Zero-loss Peak Research Articles

An energy stabilized post-column electron energy-loss spectrometer for transmission electron microscopy

An energy stabilizing system for electron energy loss spectrometers in a transmission electron microscope (TEM) is described. A wire-based fast response detector is used to position the zero-loss peak. A digital-based real-time processor is used to stabilize the energy-dispersed electron spectrum and simultaneously to acquire a drift corrected electron energy loss spectrum. The performance of the stabilizing system is evaluated.

Fourier-ratio deconvolution and its Bayesian equivalent

We discuss how an inner-shell electron energy-loss spectrum can be processed using Bayesian (maximum-entropy or maximum-likelihood) deconvolution to simultaneously remove plural scattering and improve the energy resolution. As in Fourier-ratio deconvolution, a low-loss spectrum (recorded from the same area of specimen) is used as a kernel or resolution function. This procedure avoids the need to record the zero-loss peak in the absence of a specimen and uncertainties related to the width of the zero-loss peak. Unlike the case of Fourier-ratio deconvolution, we find that core-loss data do not require pre-edge background subtraction and extrapolation towards zero intensity; simply matching the intensity at both ends of the region is usually sufficient to avoid oscillatory artifacts. Using the low-loss spectrum as both data and kernel yields a zero-loss peak whose width provides an indication of the energy resolution as a function of the number of iterations. Finally, we argue that Fourier-ratio deconvolution or its Bayesian equivalent is the correct way to remove the substrate or matrix contribution to an energy-loss spectrum recorded from a particle on a substrate or embedded in a matrix.

Nanomaterial electronic structure investigation by valence electron energy loss spectroscopy—An example of doped ZnO nanowires

The effects of doping (by ion implantation) on the electronic structure of ZnO nanowires, particularly on the defect states generation in the band gap of ZnO, are investigated using valence electron energy loss spectroscopy (VEELS) performed in a transmission electron microscope (TEM). The improved spectrum energy resolution via the introduction of a gun monochromator, together with the reduced intensity in the zero loss peak tail as realized by spectrum acquisition at non-zero momentum transfer, enable us to extract such electronic structure information from the very low loss region of the EEL spectra. We have compared the doping effects of several dopant elements, i.e., Er, Yb, and Co, and found that generation of the band tail states (∼2–3.3 eV) is a common consequence of the ion implantation process. On the other hand, specific mid-gap state(s) in the lower energy range are created only in the rare earth element doped ZnO nanowires, suggesting the dopant-sensitive nature of such state.

Electron Energy Loss Near-Edge Structures in Complex Perovskites

High-resolution electron energy loss spectroscopy (EELS) provides better sensitivity to fine structures related to bonding in complex materials [1,2]. Using monochromators and improved spectrometers it is possible to achieve 0.1eV resolution with symmetric zero-loss peak profiles leading to improved detection of both core-loss and low-loss features [3] in EELS spectra. This capability makes it possible to study systematically changes in bonding in materials that present interesting physical properties. Oxides with the perovskite structure provide an interesting playing field to study many fundamental changes in structure and physical properties when systematic substitutions of atoms in the crystals are made in a controlled fashion. Substitutions of Ti atoms in BaTiO3 with other transition metals provide ingenious ways to change the stability of the basic structure and induce phase transformations. For example, Mn substitutions for Ti in BaTiO3 stabilize the hexagonal phase at room temperature and it has been proposed that this is due to increased metal-metal bonding between adjacent face-sharing octahedra [4]. EELS measurements in Ba(Ti,Mn)O3, however, have shown that, with respect to the tetragonal structure, there are only subtle changes in the oxygen K edge fine structures of the hexagonal compound due to overlaps of the Mn 3d electrons and O 2p electrons while the Ti environment is not altered significantly [5]. Ru substitutions for Ti also lead to the stabilization of the hexagonal phase but there are no reports on the effect of the changes in the near-edge structures related to the structural changes. For this particular case, it is of great interest, to study the closely related hexagonal perovskite BaRuO3. Probing electronic structure changes in these two materials with respect to BaTiO3 makes it possible to gain insight on both the electronic structure of these closely related compounds and the origin of the structural changes. We have therefore used EELS with high energy resolution to probe the near edge structures in BaTiO3, Ba(Ti,Ru)O3 and BaRuO3.

