ADF STEM Research Articles

Dose limited reliability of quantitative annular dark field scanning transmission electron microscopy for nano-particle atom-counting

Quantitative annular dark field scanning transmission electron microscopy (ADF STEM) has become a powerful technique to characterise nano-particles on an atomic scale. Because of their limited size and beam sensitivity, the atomic structure of such particles may become extremely challenging to determine. Therefore keeping the incoming electron dose to a minimum is important. However, this may reduce the reliability of quantitative ADF STEM which will here be demonstrated for nano-particle atom-counting. Based on experimental ADF STEM images of a real industrial catalyst, we discuss the limits for counting the number of atoms in a projected atomic column with single atom sensitivity. We diagnose these limits by combining a thorough statistical method and detailed image simulations.

6D electron microscopy: combining real-space and reciprocal-space tomography

STEM and TEM-based tomography has been used widely to study the 3D morphology of a wide range of materials. Similarly reciprocal space tomography in which a tilt-series of diffraction patterns are acquired offers a powerful method for the analysis of the atomic structure of crystalline materials. The natural progression is to combine these techniques into a complete three dimensional morphology and crystallography data set, allowing both features to be studied simultaneously. Using a tilt series of scanning precession electron diffraction measurements from a commercially available Ni-base superalloy as an example, the complete reciprocal lattice orientation for a number of components embedded within the matrix could be determined. It was straightforward to identify reciprocal lattice vectors that allowed dark-field images representing each phase to be produced post-acquisition. In turn these were combined using geometric tomography methods to yield a 3-D tomogram of the superalloy. Imaging these phases using conventional ADF STEM tomography would potentially be challenging given the compositional similarity between the different phases. From the combined dataset the spatial distribution of the component phases could be easily recovered but more importantly the orientational relationships between these different components could be unambiguously determined. In this way the thermo-mechanical history of the sample could be inferred from the arrangement of coherent and semi-coherent interfaces and a previously unreported crystallographic registry between metal carbide (MC) and the matrix f.c.c. phases could been identified. The possibilities for development and applications of this technique will be discussed further.

Imaging and diffraction characterisation of 2D inorganic nanostructures

2D nanomaterials are a novel class of materials, noted for their potential in a range of high impact applications, such as sensing catalysis, and composite reinforcement. However, their structural features have not yet been fully characterised. In this work we have shown that 2D nanostructures of hBN, MoS2, and WS2, prepared by liquid exfoliation, can stack in sequences different from the bulk stacking. In 2D hBN non-sequential stacking was observed. 2D MoS2 and WS2 appear to have stacking where Mo/W atoms are located on the top of each other which differs from naturally occurring 2H and 3R polytypes. The majority of the hBN and MoS2 2D nanostructures retain the stacking of their bulk counterparts, while in 2D WS2 non-bulk stacking dominates. The conclusions are based primarily on the atomically resolved ADF STEM images with supporting evidence from electron diffraction.

Low angle ADF STEM defect imaging

Extended abstract of a paper presented at Microscopy and Microanalysis 2012 in Phoenix, Arizona, USA, July 29 – August 2, 2012.

ADF STEM imaging of screw dislocations viewed end-on

This paper presents annular dark-field scanning transmission electron microscope image simulations of a screw dislocation viewed end-on in a thin single crystal of Mo, taking into account surface relaxation (the Eshelby twist). The image contrast can be understood in terms of the effects of the displacements normal to the dislocation arising from the Eshelby twist on the channelling behaviour and interband scattering of the incident beam. With the beam focussed at the entrance surface, the image peak positions reflect the positions of the atoms at the entrance surface. For atomic columns at distances from the core less than the foil thickness, the image peak positions are predicted to lie between the perfect crystal and actual surface atom positions. The predicted intensity distribution of the image is qualitatively similar to that of a published experimental image of a screw dislocation in GaN [Phys. Rev. Lett. 91 (2003) p. 165501]. An assessment is made of the possibility of imaging core displacements by focussing near the foil centre, where surface relaxation effects should be minimised.

Uncovering Interface Structure by Column Shape Analysis in ADF STEM Images

Extended abstract of a paper presented at Microscopy and Microanalysis 2010 in Portland, Oregon, USA, August 1 – August 5, 2010.

