy filtering: quantitative assessment by means of image cross-correlation

Fourier ring correlation and root-mean-square contrast of pairs of images, taken under identical conditions, were used as criteria of image quality for comparing unfiltered with zero-loss energy-filtered imaging using a TEM equipped with a post-column energy filter. For three different specimens (amorphous carbon film, macromolecules in light negative stain, virus particles in deep negative stain) the dependence of these quantities on electron dose, specimen thickness and defocus was investigated. A model, based on simple assumptions, was used to describe quantitatively their dependence on electron dose and specimen thickness. It was found that energy filtering is most advantageous for low-dose imaging and small defocus values. The gain due to energy filtering strongly increases with specimen thickness, whereby the dependence is linear for light scattering elements. For thick specimens, the gain by energy filtering is more pronounced in the resolution range between 4 and 2 nm than for lower spatial frequencies.     

Comparison of intensity distributions in tomograms from BF TEM,ADF STEM,HAADF STEM,and calculated tilt series

The three-dimensional (3D) morphology of a nanometer-sized object can be obtained using electron tomography. Variations in composition or density of the object cause variations in the reconstructed intensity. When imaging homogeneous objects, variations in reconstructed intensity are caused by the imaging technique, imaging conditions, and reconstruction. In this paper, we describe data acquisition, image processing, and 3D reconstruction to obtain and compare tomograms of magnetite crystals from bright field (BF) transmission electron microscopy (TEM), annular dark-field (ADF) scanning transmission electron microscopy (STEM), and high-angle annular dark field (HAADF) STEM tilt series. We use histograms, which plot the number of volume elements (voxels) at a given intensity vs. the intensity, to measure and quantitatively compare intensity distributions among different tomograms. In combination with numerical simulations, we determine the influence of maximum tilt angle, tilt increment, contrast changes with tilt (diffraction contrast), and the signal-to-noise ratio (SNR) as well as the choice of the reconstruction approach (weighted backprojection (WB) and sequential iterative reconstruction technique (SIRT)) on the histogram. We conclude that because ADF and HAADF STEM techniques are less affected by diffraction, and because they have a higher SNR than BF TEM, they are better suited for tomography of nanometer-sized crystals.     

Comparison of intensity distributions in tomograms from BF TEM, ADF STEM, HAADF STEM, and calculated tilt series

The three-dimensional (3D) morphology of a nanometer-sized object can be obtained using electron tomography. Variations in composition or density of the object cause variations in the reconstructed intensity. When imaging homogeneous objects, variations in reconstructed intensity are caused by the imaging technique, imaging conditions, and reconstruction. In this paper, we describe data acquisition, image processing, and 3D reconstruction to obtain and compare tomograms of magnetite crystals from bright field (BF) transmission electron microscopy (TEM), annular dark-field (ADF) scanning transmission electron microscopy (STEM), and high-angle annular dark field (HAADF) STEM tilt series. We use histograms, which plot the number of volume elements (voxels) at a given intensity vs. the intensity, to measure and quantitatively compare intensity distributions among different tomograms. In combination with numerical simulations, we determine the influence of maximum tilt angle, tilt increment, contrast changes with tilt (diffraction contrast), and the signal-to-noise ratio (SNR) as well as the choice of the reconstruction approach (weighted backprojection (WB) and sequential iterative reconstruction technique (SIRT)) on the histogram. We conclude that because ADF and HAADF STEM techniques are less affected by diffraction, and because they have a higher SNR than BF TEM, they are better suited for tomography of nanometer-sized crystals.     

Exit wave reconstruction using a conventional transmission electron microscope

Exit wave reconstruction of a focus series of Ge in [110] using the PAMMAL algorithm was performed on a conventional electron microscope. The simulated images using the reconstructed object wave match very well with those obtained experimentally. Amplitudes from the complex wave function were extracted by means of local Fourier transformation. Crystal thickness and tilt were determined locally by quantitative comparison of the reconstructed amplitudes with amplitudes from multislice calculations. Detailed analysis yields the quasicoherent imaging approach used in the PAMMAL algorithm to produce the largest error in the analysis. For the Ge crystal specimen parameters were quantified to spatial frequencies of 5 nm¹. In the case of an object producing strong diffracted beams, the reconstruction may fail because the quasicoherent approximation will not describe correctly the nonlinear image formation.     

