Papers

AVS2001 Session AS-MoA: Quantitative Analysis and Data Interpretation II: Electron Spectroscopies

Monday, October 29, 2001 2:00 PM in Room 134

Monday Afternoon

Start	Invited?	Item
2:00 PM		AS-MoA-1 Satellite Structure in the KLL Auger spectra of Ge L. Kövér, I. Cserny, J. Tóth, D. Varga, Z. Berényi (Institute of Nuclear Research of the Hungarian Academy of Sciences, Hungary) Studying deep core Auger spectra in materials of practical interest is important for understanding solid state effects on Auger processes as well as for revealing the electronic structure of these systems, including localization and charge transfer. Very intense satellites were observed in the KLL Auger spectra of some 3d metals and alloys¹ and in the LMM spectra of 4d metals² in previous studies. An earlier work on the Ge KLL Auger spectra excited by Mo Kα radiation from a thin film of ca 10 nm thickness³ identified the presence of a strong satellite as due to the effect of inelastic electron scattering. Here we report the first high energy resolution measurements of the Ge KLL spectra. The spectra were excited by bremsstrahlung (using an X-ray source with Cu anode) from amorphous Ge overlayers of different, several ten nm thickness, deposited onto a Si substrate. Measurements of the KLL spectra were performed using a high energy hemispherical electron spectrometer.⁴ For identifying the nature of the satellite structure and the contributions from extrinsic and intrinsic processes, the spectra are compared to the corresponding energy loss spectra of backscattered electrons and the shape of the KLL spectra is analyzed using different models for inelastic background correction. This work was supported by the project OTKA T026514. ¹ ¹ L. Kövér, Zs. Kovács, J. Tóth, I. Cserny, D. Varga, P. Weightman, S. Thurgate, Surface Sci. 433-435(1999)833. ² G.G. Kleiman, S.G.C. De Castro,R. Landers, Phys. Rev. B49(1994)2753. ³ E. Sokolowski, C. Nordling, Arkiv för Fysik 14(1958)557. ⁴ L. Kövér, D. Varga, I. Cserny, J. Tóth, K. Tökési, Surf. Interface Anal., 19(1992)9.
2:20 PM		AS-MoA-2 Optimized Analysis of Spectra: Application of Reciprocal-space Approaches to Broad, Sparse, and/or Multistructured Spectra D.E. Aspnes (North Carolina State University); S.D. Yoo (Serome Ventures, Inc., Seoul, Korea) Using a procedure that we recently developed to remove endpoint-discontinuity artifacts in reciprocal-space (R-S) analysis of optical spectra, we solved the optimization problem for determining critical-point (CP) parameters for isolated structures and found that parameters such as CP energies could be obtained for a wide class of direct-space lineshapes even if the lineshapes were not known a priori.¹ Here, we extend this work to the analysis of spectra that contain structures that are broad, such as those encountered in photoemission, that are sparse (those that are represented by only a few data points), such as Raman spectra obtained with array detectors, or that contain a multiplicity of CP structures, such as those encountered in optical spectra at higher energies. For the multiple-CP case the advantages of R-S analysis are retained. We obtain a simple expression that describes the effect of overlayers on the CP energies obtained from optical spectra. For sparse spectra and those containing broad structures R-S analysis shows, not surprisingly, that the information content is low and the most efficient approach appears to be to assume a lineshape and perform curvefitting in direct space as is currently done. Extensions of the approach to two or more dimensions, of interest to image processing, are also discussed. ¹S. D. Yoo and D. E. Aspnes, J. Appl. Phys. (in press).