Dislocation tomography made easy: a reconstruction from ADF STEM images obtained using automated image shift correction

After previous work producing a successful 3D tomographic reconstruction of dislocations in GaN from conventional weak-beam dark-field (WBDF) images, we have reconstructed a cascade of dislocations in deformed and annealed silicon to a comparable standard using the more experimentally straightforward technique of STEM annular dark-field imaging (STEM ADF). In this mode, image contrast was much more consistent over the specimen tilt range than in conventional weak-beam dark-field imaging. Automatic acquisition software could thus restore the correct dislocation array to the field of view at each tilt angle, though manual focusing was still required. Reconstruction was carried out by sequential iterative reconstruction technique using FEI's Inspect3D software. Dislocations were distributed non-uniformly along cascades, with sparse areas between denser clumps in which individual dislocations of in-plane image width 24 nm could be distinguished in images and reconstruction. Denser areas showed more complicated stacking-fault contrast, hampering tomographic reconstruction. The general three-dimensional form of the denser areas was reproduced well, showing the dislocation array to be planar and not parallel to the foil surfaces.

Three-Dimensional Observation of Dislocations by Electron Tomography in a Silicon Crystal

Dislocations in a silicon single crystal introduced by three point-bending at a high temperature were observed by electron tomography in annular dark field-scanning transmission electron microscopy (ADF-STEM). Commercially available P type (001) single crystal wafers were employed. An ADF STEM tilt series was acquired from -60° to +60° in tilt range with 2° in tilt step. The diffraction vector was maintained close to g(hkl) = 220 during the acquisition by adjusting the [110] direction of the sample parallel to the tilt axis of the holder. The observed dislocations were reconstructed by simultaneous interactive reconstruction technique, exhibiting a 3-D configuration of dislocations introduced by the three-point bending.

Resolution of coherent and incoherent imaging systems reconsidered - Classical criteria and a statistical alternative

The resolution of coherent and incoherent imaging systems is usually evaluated in terms of classical resolution criteria, such as Rayleigh's. Based on these criteria, incoherent imaging is generally concluded to be 'better' than coherent imaging. However, this paper reveals some misconceptions in the application of the classical criteria, which may lead to wrong conclusions. Furthermore, it is shown that classical resolution criteria are no longer appropriate if images are interpreted quantitatively instead of qualitatively. Then one needs an alternative criterion to compare coherent and incoherent imaging systems objectively. Such a criterion, which relates resolution to statistical measurement precision, is proposed in this paper. It is applied in the field of electron microscopy, where the question whether coherent high resolution transmission electron microscopy (HRTEM) or incoherent annular dark field scanning transmission electron microscopy (ADF STEM) is preferable has been an issue of considerable debate.

EELS Spectrum Imaging and Tomography Studies of Simulated Nuclear Waste Glasses

AbstractElectron energy loss spectroscopy (EELS) has been widely used in analysis of ceramic, minerals and semiconductors. It can provide the unique advantage of sensitivity to composition, coordination and valency, while providing nano-scale spatial resolution. Due to the electron irradiation sensitivity, EELS studies on glass are quite rare. However, using special care in glass-composition selection and by recording time series of EELS-spectra to monitor any structural changes, EELS then becomes a very useful technique to study glass-chemistry and glass structure with the highest spatial resolution of all chemically sensitive techniques. Alkali borosilicate glasses (ABS) doped with Cr2O3 (2 mol%), CeO2 (4 mol%) and ZrO4 (4 mol%) were melted, cooled and annealed in the context of simulating radionuclide immobilisation glasses. Precipitates with diameter in the range of ∼20 nm to ∼200 nm were found homogeneously distributed in the glass. In preliminary studies we measured the oxidation state in the glass and in precipitates via evaluation of the Ce-M-edge double white line ratio, confirming the crystals as Ce(IV) oxide. We also found Boron K-edge ELNES spectra as a sensitive signal to boron coordination (N4 = BO4/(BO3+BO4)), which is found at N4 ratios of around 35-45%.In the present study we have extended this work to the acquisition of dense line scans (spectrum imaging) across an area of interest on our composite glass with emphasis on three questions:(i) Change of cerium valence upon transition from the glass to the precipitates?(ii) Possible oscillation of boron N4-ratio in the glass within a line scan as a consequence of random medium-order related glass fluctations when scanning the <5nm probe across a seemingly homogeneous area?(iii) N4-changes in the immediate glass layer surrounding the precipitates?Our ongoing research has produced results which prove a systematic change of the Ce oxidation state from precipitates (+IV) to glass (mixed +III/+IV). It is found that boron coordination is constant within the sensitivity given by signal-to-noise in the glass matrix. A change of the Boron K-edge fine structure upon crossing a nanoparticle during one particular linescan has been observed, which could hint to a coordination change of the glass layer surrounding the particle. Other linescan measurements have not shown any change and acquisition of an enlarged better statistic is work in progress.The ABS-glass with distributed nanoparticles provides an ideal example of a 3D nanocomposite. The only proven technique for the three-dimensional reconstruction of such materials on the nanoscale is electron tomography. We present the first results of a 3D reconstructed nuclear waste glass by using a tilt series of ADF STEM images covering a glass fragment of 2000nm field of view containing several tens of nanoparticles distributed around its volume.

