Composition analysis of semiconductor quantum wells by energy filtered convergent-beam electron diffraction

We show how energy-filtered convergent-beam electron diffraction (EFCBED) patterns can be used to determine the chemical composition of buried semiconductor strained quantum wells. Our method is based on a quantitative analysis of the intensities of high-order Bragg lines in the transmitted disc of EFCBED patterns taken from plan-view samples. This analysis makes it possible to determine the displacement vector R introduced between the top and bottom parts of the matrix by the deformation of the quantum well and consequently to determine its composition. This is illustrated in the case of an In(x)Ga(1-)(x)As quantum well buried in a GaAs matrix. A detailed analysis of the effect of experimental parameters on Bragg lines intensity is performed. In particular, the importance of the choice of the diffraction vector is pointed out. The relative uncertainty on the measurement of the indium content x is found to be lower than 5% and a possible occurrence of slight compositional fluctuations in the (001) growth plane is pointed out.     

Combining Ar ion milling with FIB lift-out techniques to prepare high quality site-specific TEM samples

Focused ion beam (FIB) techniques can prepare site‐specific transmission electron microscopy (TEM) cross‐section samples very quickly but they suffer from beam damage by the high energy Ga⁺ ion beam. An amorphous layer about 20–30 nm thick on each side of the TEM lamella and the supporting carbon film makes FIB‐prepared samples inferior to the traditional Ar⁺ thinned samples for some investigations such as high resolution transmission electron microscopy (HRTEM) and electron energy loss spectroscopy (EELS). We have developed techniques to combine broad argon ion milling with focused ion beam lift‐out methods to prepare high‐quality site‐specific TEM cross‐section samples. Site‐specific TEM cross‐sections were prepared by FIB and lifted out using a Narishige micromanipulator onto a half copper‐grid coated with carbon film. Pt deposition by FIB was used to bond the lamellae to the Cu grid, then the coating carbon film was removed and the sample on the bare Cu grid was polished by the usual broad beam Ar⁺ milling. By doing so, the thickness of the surface amorphous layers is reduced substantially and the sample quality for TEM observation is as good as the traditional Ar⁺ milled samples.     

New approach for the dynamical simulation of CBED patterns in heavily strained specimens

A new method for the dynamical simulation of convergent beam electron diffraction (CBED) patterns is proposed. In this method, the three-dimensional stationary Schrödinger equation is replaced by a two-dimensional time-dependent equation, in which the direction of propagation of the electron beam, variable z, stands as a time. We demonstrate that this approach is particularly well-suited for the calculation of the diffracted intensities in the case of a z-dependent crystal potential. The corresponding software has been developed and implemented for simulating CBED patterns of various specimens, from perfect crystals to heavily strained cross-sectional specimens. Evidence is given for the remarkable agreement between simulated and experimental patterns.     

Determination of crystal polarity from bend contours in transmission electron microscope images

A new method for determining the polarity of crystals with sphalerite structure (GaAs, GaSb, InP, etc.) within the transmission electron microscope (TEM) is presented. The method is derived from an established convergent beam electron diffraction (CBED) method (J. Appl. Crystallogr. 15 (1982) 60) and exploits the effects of the dynamical scattering on the contrast of bend contour crossings in conventional TEM images. In contrast to the CBED method, the bend contour method is performed in the image mode of the TEM. The sample can, therefore, be viewed while performing the polarity analysis. Furthermore, in the presence of strong foil bending, the bend contour method has some advantages for practical work. A general contrast rule for the bend contour intersections is stated which allows to readily obtain the crystal polarity by comparing the contrast in experimental images with the prediction of the rule. Exemplarily, the polarity of GaAs in TEM samples prepared for investigation in the two frequently used projections < 001 > and < 110 > is determined. The validity of the rule for the cases studied is confirmed by simulations of the dynamical scattering process. Furthermore, an independent analysis of the crystal polarity by making use of a long-range-ordered (GaIn)P layer grown on top of the GaAs confirms the results obtained with the bend contour method. As an example, the usefulness of the method is demonstrated in an analysis of the alpha/beta-character of misfit dislocations at the interface between the GaAs substrate and the (GaIn)P layer.     