High-precision tilt stage for the high-voltage electron microscope

The high-voltage electron microscope is used to study thick samples (0.25 to several micrometers) to obtain three-dimensional information at ultrastructural resolution. Three-dimensional image reconstructions are often employed to extract, process and display this information. The sets of images used to form reconstructions must be recorded for precisely known specimen-beam orientations, especially if tomographic methods are employed. The design and operation of a precision (+/- 0.06 degrees) single-tilt stage to support this type of imaging is reported. All motions including two translations, height adjustment and tilting are accomplished via a single objective lens entry port. The specimen rod is supported on two rubber gaskets for vibration isolation, and motorized precision micrometers with encoder readouts for position monitoring drive the motions. The stage is stable to 0.6 nm for at least 16 s and is capable of tilt angles of +/- 70 degrees.     

Optimal strategies for imaging thick biological specimens: exit wavefront reconstruction and energy-filtered imaging

In transmission electron microscopy (TEM) of thick biological specimens, the relationship between the recorded image intensities and the projected specimen mass density is distorted by incoherent electron–specimen interactions and aberrations of the objective lens. It is highly desirable to develop a strategy for maximizing and extracting the coherent image component, thereby allowing the projected specimen mass density to be directly related to image intensities. For this purpose, we previously used exit wavefront reconstruction to understand the nature of image formation for thick biological specimens in conventional TEM. Because electron energy-loss filtered imaging allows the contributions of inelastically scattered electrons to be removed, it is potentially advantageous for imaging thick, biological samples. In this paper, exit wavefront reconstruction is used to quantitatively analyse the imaging properties of an energy-filtered microscope and to assess its utility for thick-section microscopy. We found that for imaging thick biological specimens (> 0.5 μm) at 200 keV, only elastically scattered electrons contribute to the coherent image component. Surprisingly little coherent transfer was seen when using energy-filtering at the most probable energy loss (in this case at the first plasmon energy-loss peak). Furthermore, the use of zero-loss filtering in combination with exit wavefront reconstruction is considerably more effective at removing the effects of multiple elastic and inelastic scattering and microscope objective lens aberrations than either technique by itself. Optimization of the zero-loss signal requires operation at intermediate to high primary voltages (> 200 keV). These results have important implications for the accurate recording of images of thick biological specimens as, for instance, in electron microscope tomography.     

Computation of contrasts in atomic resolution electron spectroscopic images of planar defects in crystalline specimens

The image obtained in a conventional transmission electron microscope contains contributions from elastically and from inelastically scattered electrons. The electron spectroscopic imaging mode of an energy-filtering transmission electron microscope allows us to separate these two different contributions by inserting an energy-selecting slit in the energy-dispersive plane of an imaging energy filter. Selecting a specific energy loss corresponding to the ionization of the inner shell of a particular element one can obtain information on the distribution of the element within the specimen. The contrast is then caused by inelastically scattered electrons. For crystalline specimens, however, the contrast will be influenced additionally by the elastic contrast. This elastic contrast arises from electron diffraction and increases with increasing crystal thickness. Therefore the intensity distribution in the image cannot directly be interpreted as an elemental map. For a reliable interpretation of contrast formation in elemental maps it is therefore necessary to compute theoretical energy-loss images for various crystal thicknesses and compare these images with the experimental images. As an example we discuss the influence of electron diffraction effects on energy-loss images of two crystals with planar defects. Linescans are computed for various thicknesses of these crystals. Our calculations are performed using first-order perturbation theory to describe the transitions between the Bloch-wave states of the incident electron. The computed linescans for various crystal thicknesses show clearly that the influence of the elastic contrast on an image increases when we investigate thicker specimens. Furthermore, the comparison between elastic and energy-loss images demonstrates the partial preservation of the elastic contrast as a function of thickness. We find that for specimens thicker than about one third of the extinction length (here approximately 80-100 A) it is impossible to interpret an energy-loss image directly as elemental map.     

Automated high-throughput electron tomography by pre-calibration of image shifts

Electron tomography is a versatile method for obtaining three‐dimensional (3D) images with transmission electron microscopy. The technique is suitable to investigate cell organelles and tissue sections (100–500 nm thick) with 4–20 nm resolution. 3D reconstructions are obtained by processing a series of images acquired with the samples tilted over different angles. While tilting the sample, image shifts and defocus changes of several µm can occur. The current generation of automated acquisition software detects and corrects for these changes with a procedure that incorporates switching the electron optical magnification. We developed a novel method for data collection based on the measurement of shifts prior to data acquisition, which results in a five‐fold increase in speed, enabling the acquisition of 151 images in less than 20 min. The method will enhance the quality of a tilt series by minimizing the amount of required focus‐change compensation by aligning the optical axis to the tilt axis of the specimen stage. The alignment is achieved by invoking an amount of image shift as deduced from the mathematical model describing the effect of specimen tilt. As examples for application in biological and materials sciences 3D reconstructions of a mitochondrion and a zeolite crystal are presented.