Detector geometry, thermal diffuse scattering and strain effects in ADF STEM imaging

Intensities of atomic-resolution Annular Dark Field Scanning Transmission Electron Microscopy (ADF STEM) images of zone-axis-oriented specimens change with defocus at rates that depend on lattice spacing. Thickness and strain effects on the intensities have been demonstrated. In this paper, image simulations (with some experimental basis) are presented that consider the dimensions of the ADF detector. Changing the inner radius of the detector seems to have relatively small effect on the image except to lower the detected intensity. Probe size was explored and a case identified where multiple scattering was important in the image. Thermal diffuse scattering (TDS) is important in high-angle scattering at room temperatures but it does not seem to alter the image appearance markedly. Finally, the image arising from the strain field around a single boron atom has been simulated and the results suggest increased scattering in agreement with observations. This mechanism may be adequate for single impurity atom detection at low temperatures and with special detector angles.

Annular Dark Field Imaging in Stem

ABSTRACTAnnular dark field scanning transmission electron microscopy (ADF STEM) is chemically sensitive at high spatial resolution (e.g., 1.8ë at 100keV). Images can be digitally acquired and recorded, permitting quantitative analysis. It is particularly powerful when used in combination with complementary analysis modes such as x-ray microanalysis and transmission electron energy loss spectroscopy. Critical to the interpretation of these data is an understanding and determination of the electron probe intensity, shape and propagation characteristics inside the specimen. Quantitative measurements of diffraction patterns and images in comparison with computer-based simulations (including phonon scattering) provide a basis for developing that information. Results of a series of studies are reviewed that address questions such as defocus and other instrumental factors, and also the formation of channeling peaks that appear on the atomic columns along zone axes. For example, along Si(100) a peak forms and penetrates over 500ë whereas along Ge(100) it developes rapidly but disappears in less than 200ë. In higher atomic number elements, the penetration is even less (e.g. 1 O0ë for In).

Thickness effects in ADF STEM zone axis images

Quantitative measurements of intensities in annular dark field scanning transmission electron microscope images of (100) indium phosphide are reported. Fourier analysis of the experimental images gives the lattice fringe intensities which are related to the intensities of individual atomic columns. Electron energy loss measurements of the plasmon intensities provide a thickness. Analysis of simulated images for the same instrumental conditions provides a comparison that is consistent with an earlier prediction that electrons channel for considerably different distances along atomic columns of differing atomic number. This effect has considerable implications for image interpretation and analytical studies. The measurements indicate that reasonable partial coherence effects give rise to serious limitations on imaging at the highest resolution levels.

Thickness-dependent annular dark-field STEM

The demonstration of Z-contrast atom structure images and high spatial resolution using annular dark field STEM has drawn considerable interest. This approach is well suited for quantitative image and diffraction pattern acquisition. Experimental data has been used to validate calculations of convergent beam electron diffraction (CBED) patterns in silicon and ADF STEM images of (100) InP. These procedures can be extended to explore the dependence of images on factors such as thickness.Energy filtered CBED patterns over an angular range of over 150 mrads and covering a range of over four orders of magnitude in intensity from silicon were recorded with a VG HB501A STEM for thicknesses of 217Å, 326Å and 543Å. Simulation using a “frozen phonon” approximation successfully modeled this data provided an accurate phonon amplitude was used. A similar quantitative study for images has been carried out for (100) oriented InP. This sample contains indium columns at the comer of a 2.9Å square and phosphorus atoms at the square center.

Annular dark-held images of crystals

The demonstration of Z-contrast atom structure images and high spatial resolution using annular dark field STEM has drawn considerable interest. This approach is well suited for quantitative image and diffraction pattern acquisition. It has thus proven possible to use experimental data to validate calculations of convergent beam electron diffraction (CBED) patterns in silicon and ADF STEM images of (100) InP. Such simulations can then be extended, as reported here, to explore the dependence of images on factors such as thickness.Energy filtered CBED patterns over an angular range of over 150 mrads and covering a range of over four orders of magnitude in intensity from silicon were recorded with a VG HB501A STEM for thicknesses of 217Å, 326Å and 543Å. Simulation using a “frozen phonon” approximation successfully modeled this data provided an accurate phonon amplitude was used. A similar quantitative study for images has been carried out for (100) oriented InP.