Active pixel sensor array as a detector for electron microscopy

A new high-resolution recording device for transmission electron microscopy (TEM) is urgently needed. Neither film nor CCD cameras are systems that allow for efficient 3-D high-resolution particle reconstruction. We tested an active pixel sensor (APS) array as a replacement device at 200, 300, and 400 keV using a JEOL JEM-2000 FX II and a JEM-4000 EX electron microscope. For this experiment, we used an APS prototype with an area of 64 x 64 pixels of 20 microm x 20 microm pixel pitch. Single-electron events were measured by using very low beam intensity. The histogram of the incident electron energy deposited in the sensor shows a Landau distribution at low energies, as well as unexpected events at higher absorbed energies. After careful study, we concluded that backscattering in the silicon substrate and re-entering the sensitive epitaxial layer a second time with much lower speed caused the unexpected events. Exhaustive simulation experiments confirmed the existence of these back-scattered electrons. For the APS to be usable, the back-scattered electron events must be eliminated, perhaps by thinning the substrate to less than 30 microm. By using experimental data taken with an APS chip with a standard silicon substrate (300 microm) and adjusting the results to take into account the effect of a thinned silicon substrate (30 microm), we found an estimate of the signal-to-noise ratio for a back-thinned detector in the energy range of 200-400 keV was about 10:1 and an estimate for the spatial resolution was about 10 microm.     

Novel method for the plan-view TEM preparation of thin samples on brittle substrates by mechanical and ion beam thinning

A new preparation method has been developed in order to avoid the breaking of brittle samples for plan-view TEM investigation during and after mechanical and ion beam thinning. The thinning procedure is carried out on a reduced size piece of the sample (about 1.6 x 0.8 mm(2) or about 1-1.6 mm diameter) that is embedded into a 3-mm-diameter Ti disk, which fits the sample holder of the TEM. The small sample size and the supporting metal disk assure the mechanical stability and minimize the possibility of breaking during and after the preparation: The Ti disk is placed on adhesive kapton tape, a cut piece of the sample is put into the slot of the disk, pressed onto the tape and embedded with glue. The tape keeps the parts in place and in the same plane, keeps the sample surface safe from the embedding glue and can be removed easily after the glue solidifies. Subsequently, the embedded sample is thinned from the rear by well-known mechanical and ion beam techniques until electron transparency. This simple solution lowers the risk of failed sample preparation remarkably and makes it possible to reduce the thickness of the sample to about 50 microm by mechanical thinning. As a result, dimpling becomes unnecessary and low angle ion milling gives a large transparent area for TEM. Its efficiency has been proved by successful preparation of numerous thin film samples on Si, sapphire, and glass substrates. The method is compatible with the widespread cross-sectional thinning procedures, and can be easily adopted by TEM laboratories.     

Site-specific fracture plane determination using the FIB/TEM

A new method for the determination of the crystallographic indices of planar fracture surfaces is described. The key innovation is the use of a focused ion beam instrument to extract two transmission electron microscopy (TEM) foils from the fracture surface. Selected area diffraction of these foils in the TEM allows the determination of the fracture plane from the cross product of two crystallographic line directions contained within the plane. This allows the indices to be determined from relatively small fracture surfaces, affording fracture plane determinations from facets on polycrystalline samples. The validation of this method using cleavage fracture in pure zinc is described.     

Imaging of fullerene-like structures in CNx thin films by electron microscopy; sample preparation artefacts due to ion-beam milling

The microstructure of CN(x) thin films, deposited by reactive magnetron sputtering, was investigated by transmission electron microscopy (TEM) at 200kV in plan-view and cross-sectional samples. Imaging artefacts arise in high-resolution TEM due to overlap of nm-sized fullerene-like features for specimen thickness above 5nm. The thinnest and apparently artefact-free areas were obtained at the fracture edges of plan-view specimens floated-off from NaCl substrates. Cross-sectional samples were prepared by ion-beam milling at low energy to minimize sample preparation artefacts. The depth of the ion-bombardment-induced surface amorphization was determined by TEM cross sections of ion-milled fullerene-like CN(x) surfaces. The thickness of the damaged surface layer at 5 degrees grazing incidence was 13 and 10nm at 3 and 0.8keV, respectively, which is approximately three times larger than that observed on Si prepared under the same conditions. The shallowest damage depth, observed for 0.25keV, was less than 1nm. Chemical changes due to N loss and graphitization were also observed by X-ray photoelectron spectroscopy. As a consequence of chemical effects, sputtering rates of CN(x) films were similar to that of Si, which enables relatively fast ion-milling procedure compared to carbon compounds. No electron beam damage of fullerene-like CN(x) was observed at 200kV.     

Novel techniques of preparing TEM samples for characterization of irradiation damage

Focus ion beam preparation of transmission electron microscopy (TEM) samples has become increasingly popular due to the relative ease of extraction of TEM foils from specific locations within a larger sample. However the sputtering damage induced by Ga ion bombardment in focus ion beam means that traditional electropolishing may be a preferable method. First, we describe a special electropolishing method for the preparation of irregular TEM samples from ex‐service nuclear reactor components, spring‐shaped spacers. This method has also been used to prepare samples from a nonirradiated component for a TEM in situ heavy ion irradiation study. Because the specimen size is small (0.7 × 0.7 × 3 mm), a sandwich installation is adopted to obtain high quality polishing. Second, we describe some modifications to a conventional TEM cross‐section sample preparation method that employs Ni electroplating. There are limitations to this method when preparing cross‐section samples from either (1) metals which are difficult to activate for electroplating, or (2) a heavy ion irradiated foil with a very shallow damage layer close to the surface, which may be affected by the electroplating process. As a consequence, a novel technique for preparing cross‐section samples was developed and is described.     

