Core-loss Spectra Research Articles

Characterization of Ga-Mn Decagonal Quasicrystals in GaAs

AbstractGa-Mn decagonal quasicrystals (DQC), as well as a Ga-Mn approximant and a normal crystal in GaAs are investigated by electron energy-loss spectroscopy (EELS) and energy dispersive X- ray spectroscopy (EDS) combined with Z-contrast imaging. Plasmon peak positions (Ep), full- width-half-maxima (FWHM) and Mn L3/L2ratios of these three phases are derived from their low-loss spectra and core-loss spectra respectively. Mn, Ga and As distributions in ion implanted GaAs layers are characterized by EDS at line-scan mode. These results show that the Ga-Mn DQC has higher Ep and FWHMs than those of its normal crystal counterpart, as well as all other reported QCs. The much larger L3/L2 intensity ratio of the Ga-Mn DQC over that of the Al-Mn icosahedral quasicrystals (IQC) may suggest Mn atoms in the Ga-Mn DQC have much larger local magnetic moments.

Optimization of ELNES processing using a four‐parameter function: application to SiC fibers

In the explanation of EELS spectra in terms of molecular orbitals, where the local band structure is approximated as a linear combination of atomic orbitals of the excited atom and its immediate neighbours, K and L core-loss spectra are decomposed using a four-parameter function that models the ionization continuum in a better way than ‘classical’ step functions. The peaks corresponding to particular chemical bondings are extracted and then fitted by Gaussian functions. The accuracy of such a technique is discussed in the case of Nicalon SiC fibre compared to pure α-SiC. Because of this ionization continuum subtraction we found, in agreement with previous XPS data, that the fibre contains a particular SiCO bonding that is different from SiC and SiO and was not visible without this treatment. Moreover, the CK-edge analysis shows the presence of graphite-like carbon. The existence of CH bonding is also discussed. © 1999 John Wiley & Sons, Ltd.

Optimization of ELNES processing using a four-parameter function: application to SiC fibers

In the explanation of EELS spectra in terms of molecular orbitals, where the local band structure is approximated as a linear combination of atomic orbitals of the excited atom and its immediate neighbours, K and L core-loss spectra are decomposed using a four-parameter function that models the ionization continuum in a better way than ‘classical’ step functions. The peaks corresponding to particular chemical bondings are extracted and then fitted by Gaussian functions. The accuracy of such a technique is discussed in the case of Nicalon SiC fibre compared to pure α-SiC. Because of this ionization continuum subtraction we found, in agreement with previous XPS data, that the fibre contains a particular SiCO bonding that is different from SiC and SiO and was not visible without this treatment. Moreover, the CK-edge analysis shows the presence of graphite-like carbon. The existence of CH bonding is also discussed. © 1999 John Wiley & Sons, Ltd.

Importance of relativistic effect on inelastic scattering cross-sections for quantitative microanalysis

For quantitative microanalysis of specimens using core-loss spectra, elemental maps by electron energy-loss spectroscopy or energy-filtering transmission electron microscopy, accurate values of partial cross-sections are required. Upon increasing the incident electron energy, the relativistic effect including retardation and magnetic interactions on inelastic scattering cross-sections can not be ignored. In the present work, we report the direct observation of the relativistic effect as it appears in the angular distribution of inner-shell electron excitation by using the energy-filtering diffraction technique performed with a high-voltage electron microscope (JEOL-ARM1000). The experimental result confirms a theoretical prediction proposed previously. The importance of the relativistic correction on partial cross-section is discussed in relation to quantitative microanalysis.