2:40 PM		AS-MoA-3 Quantitative AES and XPS - Databases Test Quantification Validity M.P. Seah, I.S. Gilmore, S.J. Spencer (National Physical Laboratory, UK) For quantitative analysis by AES and XPS we may either use theoretical or experimental sensitivity factors (SFs). In the past, many analysts have used experimental SFs deduced from data for pure elements or have calculated such data. Both of these will lead to erroneous results. It is shown that the correct SFs for analysing homogeneous mixtures are those deduced for a common, or average, matrix.¹ These may be calculated according to easily defined rules.^1,2 Experimental SFs from NPL databases for AES and XPS agree closely with theory apart from one factor that varies from element to element but is the same for AES as for XPS and is independent of the electron emission energy. This factor is attributed to an inadequacy either of the background subtraction method for the experimental peak areas or of the material-to-material dependence of the inelastic mean free path (IMFP) for the theory. A number of improvements have been made in the background subtraction method which now uses data from an angle-averaged REELS database³ instead of the Tougaard method. The background subtracted XPS spectra have intrinsic loss intensities that are consistent with theoretical predictions. Analyses show that the main error may come from the material-to-material dependence of the IMFP but that the new average matrix sensitivity factors avoid this error. Apart from the single factor for each element, which is the same for AES as for XPS, the correlations between experiment and theory exhibit scatters of 8% and 11% for AES and XPS, respectively. These correlations are shown to be valid in AES for K, L, M or N shell Auger electron peaks with kinetic energies above 180 eV and in XPS for all shells except the weak s shells for Z > 20. ¹M P Seah and I S Gilmore, Surf. Interface Anal. 26(1998)908. ²M P Seah, I S Gilmore and S J Spencer, J. Elec. Spectrosc. to be published. ³M P Seah, Surf. Sci. 471(2001)782.
3:00 PM		AS-MoA-4 Surface Excitations of Medium Energy Electrons in Metals and Semiconductors W.S.M. Werner, W. Smekal, C. Tomastik, H. Stoeri (Vienna University of Technology, Austria) Reflection energy electron loss spectra (REELS) have been measured for several metals and semiconductors (Be, Al, Si, V, Fe, Co, Ni, Cu, Ge, Mo, Pd, Te, Ta, W, Au, Pb) in the medium energy range (150-3400 eV) for normal incidence and an emission direction of 60 with respect to the surface normal. The ratio of the number of electrons that induced a surface excitation to the intensity of the elastic peak was extracted from each spectrum providing the total surface excitation probability (SEP). For the nearly free electron materials the results agree reasonably with free electron theory while significant deviations are observed for the other materials. In all cases the SEP is proportional to the time the probing electron spends in the vicinity of the surface. It is generally found that the surface excitation probability decreases with the generalized plasmon energy. A simple predictive formula to estimate the surface excitation probability for medium energy electrons entering or leaving an arbitrary material is proposed.
3:20 PM		AS-MoA-5 The Attenuation Length Revisited A. Jablonski (Polish Academy of Sciences); C.J. Powell (National Institute of Standards and Technology) The attenuation length, and the preferred replacement term effective attenuation length (EAL), have been a source of confusion in surface analysis by Auger-electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS). The EAL is formally defined in terms of the local slope of the emission depth distribution function (EDDF). This definition is relevant to the determination of the depth of a thin marker layer. The EAL is also frequently used as a term to determine the thicknesses of thin overlayer films from measurements of AES or XPS signal intensities. We show that while the formal EAL definition does not correspond conceptually to this important application, numerical EAL values calculated from changes of signal intensities are similar to values obtained from the formal EAL definition for some experimental configurations. The EALs for the marker-layer and overlayer-film applications are what we term "local" EALs since they were derived from the reasonable assumption that the EDDF could be regarded as an exponential function of marker depth or overlayer-film thickness for small ranges of depths or thicknesses. We point out that it is also useful to define "practical" EALs for a wider range of depths or thicknesses. We have made calculations of local and practical EALs for XPS with Mg Kα X-rays in a range of measurement configurations for Si 2s photoelectrons in Si (for which elastic-electron scattering effects are relatively weak) and for Au 4s photoelectrons in Au (for which the elastic-scattering effects are relatively strong). In general, the local and practical EALs for each solid vary with the measurement conditions. Nevertheless, the practical EALs for overlayer- film thickness measurements in conventional XPS experiments do not vary appreciably with emission angle for emission angles between 0 deg and about 65 deg with respect to the surface normal. A similar result was found for three Auger transitions in Au.