Resolution limits in annular dark-field STEM

Annular dark field scanning transmission electron microscopy (ADF STEM) is capable of resolving atomic structures with Z contrast. Better than 2 Å resolution at 100 kV has been demonstrated with a Cs of 1.3 mm and 0.7 mm. The images are apparently simple to interpret and change little with thickness or defocus. However, much of the reported work is qualitative, or semi-quantitative at best, so that the customarily adopted incoherent imaging model is not well established and a clear understanding of the limits (e.g., detectable AZ at a given ratio of spacing to resolution) is lacking. We present an exploration of such limits for the simple specimen, (100) InP, using quantitative ADF STEM and image simulation.

Thermal vibrations in simulated ADF STEM

Thermal vibrations reduce Bragg scattering, create a thermal diffuse background, and disrupt channeling, all of which may affect the ADF STEM signal. There are claims that thermal vibrations can change the relative contrast of different elements in the ADF STEM image, which is different from the simple Z- contrast interpretation. Evaluating the effects of thermal vibrations may be necessary for correct interpretation of experimental results. We present an investigation of ADF STEM imaging including thermal vibrations based on the frozen phonon model.Thermal vibrations can be included in the multislice calculation with a Monte Carlo technique. Since, the interaction time for 100 keV electrons is ≤ 10-4 atomic vibration periods, the atoms may be considered stationary. The time between successive electrons in the STEM is ≥ 102 vibration periods, so the atomic positions seen by different electrons are uncorrelated. The technique is to offset every atom in the specimen (many slices) by a small random amount perpendicular to the beam before performing a standard multislice calculation. A set of random offsets freezes one phonon configuration into the specimen. The calculation is then repeated and averaged over an ensemble of different configurations.

Hollow-cone illumination in simulated ADF STEM

The overlap regions between CBED disks are a source of lattice fringe contrast in ADF STEM images of crystals. For sufficiently small lattice spacings, the Scherzer focus convergence angle is too small for the CBED disks to overlap, and the lattice is not visible. By increasing the convergence angle beyond the Scherzer angle to create an overlap region, and by using stationary phase defocus, smaller lattices can be imaged However, with larger convergence angles, lens aberrations spread the incident beam and small aperiodic structures, such as adatoms, become difficult to image. If the specimen Bragg angle is known, an annular objective aperture can be designed to pass the overlap regions of the incident beam, while blocking the central, non-overlap region which produces a large constant background. The annular probe size is minimized by choosing an annulus width and defocus which balance the spherical aberration across the annulus (close to stationary phase defocus). Simulations indicate hollow cone illumination may allow both adatoms and smaller lattices to be imaged simultaneously. This approach presumes knowledge of the specimen, and must be used in conjunction with standard methods for correct image interpretation. An incorrectly applied annulus will produce image artifacts.

Wavefunction and CBED results of the STEM multislice calculation

The multislice algorithm has been used to simulate ADF STEM images. Examination of the evolution of the electron wavefunction as it propagates through the specimen, can provide insight into the sources of contrast in the STEM image. Plots of the wavefunction intensity as a function of position and as a convergent beam electron diffraction (CBED) pattern are two complementary views of the diffraction process. Examples from the large number of these plots that are calculated during the ADF STEM calculations will be presented.The simulated specimen consists of multiples of 47 Å (15 slices) of silicon (111). The slices are 65 Å × 66 Å(512 × 512 pixels) in size, setting the maximum included scattering angle to 95 mrad. The incident probe models a 100 keV VG-HB501 STEM at Scherzer focus with either the low resolution pole piece (Cs=3.3 mm, Δf=1100 Å, αap = 8.2 mrad) or the high resolution pole piece (Cs = 0.7 mm, Δf=510 Å, αap=12.1 mrad). The beam and specimen are aligned exactly along the (111) zone axis (no tilt).

Visibility of single heavy atoms on thin crystalline silicon in simulated annular dark-field STEM images

The multislice method, pioneered by Cowley and Moodie, has recently been adapted to simulate annular dark-field scanning transmission electron-microscope (ADF STEM) images. This paper presents a series of calculations using this new approach with experimental parameters appropriate for a VG-HB501 STEM to investigate the visibility of single heavy adatoms on thin crystalline silicon membranes. The tendency for electrons to channel along columns of atoms in crystals can greatly increase the intensity incident on an adatom on the exit surface, thereby increasing the adatom visibility. The simulations indicate that an adatom on the exit surface on a column of crystal atoms is up to three times as visible as an adatom on the entrance surface, and that the adatom remains highly visible as the crystal thickness is increased. Tilting the specimen or displacing the adatom from the column appears to lower the visibility of the adatom dramatically. These calculations suggest that, with the appropriate imaging conditions, a single gold adatom may be visible on at least 235 of (111) silicon.