Plasmon response and structure of nanocrystalline diamond powder

Abstract High-resolution transmission electron microscopy, electron diffraction and parallel electron-energy-loss spectroscopy are used to analyse nanocrystalline diamond powder. Grains of diameter in the range 2–10 nm were found aggregated together; the grain boundaries were essentially a grossly disordered (amorphous) intergranular phase. Analysis of the plasmon loss function indicated a mass density of 3·30 g cm−1, compared with 3·51 g cm−3 for a chemically vapour-deposited diamond. The core loss spectra showed virtually pure sp3 bonding overall although some exposed surfaces were coated with two or three graphitic layers. Two peaks were observed in the low energy loss function; one at 34 eV was characteristic of the volume plasmon typically observed in crystalline diamond, and a second peak at approximately 23 eV for larger grains shifted to lower energies as the particle size decreased (to 19·5 eV for 2·8 nm diameter) and at the same time increased in intensity, becoming stronger than the volume plas...

Thickness measurement of hydrated and dehydrated cryosections by EELS.

Electron energy-loss spectroscopy (EELS) provides a useful method for determining the thickness of frozen-hydrated and dehydrated cryosections in terms of the inelastic mean free path. Cryosection thickness is an important parameter because plural inelastic scattering limits the sensitivity of elemental microanalysis based on core-loss EELS, and because overlapping structures can affect interpretation of microanalytical data as well as the quality of electron images. The purpose of this work was to establish the minimum practical thickness for cutting cryosections and to explain the measured values for hydrated and dehydrated specimens. Hydrated sections were typically found to be between 1.5-2.5 times thicker than expected from the nominal microtome setting; this difference can be largely explained by compression during cutting. Comparison of micrographs from hydrated and dehydrated cryosections of rapidly-frozen, vitrified liver revealed a lateral shrinkage of approximately 20% on drying. The measured compression and shrinkage factors are consistent with dark-field scanning transmission electron microscopy (STEM) mass measurements on freeze-dried sections. Freeze-dried cryosections, cut to a nominal thickness of 90 nm and supported on thin Formvar/carbon films, had a relative thickness t/lambda i in the range of 0.5 for cytoplasm to 0.9 for mitochondria when analyzed at 100 keV beam energy. Mass loss of approximately 30% occurring at high electron dose enabled useful core-loss spectra to be recorded even from high-mass compartments such as mitochondria without excessive plural scattering.

Thickness measurement of hydrated and dehydrated cryosections by EELS

Electron energy-loss spectroscopy (EELS) provides a useful method for determining the thickness of frozen-hydrated and dehydrated cryosections in terms of the inelastic mean free path. Cryosection thickness is an important parameter because plural inelastic scattering limits the sensitivity of elemental microanalysis based on core-loss EELS, and because overlapping structures can affect interpretation of microanalytical data as well as the quality of electron images. The purpose of this work was to establish the minimum practical thickness for cutting cryosections and to explain the measured values for hydrated and dehydrated specimens. Hydrated sections were typically found to be between 1.5–2.5 times thicker than expected from the nominal microtome setting; this difference can be largely explained by compression during cutting. Comparison of micrographs from hydrated and dehydrated cryosections of rapidly-frozen, vitrified liver revealed a lateral shrinkage of ∼20% on drying. The measured compression and shrinkage factors are consistent with dark-field scanning transmission electron microscopy (STEM) mass measurements on freeze-dried sections. Freeze-dried cryosections, cut to a nominal thickness of 90 nm and supported on thin Formvar/carbon films, had a relative thickness t/λi in the range of 0.5 for cytoplasm to 0.9 for mitochondria when analyzed at 100 keV beam energy. Mass loss of ∼30% occurring at high electron dose enabled useful core-loss spectra to be recorded even from high-mass compartments such as mitochondria without excessive plural scattering. © 1996 Wiley-Liss, Inc.