Retardation, Surface and Interface Effects in VEELS

R. Erni,* N. D. Browning**, P. Specht***, and C. Kisielowski**** * EMAT, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium ** Materials Science and Technology Division, Lawrence Livermore National Laboratory, P.O. Box 808, L-356, Livermore, CA 94550, and Department of Chemical Engineering and Materials Science, University of California Davis, One Shields Ave., Davis, CA 95616 *** Materials Sciences Division, Lawrence Berkeley National Laboratory (LBNL), One Cyclotron Road, Berkeley, CA 94720 **** National Center for Electron Microscopy, LBNL, One Cyclotron Road, Berkeley, CA 9420 Valence electron energy-loss spectroscopy (VEELS) in (scanning) transmission electron microscopy (STEM) offers the possibility to measure band-structure information and in particular band-gap and band-transition energies of semiconductor materials with (sub-)nanometer spatial resolution. Using an electron monochromator implemented in the gun area of a (scanning) transmission electron microscope, VEEL spectra can be recorded that reveal in an unprecedented way the spectral fine structure of the low-loss area. Down to an energy loss of about 1.0 eV the signal is hardly affected by the low-energy tail of the zero-loss peak. For the case that electrons are detected that are inelastically scattered at very small angles, VEELS approaches the optical limit providing information about the complex dielectric function e(

A method for thin foil thickness determination by transmission electron microscopy

With the intention of determining the local thickness within a crystalline thin foil specimen, by means of transmission electron microscopy (TEM), a method previously proposed by Zuo and Shi [J.M. Zuo, Y.F. Shi, Microsc. Microanal. 7 (Suppl. 2) (2001) 224–225] was applied. Using the convergent beam technique, with the incident beam parallel to a zone axis with low indices, diffraction patterns were obtained for some aluminum alloys with low solute content. These patterns were contrasted with those obtained from simulations based on the dynamic theory with Bloch's waves formalism. The local thickness of the thin foil was then obtained by visually comparing the simulated patterns with the experimental one. Comparison of the proposed method with that based on the analysis of two-beam convergent beam patterns [P.M. Kelly, A. Jostsons, R.G. Blake, J.G. Napier, Phys. Stat. Solidi (a) 31 (1975) 771–780] and with that based on the ratio of intensity of the zero loss peak to the total intensity in an electron energy loss spectrum [R.F. Egerton, Electron Energy Loss Spectroscopy in the Electron Microscope, second ed., Plenum Press, New York, 1996] was carried out. A very good agreement between thicknesses determined using the different methods was found. The sensitivity of the method of Zuo et al. was found to be about 1 or 2 nm. The advantages and limitations of the different methods are discussed. The method of Zuo et al. can provide fast and reliable results and can be applied in all modern instruments.

Electron energy loss spectroscopy of defects in diamond

AbstractLocalised energy loss spectroscopy (EELS) is carried out in order to extract information of the effect of extended defects on the electronic bandstructure in diamond. EELS is performed in a dedicated cold field emission scanning transmission electron microscope (STEM), equipped with a Gatan Enfina spectrometer, at high energy dispersions, allowing energy resolutions (FWHM of the zero loss peak) of around 0.3 eV to be achieved without monochromation, thus enabling low EELS to be carried out in the bandgap regime of diamond at high brightness. Experiments are carried out on single crystal CVD and brown and colourless natural diamonds. Experimental details and evaluation of spectra are discussed, and the effects of diffraction contrast on the EEL intensity in the vicinity of defects are critically examined. A general characteristic of extended defects in CVD diamonds appears to be a relative increase of the loss signal below 6 eV and a distinct feature at 6–7 eV, concomitant with sp2‐bonds. It emerges that these states do not reside at cores of single dislocations, but near dislocation arrangements, in particular at nodes. A stark phenomenon in brown natural diamond is the omni‐present increase in the EEL intensity below 6 eV over that of colourless samples, suggesting the existence of spatially dispersed, non‐dislocation related centres, with energies throughout the bandgap. Recent theoretical and experimental investigations have revealed strong links with vacancy clusters. In order to further support the vacancy cluster hypothesis high resolution electron transmission microscopy (HR TEM) imaging in combination with vacancy clusters simulations at high defocus values is performed. The simulations predict characteristic small‐scale contrast, which can be experimentally observed in brown, but not in colourless diamond. (© 2007 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)