001] zone-axis bright-field diffraction contrast from coherent Ge(Si) islands on Si(001)

Coherent Ge(Si)/Si(001) quantum dot islands grown by solid source molecular beam epitaxy at a growth temperature of 700 degrees C were investigated using transmission electron microscopy working at 300kV. The [001] zone-axis bright-field diffraction contrast images of the islands show strong periodicity with the change of the TEM sample substrate thickness and the period is equal to the effective extinction distance of the transmitted beam. Simulated images based on finite element models of the displacement field and using multi-beam dynamical diffraction theory show a high degree of agreement. Studies for a range of electron energies show the power of the technique for investigating composition segregation in quantum dot islands.     

Diffracted phase and amplitude measurements by energy-filtered convergent-beam holography (CHEF)

Interference between transmitted and diffracted disks in convergent-beam electron diffraction (CBED) patterns using the CBED+EBI method proposed by Herring et al. is explored using different optical configurations on a spherical aberration corrected transmission electron microscope equipped with a biprism and imaging energy filter: the SACTEM-Toulouse. We will relate the amplitude and phase of these interference patterns, which we call convergent-beam holography (CHEF), to microscope transfer theory and the complex amplitudes of the diffracted beams. Experimental CHEF patterns recorded in the absence of aberration correction will be compared with simulations to validate the theory concerning the effect of microscope aberrations and current instabilities. Then, using aberration correction, we propose a scheme for eliminating the effect of the microscope, so that the diffracted amplitudes and phase due to dynamical scattering within the specimen can be studied. Experimental results are compared with simulations performed using the full dynamical theory. The potential for studying diffracted amplitudes and phases using CHEF analysis is discussed.     

A simple method for depositing thin films on the surface of transmission electron microscope specimens

Thin-film deposition on already thinned transmission electron microscope (TEM) samples is generally difficult as these samples are often small and fragile. To overcome this difficulty, we have developed a new method for depositing thin films of practically any material onto TEM samples using an ion-milling machine. This principle utilizes the fact that an ion-milling machine can be used not only for removal but also for deposition of materials. Fine-grained thin films were found to grow on TEM samples by this method.     

In situ measurements and transmission electron microscopy of carbon nanotube field-effect transistors

We present the design and operation of a transmission electron microscopy (TEM)-compatible carbon nanotube (CNT) field-effect transistor (FET). The device is configured with microfabricated slits, which allows direct observation of CNTs in a FET using TEM and measurement of electrical transport while inside the TEM. As demonstrations of the device architecture, two examples are presented. The first example is an in situ electrical transport measurement of a bundle of carbon nanotubes. The second example is a study of electron beam radiation effect on CNT bundles using a 200 keV electron beam. In situ electrical transport measurement during the beam irradiation shows a signature of wall- or tube-breakdown. Stepwise current drops were observed when a high intensity electron beam was used to cut individual CNT bundles in a device with multiple bundles.     

Two‐In‐one sample preparation for plan‐VIew TEM

Transmission electron microscopy (TEM) sample preparation requires special skills, it is time consuming and costly, hence, an increase of the efficiency is of primary importance. This article describes a method that duplicates the yield of the conventional mechanical and ion beam preparation of plan‐view TEM samples. As a modification of the usual procedures, instead of one two different samples are comprised in a single specimen. The two pre‐cut slabs, one from each samples, are embedded side by side in the window of a 3 mm dia Ti disk and the specimen is thinned mechanically and by ion milling until perforation that occurs at the interface of the two different slabs. That, with proper implementation, provides acceptable size thin area for the TEM study of both samples. The suitability of the two‐in‐one method has been confirmed through examples. Microsc. Res. Tech. 78:599–602, 2015. © 2015 Wiley Periodicals, Inc.     

An efficient approach to characterize pseudo-merohedral twins by precession electron diffraction: Application to the LaGaO3 perovskite

Pseudo-merohedral twins are frequently observed in crystals displaying pseudo-symmetry. In these crystals, many [u v w] zone axis electron diffraction patterns are very close and can only be distinguished from intensity considerations. On conventional diffraction patterns (selected-area electron diffraction or microdiffraction), a strong dynamical behaviour averages the diffracted intensities so that only the positions of the reflections on a pattern can be considered. On precession electron diffraction patterns, the diffracted beams display an integrated intensity and a “few-beam” or “systematic row” behaviour prevails which strongly reduces the dynamical interactions. Therefore the diffracted intensity can be taken into account. A procedure based on observation of the weak extra-reflections connected with the pseudo-symmetry is given to identify without ambiguity any zone axis. It is successfully applied to the identification and characterization of {1 2 1} reflection twins present in the LaGaO₃ perovskite.     