Some practical consequences of the Lorentzian angular distribution of inelastic scattering

Although inelastic scattering of fast electrons occurs mainly at small angles, a significant fraction involves angles of several milliradians and contributes to the diffuse background in a diffraction pattern; in fact, this electronic contribution predominates over quasielastic (phonon) scattering in the case of very light elements. Chromatic aberration of TEM imaging lenses results in an inelastic point-spread function (PSF) which is similar in shape to the angular distribution of inelastic scattering, giving rise to simple formulas for estimating the spatial resolution in energy-selected images and core-loss spectra. We show that the Lorentzian angular distribution enables Poisson statistics to be applied to energy-loss data recorded with a normal-sized collection aperture, allowing single-scattering distributions to be derived from angle-limited spectra by simple (one-dimensional) deconvolution. Making use of the Lenz model of elastic scattering in an amorphous material, we give approximate expressions for the angular distribution of plural elastic and mixed (elastic + inelastic) scattering and explain why elastic scattering does not invalidate the measurement of specimen thickness from a low-loss spectrum. We describe a procedure for including elastic scattering in quantitative elemental analysis of thick amorphous specimens and demonstrate its validity.

Analyzing line scan EELS data with neural pattern recognition

Neural Pattern Recognition was used for extracting chemical state information from electron energy-loss (EEL) spectra. The purpose was to obtain a quantitative composition profile from sets of low-loss and core-loss EEL spectra measured along a line across an amorphous inclusion at a grain boundary in a silicon bicrystal. The spectra were presented serially to the artificial neural network to obtain the number and shape of the spectra, whose linear combinations reproduce each single spectrum. The results indicate the existence of a different chemical environment at the interfaces between inclusion and crystal. The data analysis proved to be fast, robust, relatively immune to noise or artifacts and capable of extracting relevant information from subtle spectral features.

Transmission EELS of oxide superconductors with a cold field emission TEM

Electron energy loss spectrometry (EELS) with a cold field emission gun (cFEG) transmission electron microscope (TEM) is implemented to analyze the evolution of the electronic structure and dielectric function of oxide superconductors. The O-K core loss spectra of p-type doped oxide superconductors are analyzed in terms of holes formation on oxygen sites, while low loss spectra are analyzed for free carrier plasmas, other spectral excitations, and their crystallographic confinement. It is illustrated that the transmission EELS with a cFEG TEM very much complement soft X-ray absorption spectroscopy and optical spectroscopy, with the added advantages of high spatial resolution (approximately 1-100 nm), and is compatible with other analytical, diffraction, and imaging techniques, which are readily available in a cFEG TEM.

Transmission electron spectroscopy and interferometry of electroceramic oxides

Research on ferroelectricity and related phenomena has come a long way since the discovery of hysteretic nonlinearity of polarization in "Rochelle Salt" by J. Valasek in 1920's. Ever since the discovery, microstructural issues have dominated the field. With the current trends in using ferroelectric and related oxides in nanocrystalline and thin film forms, techniques for microstructural analysis for such specimens are becoming more sophisticated.We are currently investigating nanocrystalline TiO2 as a model system to examine size dependency of electronic structure in nanocrystals, as reflected in their energy loss spectra. Electron beam evaporated nanocrystals of TiO2 (size ranging from 2 - 10 nm) have been examined with transmission EELS with a ~ 0.8 nm electron probe of our HF-2000 FE TEM. The particles were dispersed on a holey carbon grid and those protruding into the holes were analyzed for changes in valence and core loss spectra as a function of nominal particle size. The plasmon loss energies were observed to increase with decrease in particle size.

Investigations of thin hydrogenated amorphous carbon films using electron energy loss spectroscopy (EELS)

Transmission EELS measurements have been carried out on a-C:H films deposited by ion plating from acetone or n-hexane plasma using various substrate potentials. The calculated dielectric functions show low optical absorption in the range of 4 eV to 8 eV. The amount of sp3 hybridizised C atoms has been estimated by integration over the π* and σ* band ranges in the core loss spectra.