On the formation mechanisms, spatial resolution and intensity of backscatter Kikuchi patterns

The present paper is divided into two main sections. In the first, the formation mechanisms of backscatter Kikuchi patterns (BKP) are discussed on the basis of measurements on the sharpness of Kikuchi lines and on the spatial, that means the lateral and depth resolution of the technique. We propose that thermal diffuse scattering is the important incoherent scattering mechanism involved in pattern formation. This mechanism is not considered in the classical description of the origin of backscattered electrons in the scanning electron microscope (SEM) which is why there is in some important points no agreement between classical Monte-Carlo-type electron trajectory simulations and experimental results. We assume that the energy spectrum of the backscattered electrons shows, similar to the spectra in transmission electron microscopy, a sharp zero-loss peak. In the second section, we discuss the intensity of Kikuchi bands in BKP. It is shown that the kinematical theory gives—of course—not the correct intensities, but that these intensities are, on the other hand, not too far off the experimental ones. We subsequently introduce a simple intensity correction procedure that is based on the two-beam dynamical theory, originally proposed by Blackman for transmission electron diffraction patterns. It is shown by examples of diffraction patterns of niobium and silicon that this procedure leads to satisfying results, once two unknown variables (a universal constant and the exit depth of the electrons) have been empirically fit. It is assumed that in the future, this correction will improve the possibilities of phase identification by backscatter Kikuchi diffraction patterns.

Enhancement of resolution in core-loss and low-loss spectroscopy in a monochromated microscope

The significant enhancement of the energy resolution in the new generation of commercially available monochromated transmission electron microscopes presents new challenges in term of selecting the correct experimental conditions and understanding the various effects that can potentially influence the quality of the EELS data. In this respect we investigated the effect of point spread function of the detector and spectrum-diffraction mixing on the energy resolution and the intensity of the zero loss peak tails. Alternative approaches to improve the energy resolution by mathematical methods have been tested. By using a simple and commonly available test case (Si L 2,3 edges) we assessed the efficiency of the deconvolution algorithms to improve the resolution. The results show that the deconvolution is not always successful in improving the resolution of the core loss EELS data and the results may not always be reliable. Contrary to this, the application of the Richardson–Lucy deconvolution algorithm on some bandgap measurements data appears to be very effective. The procedure proved successful in removing the contribution of the zero-loss peak tails and allows an easier access to spectroscopic information starting at energy losses as low as of 0.5 eV with monochromated spectra and 1 eV with the non-monochromated spectra.

Improving the energy resolution of X-ray and electron energy-loss spectra

We discuss some practical problems of improving the resolution of X-ray and electron spectra. Iterative Bayesian methods promise greater resolution enhancement than Fourier techniques but they also give rise to spectral artifacts. Satellite peaks are generated adjacent to strong peaks in the original spectrum and oscillatory artifacts become prominent after a large number of iterations, particularly when the original data contain high noise content. In the case of valence-electron energy-loss spectra, satellite peaks are reduced by removing the zero-loss peak prior to spectral sharpening. Even so, care should be exercised in interpreting low intensity at low energy loss (after sharpening) as evidence for a bandgap in the electronic density of states.

Test of dielectric–response model for energy and angular dependence of plasmon excitations in core-level photoemission

We compare experimental measurements of the loss structure appearing in the Si 2p and Si 1s photoemission lines for a large range of emission angles (between 0° and 82°) and kinetic energies (125–3660 eV) with calculations derived within a semiclassical dielectric–response formalism [A. Cohen Simonsen, F. Yubero, S. Tougaard, Phys. Rev. B 56 (1997) 1612]. It is found that this semi-classical dielectric description of the energy-loss processes reproduces the relative intensity of surface to bulk energy losses appearing in the lower kinetic energy side of the main photoelectron peaks as well as the relative intensity of the losses with respect to the zero-loss peak. The absolute ratio of intensity of the loss structure to the zero-loss is ∼25–35% lower compared to experiment using self-consistent inelastic mean free paths in the analysis.