Towards sub-A atomic resolution electron diffraction imaging of metallic nanoclusters: a simulation study of experimental parameters and reconstruction algorithms

A simulation study is carried out to elucidate the effects of dynamical scattering, electron beam convergence angle and detection noise on atomic resolution diffraction imaging of small particles and to develop effective reconstruction procedures. Au nanoclusters are used as model because of their strong scattering. The results show that the dynamical effects of electron diffraction place a limit on the size of Au nanoclusters that can be reconstructed from the diffraction intensities with sufficient accuracy. For smaller Au nanoclusters, the simulations show that diffraction patterns recorded under the experimental conditions can be reconstructed using a combination of phase retrieval algorithms. The use of a low-resolution image is shown to be effective for reconstructing diffraction patterns without the central beam. A new algorithm for estimating the object support is proposed.     

Making the practically impossible “Merely difficult”—Cryogenic FIB lift‐out for “Damage free” soft matter imaging

The preparation of thinned lamellae from bulk samples for transmission electron microscopy (TEM) analysis has been possible in the focussed ion beam scanning electron microscope (FIB‐SEM) for over 20 years via the in situ lift‐out method. Lift‐out offers a fast and site specific preparation method for TEM analysis, typically in the field of materials science. More recently it has been applied to a low‐water content biological sample (Rubino 2012). This work presents the successful lift‐out of high‐water content lamellae, under cryogenic conditions (cryo‐FIB lift‐out) and using a nanomanipulator retaining its full range of motion, which are advances on the work previously done by Rubino (2012). Strategies are explored for maintaining cryogenic conditions, grid attachment using cryo‐condensation of water and protection of the lamella when transferring to the TEM. Microsc. Res. Tech. 79:298–303, 2016. © 2016 Wiley Periodicals, Inc.     

The quantitative characterization of SiGe layers by analysing rocking profiles in CBED patterns

Convergent beam electron diffraction is used for the quantitative determination of layer thickness, composition and strain in pseudomorphic Si/SiGe two- and three-layer systems grown by molecular beam epitaxy. By using plan-view specimens, we are able to avoid the influence of surface relaxation which generally complicates the determination of strains in cross-sectional specimens. For quantitative strain determination, rocking curves of Bragg lines in energy-filtered convergent beam electron diffraction patterns are analysed. The experimentally obtained rocking curves are compared with kinematical calculations by a computerized fit procedure. The resulting layer parameters are then further refined by a dynamical simulation. Results for the strains obtained with this technique are in good agreement with theoretical values. With this method layer thickness is measured down to monolayer precision. The accuracy of the strain analysis depends on the sequence and thickness of the layers.     

Precession technique and electron diffractometry as new tools for crystal structure analysis and chemical bonding determination

We have developed a new fast electron diffractometer working with high dynamic range and linearity for crystal structure determinations. Electron diffraction (ED) patterns can be scanned serially in front of a Faraday cage detector; the total measurement time for several hundred ED reflections can be tens of seconds having high statistical accuracy for all measured intensities (1-2%). This new tool can be installed to any type of TEM without any column modification and is linked to a specially developed electron beam precession "Spinning Star" system. Precession of the electron beam (Vincent-Midgley technique) reduces dynamical effects allowing also use of accurate intensities for crystal structure analysis. We describe the technical characteristics of this new tool together with the first experimental results. Accurate measurement of electron diffraction intensities by electron diffractometer opens new possibilities not only for revealing unknown structures, but also for electrostatic potential determination and chemical bonding investigation. As an example, we present detailed atomic bonding information of CaF(2) as revealed for the first time by precise electron diffractometry.     

Surface damage formation during ion-beam thinning of samples for transmission electron microscopy

All techniques employed in the preparation of samples for transmission electron microscopy (TEM) introduce or include artifacts that can degrade the images of the materials being studied. One significant cause of this image degradation is surface amorphization. The damaged top and bottom surface layers of TEM samples can obscure subtle detail, particularly at high magnification. Of the techniques typically used for TEM sample preparation of semiconducting materials, cleaving produces samples with the least surface amorphization, followed by low-angle ion milling, conventional ion milling, and focused ion beam (FIB) preparation. In this work, we present direct measurements of surface damage on silicon produced during TEM sample preparation utilizing these techniques. The thinnest damaged layer formed on a silicon surface was measured as 1.5 nm thick, while an optimized FIB sample preparation process results in the formation of a 22 nm thick damaged layer. Lattice images are obtainable from all samples.