Electronic structure and dielectric function of oxide superconductors via transmission EELS with a cold field emission TEM

Transmission high-energy electron energy loss spectrometry (EELS) with a cold field emission transmission electron microscope (TEM) is employed to analyze the evolution of electronic structure and dialectric function of oxide superconductors. The O-K core loss spectra of p-type doped oxide superconductors are analyzed in terms of formation of holes on oxygen sites while momentum-controlled EELS is utilized to identify their crystallographic confinement in YBa 2Cu 4O 8 (Y124). The semiconductor-to-metal transition in Ba 1- x K x BiO 3 (BKBO) is monitored via low-loss EELS. Emergence of free carrier plasmons for x > 0.35 is confirmed using momentum-resolved low-loss EELS. It is argued that transmission EELS with a cold field emission TEM very much complements soft X-ray absorption and optical spectroscopy, with added advantages of high spatial resolution and its compatibility with other analytical and high-resolution TEM techniques readily available in a TEM.

Anisotropic dielectric function and electronic structure of the infinite-layer compound (Sr1-xCax)yCuO2.

The electronic structure and dielectric functions of single crystallites of the infinite-layer compound (Sr[sub 1[minus][ital x]]Ca[sub [ital x]])[sub [ital y]]CuO[sub 2] ([ital y]=0.90, [ital x]=0.3) are investigated by transmission-electron-energy-loss spectroscopy with a cold-field emission-gun TEM. The band gap along the [ital c] axis is estimated to be significantly greater than in the [ital aa] plane, which suggests anisotropic electronic properties. Unlike other cuprates, oxygen core-loss spectra of this compound display two distinct preedge peaks. A close correlation was observed between the valence-band excitations below 10 eV in the [ital aa] plane and the oxygen [ital K] core-loss fine structure. The [similar to]2 eV excitation in the valence-band spectrum is associated with a transition from the lower Hubbard band to the upper Hubbard band and is detected only in the [ital aa] plane. This excitation coincides with the first preedge peak in O [ital K] core-loss spectra, which is assigned to a transition from the O 1[ital s] state to the upper Hubbard band. The results are consistent with the band-structure calculations, for the infinite-layer compound, which indicate strong Cu 3[ital d]--O 2[ital p] hybridization within the [ital aa] plane, but little overlap of the wave function along the [ital c]more » axis.« less

Anisotropy of charge carriers and dielectric function of YBa 2Cu 4O 8 (Y124)

Orientation dependent (momentum transfer resolved) electron energy loss spectrometry (EELS) has been utilized to probe the localization of free carrier plasmons in single crystals of YBa 2Cu 4O 8 (Y124). Two low energy excitations are observed in EELS spectra, one at about 1.4 eV which is the free carrier plasmon,and another excitation at about 3.2 eV. The oscillator strength of the free carrier plasmon is observed to vary considerably with momentum transfer. Maximum oscillator strength of the free carrier plasmon is observed when the momentum transfer ( q) is confined to the ab planes, while the minimum strength is observed for q parallel to the c-axis. Dielectric function calculated using the Kramers-Kronig analysis indicates that the major differences in the dielectric function ofY124 within the ab plane versus along the c-axis are confined only within the first few eVs. The O K core loss spectra exhibit subtle but important deviations from that of YBa 2Cu 3O 7-δ (Y123). Possible EELS spectral signatures of the double chains in Y124 are discussed.

Nickel L2,3 core energy loss studies in NiSi2

The L2 and L3 edges in EELS spectra of 3d transition metals arise primarily from dipole excitation of the 2p1/2 and 2p3/2 electrons to unfilled 3d states. Since the initial states have narrow energy widths, the L2,3 edges reflect the conduction band density of states (DOS) with d character. By comparing the Ni L2,3 edges in NiSi2 to those in pure Ni, the bonding between Ni and Si can be studied.Experiments were carried out in an UHV HB501A STEM operating at 100 keV with a serial EELS spectrometer. The spot size and the current were estimated to be 5 Å and 0.1 nA respectively. Both the convergence and the detector angles were about 11 mrad. The spectra were recorded in steps of 0.25 eV for Ni and 0.5 eV for NiSi2. Long dwell times were necessary to achieve high signal to noise ratio in the core loss spectra and typically 5 sec per energy step was used.