Study of the electronic structure of carbon materials near the bandgap using electron energy loss spectroscopy

The detailed understanding of the electronic properties of semiconducting materials requires the determination of their joint density of states (JDOS) and dielectric constant with high spatial resolution. Low electron energy loss spectroscopy (EELS) provides a continuous spectrum which represents all electronic excitations with energies ranging between zero and about 100 eV. Therefore EELS is potentially more powerful than conventional optical spectroscopy which has an intrinsic upper information limit of about 6 eV due to absorption of light from the optical components of the system or the ambient. However, when analysing EELS data, the extraction of the single scattered data needed for Kramers–Kronig calculations is subject to the deconvolution of the zero loss peak from the raw data. This procedure is particularly critical when attempting to study the near-bandgap region of materials with a bandgap below 1.5 eV. In this paper calculation of the electronic properties of some widely studied carbon materials; namely amorphous carbon (a-C), tetrahedral amorphous carbon (ta-C), C60 fullerite crystal as well as synthetic and natural diamond, are presented. Extrapolation of a fit to the absorption data at the onset of an interband transition was used to get an accurate value for the bandgap energy and to determine whether the bandgap is direct or indirect in character. Particular problems relating to the extraction of the single scattered data for these materials are also discussed. The ta-C and C60 fullerite materials are found to be direct bandgap-like semiconductors, having a bandgap of 2.63 and 1.59 eV, respectively. On the other hand, the electronic structure of a-C was unobtainable because it had such a small bandgap that most of the information is contained in the first 1.2 eV of the spectrum, which is a region removed during the zero loss peak deconvolution. The processing of the data obtained from the ta-C material revealed another benefit of using EELS: the electrons of the beam can transfer momentum and energy to the electrons of the material, thus enabling the probing of indirect interband transitions. Use of this extra feature of EELS coupled with the extremely high spatial resolution obtainable in a scanning electron microscope has indicated that tetrahedral carbon thin films might not be amorphous after all, implying the need for a fresh look at their use as electronic materials.

Extended defect related energy loss in CVD diamond revealed by spectrum imaging in a dedicated STEM

This article aims at investigations of the low EEL region in the wide band gap system diamond. The advent of the UHV Enfina electron energy loss spectrometer combined with Digital Micrograph acquisition and processing software has made reliable detection of absorption losses below 10 eV possible. Incorporated into a dedicated STEM this instrumentation allows the acquisition of spectral information via spectrum maps (spectrum imaging) of sample areas hundreds of nanometers across, with nanometers pixel sizes, adequate spectrum statistics and 0.3 eV energy resolution, in direct correlation with microstructural features in the mapping area. We aim at discerning defect related losses at band gap energies, and discuss different routes to simultaneously process and analyse the spectra in a map. This involves extracting the zero loss peak from each spectrum and constructing ratio maps from the intensities in two energy windows, one defect related and one at a higher, crystal bandstructure dominated energy. This was applied to the residual spectrum maps and their first derivatives. Secondly, guided by theoretical EEL spectra calculations, the low loss spectra were fitted by a series of gaussian distributions. Pixel maps were constructed from amplitude ratios of gaussians, situated in the defect and the unaffected energy regime. The results demonstrate the existence of sp 2-bonded carbon in the vicinity of stacking faults and partial dislocations in CVD diamond as well as additional states below conduction band, tailing deep into the band gap, at a node in a perfect dislocation. Calculated EEL spectra of shuffle dislocations give similar absorption features at 5–8 eV, and it is thought that this common feature is due to sp 2-type bonding.

Contributions of Inelastically Scattered Electrons to Defect Images

We have made measurements of the contribution of inelastically scattered electrons to images of dislocations in Ni{sub 3}Ga and nanometer-sized defects in ion-irradiated Au under weak-beam dark-field diffraction conditions [1]. The purpose is to determine the conditions for data acquisition required to eventually make detailed and quantitative comparisons to simulations of images for various defect models, thus determining defect structure, composition, and local strain field. Such image simulations usually consider only elastically scattered electrons, and thus it is important to understand and possibly eliminate the contribution of inelastically scattered electrons to the experimental images for quantitative comparisons with image simulations. Experimental data have been acquired with either JEOL 2010F or 3000F microscopes, both equipped with Gatan Imaging Filter electron spectrometers. Samples examined in the 2010F were Au, ion-irradiated to low dose (10{sup 11}Kr ions at 1 MeV energy) to form individual defects (1-10nm sized Frank dislocation loops and partial stacking fault tetrahedra). Samples examined in the 3000F were Ni{sub 3}Ga with long dislocation defects. Imaging conditions included weak-beam dark-field with deviation parameter generally > 0.2 nm{sup -1}. Energy filter slit width was set to 10 eV and centered on the zero loss peak in both instruments to obtain images produced by elastically scattered electrons. In the Au experiment an energy window was also set to image the 10-40 eV loss range to investigate the contribution to the defect images of plasmon-loss electrons. Electron intensity in defect images and backgrounds was measured by summing pixel values in appropriately sized rectangles using Digital Micrograph (Gatan) software.

A fast beam switch for controlling the intensity in electron energy loss spectrometry

The fast deflection system described in this paper is suitable for controlling the intensity reaching the detector of a magnetic sector electron spectrometer mounted below an analytical transmission electron microscope. Amongst other things, this allows the low loss region of the spectrum to be recorded with the same electron probe conditions used to record core losses, something that is essential for high spatial resolution studies. The plate assembly restricts the width of the electron distribution reaching the viewing screen to a strip ∼17 mm wide in the direction approximately normal to the dispersion direction of the spectrometer. The resulting deflection has no detectable effect on the FWHM of the zero-loss peak for exposure times as short as 1 μs. At incident energies up to 300 keV, positioning the deflection plates in the 35 mm camera port above the viewing chamber allows voltages of <±3 kV to deflect the electrons out of the spectrometer and beyond the edge of the annular detector. When the deflection is switched on, the electrons are deflected out of the spectrometer in ⪡40 ns and when the deflection is switched off, the electrons return to within 10 μm of the undeflected position within 100 ns. Thus, even at an exposure time of 30 μs, the smallest time likely to be used in practice with a GATAN 666 spectrometer, <1% of the signal in the spectrum is from electrons whose scattering conditions differ from those in the undeflected position. The performance of the deflection system is such that it will also be suitable for use with the new and much faster GATAN ENFINA spectrometer system. At incident energies up to 200 keV and possibly up to 300 keV, deflection voltages of ±3 kV are sufficient to deflect the electrons off a 1 k×1 k charge coupled device (CCD) camera placed below the photographic camera. Thus the deflection system can be used as a very fast, non-mechanical shutter for such a CCD camera.

Background subtraction for low-loss transmission electron energy-loss spectroscopy

We present a technique for removing the zero-loss background from electron energy-loss spectra at very low energies (down to ∼2 eV), generating results that are superior in a number of ways to the results of standard Fourier deconvolution techniques. Our technique is based on a separately measured background spectrum which is spline-interpolated and matched to the zero-loss peak in the low-loss spectrum using curve-fit techniques. The data points are weighted with the use of a semi-empirical model of the random error in the data produced by a spectrometer. We demonstrate in tests on real-world data that this model accounts for the random error within the energy range of interest. We discuss practical details of implementation and present detailed comparisons of the results of various algorithms on a piece of test data obtained from a carbon nanotube sample. Compared to the standard techniques, our algorithm tends to be more consistent, less dependent on arbitrary parameters, and better able to quantify spectral features with small signal-to-noise ratios, particularly those at very low energies.

Calculation of the electronic structure of carbon films using electron energy loss spectroscopy