Elemental analysis of telatively thick specimens by EELS

Core-loss spectra of thicker specimens are strongly influenced by plural scattering. Plural inelastic events increase the background Ibunderneath an ionization edge by an amount dependent on t/λ, t being the specimen thickness and λ the total-inelastic mean free path. However, plural scattering also contributes to the integral core-loss signal Ic. In fact, if the latter were integrated over a sufficiently large energy window Δ, the signal/background ratio (SBR=IC/Ib) might be expected to be independent of t.Figure 1 shows K-edge signal/background ratios for elemental carbon and silicon, presented as measured data points (for a collection semi-angle of l0mrad, 120keV incident energy and Δ=100eV) and as solid curves calculated by convolving a power-law edge profile with a delta-function approximation of the low-loss region. For silicon, SBR falls off more slowly with increasing t/λ, as expected from its higher edge energy. This trend is confirmed by measurements on amorphous silicon dioxide depicted in Fig.2, which shows the signal/background ratios of the oxygen and silicon K-edges as a function of thickness.

Transmission electron microscopy of chromatographically purified solid state C 60 and C 70

Solid state forms of pure C 60 and C 70 were investigated at room temperature using transmission electron microscopy (TEM) techniques incorporating electron diffraction, high resolution TEM (HRTEM) and electron energy loss spectrometry (EELS). Electron diffraction patterns recorded along several zone axes indicated that C 60 has a face centered cubic unit cell with Fm3m space group symmetry and lattice parameter of 1.42 nm within 5% experimental error. C 70 appears to adopt close packed hexagonal symmetry (P6 3/mmc space group) with a 0 = 1.01 nm and n 0 = 1.68 nm within 5% experimental error. HRTEM imaging revealed the presence of extensive periodicity associated with C 60 and C 70 crystallites as well as numerous lattice defects and faults. EELS low loss spectra for C 60 and C 70 solids exhibit excitations analogous to graphite and amorphous carbon, such as interband π and π + σ plasmons, though at slightly lower energy losses. C 70 exhibits a more intense “π plasmon” than C 60. The 1s core loss spectra of both C 60 and C 70 show a π* transition at 285 ± 0.5 eV, and splitting of a σ* transition centered at 291 ± 0.5 eV.

Electron-energy-loss study of the TiO2(110) surface.

Electron-energy-loss measurements on the stoichiometric ${\mathrm{TiO}}_{2}$(110) surface which include electronic excitations from both the O 2p valence states and the O 2s, Ti 3p, and Ti 3s core levels have been performed. The observed features have been empirically interpreted within the framework of the molecular-orbital description of an octahedral complex. The effects of surface defects created by ${\mathrm{Ar}}^{+}$-ion bombardment and by the evaporation of a thin Ti overlayer on the energy-loss results have also been studied and were found to cause marked changes in both the valence and core loss spectra. This sensitivity arises from changes in the screening between surface cations that occur when the oxygen-titanium ratio at the surface is varied.

Electronic excitations in the transition metals V, Zr, Nb, Mo, and Ta.

The np core-loss spectra and the associated deexcitation processes for the transition metals V (n=3), Zr, Nb, Mo (n=4), and Ta (n=5) have been studied in low-energy electron scattering experiments in order to investigate the dependence of their nature on the principle quantum number n. For the metals with n=3 and n=4, the measurements are accounted for in terms of atomic transitions involving the excitation and the decay of localized d electrons. The observations on V can qualitatively be compared with published computations for isolated atoms. While the 5p core excitations for Ta resemble the gross features of the unoccupied d-band density of states, the decay processes involve, besides the conventional ${O}_{2}$VV and ${O}_{3}$VV Auger transitions, autoionization emission related to 5${\mathit{p}}^{6}$5${\mathit{d}}^{\mathit{N}}$-5${\mathit{p}}^{5}$5${\mathit{d}}^{\mathit{N}+1}$-5${\mathit{p}}^{6}$5${\mathit{d}}^{\mathit{N}\mathrm{\ensuremath{-}}1}$+\ensuremath{\epsilon}f transitions.