The detailed understanding of the electronic properties of carbon-based materials requires the determination of their electronic structure and more precisely the calculation of their joint density of states (JDOS) and dielectric constant. Low electron energy loss spectroscopy (EELS) provides a continuous spectrum which represents all the excitations of the electrons within the material with energies ranging between zero and about 100 eV. Therefore, EELS is potentially more powerful than conventional optical spectroscopy which has an intrinsic upper information limit of about 6 eV due to absorption of light from the optical components of the system or the ambient. However, when analysing EELS data, the extraction of the single scattered data needed for Kramers–Kronig calculations is subject to the deconvolution of the zero loss peak from the raw data. This procedure is particularly critical when attempting to study the near-bandgap region of materials with a bandgap below 1.5 eV. In this paper, we have calculated the electronic properties of three widely studied carbon materials; namely amorphous carbon (a-C), tetrahedral amorphous carbon (ta-C) and C 60 fullerite crystal. The JDOS curve starts from zero for energy values below the bandgap and then starts to rise with a rate depending on whether the material has a direct or an indirect bandgap. Extrapolating a fit to the data immediately above the bandgap in the stronger energy loss region was used to get an accurate value for the bandgap energy and to determine whether the bandgap is direct or indirect in character. Particular problems relating to the extraction of the single scattered data for these materials are also addressed. The ta-C and C 60 fullerite materials are found to be direct bandgap-like semiconductors having a bandgaps of 2.63 and 1.59 eV, respectively. On the other hand, the electronic structure of a-C was unobtainable because it had such a small bandgap that most of the information is contained in the first 1.2 eV of the spectrum, which is a region removed during the zero loss deconvolution.

Electron energy-loss spectroscopy of silicate perovskite-magnesiowüstite high-pressure assemblages

Silicate perovskite-magnesiowustite assemblages synthesized from natural olivine in the multi-anvil press and diamond-anvil cell were studied by electron energy-loss spectroscopy (EELS). Spectra of crystalline silicate perovskite, and its post-amorphization phase, as well as magnesiowustite were collected at the Fe and Si L 2,3 edge, and in the low loss (<50 eV) domain. The technique of line spectra ensuring very low beam doses allows good quality spectra to be collected from crystalline perovskite prior to amorphization and permits characterization of coexisting crystals of perovskite and magnesiowustite. Spectra at the Si L 2,3 edge show that the beam-induced amorphization of silicate perovskite is accompanied by a change from sixfold to fourfold oxygen coordination of silicon atoms. Spectra at the Fe L 2,3 edge show that Fe2+ is the major form of Fe in olivine, ringwoodite, and magnesiowustite, whereas Fe3+ is dominant in crystalline silicate perovskite and its amorphization products. In magnesiowustite and silicate perovskite observed in contact in these samples, Fe3+ is strongly partitioned into the silicate phase. Careful experimental substraction of zero-loss peak by off Bragg acquisition of electron energy-loss spectra allows good quality low loss spectra to be collected from crystalline silicate perovskite and magnesiowustite. In magnesiowustite, interband transitions are well characterized, leading to a measured gap of 7.8 eV, in agreement with previous theoretical calculations. Interband transitions at 10 eV and 12.5 eV are also well resolved in crystalline silicate perovskite, leading to a gap of about 9.5 eV.

Determination of Mean Free Path for Inelastic Scattering in Poly(2-Vinyl Pyridine) by Low-Loss Spectrum Imaging

Abstract The common experimental method to determine the total inelastic mean free path i by electron energy-loss spectroscopy (EELS) is by the relation : t/λi= ln(It/IO) [1] where t is the specimen thickness, It, is the total integrated intensity, and Io is the intensity of the zero-loss peak. The accuracy of this measurement depends on the thickness determination. Model geometries like cubes, wedges, and spheres enable accurate thickness determination from transmission images. Spherical polymers with diameters of order 10-200nm can be made from a number of high-Tg polymers by solvent atomization. This research studied atomized spheres of poly(2-vinyl pyridine) [PVP]. A solution of 0.1% PVP in THF was nebulized. After solvent evaporation during free fall within the chamber atmosphere, solid spherical polymer particles with a range of diameters were collected on holey-carbon TEM grids at the bottom of the atomization chamber.

Maximum-entropy deconvolution applied to electron energy-loss spectroscopy

In the present paper, we investigate the performance of maximum-entropy deconvolution, in removing the instrumental response function from electron energy-loss spectra. To this end we make use of spectra acquired from the carbon K-edge in graphite for a range of signal-to-noise ratios. The zero-loss peak is used as the instrumental profile. The resolution improvement obtained through the application of the deconvolution algorithm as a function of the signal-to-noise ratio is well described by a logarithmic dependency. The claimed resolution improvement is further substantiated by demonstrating the consistency between improvement obtained for the width of the instrumental response function, the width of the π ∗ peak and the splitting of the σ ∗ peaks for a range of signal-to-noise ratios.