3:40 PM		AS-MoA-6 Comparisons of Practical Effective Attenuation Lengths and Inelastic Mean Free Paths for Applications in AES and XPS C.J. Powell (National Institute of Standards and Technology); A. Jablonski (Polish Academy of Sciences) The practical effective attenuation length (EAL), defined in terms of changes in signal-electron intensities, is needed (as the "lambda parameter") for measurements of overlayer thicknesses by AES and XPS.¹ We have calculated practical EALs for principal photoelectron and Auger-electron lines in Si, Cu, Ag, and W using an algorithm based on solution of the kinetic Boltzmann equation within the transport approximation.² These calculations were made for values of the angle of emission α between 0 and 80° and, for XPS, a configuration in which the angle between the x-ray source and the analyzer axis was 55°. For α < 60°, the ratio of the practical EAL to the corresponding inelastic mean free path (IMFP) is nearly constant with emission angle, and varies in magnitude between 0.68 and 0.91. The deviation of this ratio from unity is due to the effects of elastic-electron scattering. We find that the ratio varies approximately linearly as a function of the single-scattering albedo that is defined in terms of the IMFP and the transport mean free path (TMFP). While the albedo varies in a complicated manner with atomic number and electron energy, it provides a convenient measure of the effects of elastic-electron scattering. For α > 60°, the ratio of the practical EAL to the IMFP increases rapidly with emission angle. At these more grazing emission angles, practical EALs should be determined for the specific film thicknesses and emission angles of interest in order to obtain reliable measurements of film thicknesses. ¹A. Jablonski and C. J. Powell (to be published). ²I. S. Tilinin, J. Zemek, and S. Hucek, Surf. Interface Anal. 25, 683 (1997).
4:00 PM		AS-MoA-7 Surface Analysis of Hafnium Compounds by XPS using High Energy Zr Source P. Mrozek, D.F. Allgeyer, B. Vaartstra (Micron Technology, Inc.) The deep 3d5/2 core level at Eb = 1662 eV of Hf in a metallic state, as HfO2, and as HfSi2 were investigated by X - ray photoemission using the Zr LÎ± source. The study was undertaken for the sake of quantitative analysis of hafnium thin films used as common diffusion barriers.¹ Chemical shifts of Hf 3d5/2 were determined together with Auger parameter values for M3N6,7N6,7 (Ek = 1615 eV) Auger electron transition and used to probe the evolution of chemical bonding in hafnium films. The effect of initial state from the configuration interaction within the Hf 4f shell onto the quantitative analysis was examined. Common spectral overlaps of the Auger electron and photoelectron lines in hafnium compounds were discussed. ¹Ken-ichi Yoshimoto, Satoko Shinkai, Katsutaka Sasaki, Jpn. J. Appl. Phys. 39 (2000) 1835.
4:20 PM		AS-MoA-8 VAMAS TWA2 Project A2, Evaluation of Static Charge Stabilization and Determination Methods in XPS on Non-conducting Samples: Report on an Inter-laboratory Comparison W.E.S. Unger, Th. Gross, O. Boese, A. Lippitz, Th. Fritz (Bundesanstalt fuer Materialforschung und -pruefung (BAM), Germany); U. Gelius (Uppsala Universitet, Sweden) Results of an inter-laboratory comparison (27 participants in Europe, Japan and USA) of XPS data obtained with non-conductive samples are presented. Binding energies of Al 2s for alumina, N 1s and imide C 1s for Kapton and Sr 3p3/2 for a strontium titanate film on glass were obtained after static charge referencing with the help of 15 nm gold particles deposited at the surface of the test samples. For the alumina sample C 1s static charge referenced data are presented, too. In any case repeat standard deviations, between standard deviations, reproducibility standard deviations and total means are evaluated from the experimental data. It can be stated, that though the repeat standard deviation is small as 0.05 eV in the best case, the standard devia-tion characterizing the reproducibility of the method is obviously not better than 0.15 eV, also in the best case, at the present time. The knowledge of these standard deviations is important for metrology, validations of analytical procedures relying on qualitative photoelectron spectroscopy and XPS databanking.
4:40 PM		AS-MoA-9 Real and Gedanken Experiments Related to Surface Charging during XPS Measurements of Insulating Materials and Thin Films D.R. Baer, M.H. Engelhard, Y. Liang, A.S. Lea, D.J. Gaspar, C.F. Windisch (Pacific Northwest National Laboratory) The ability to accurately obtain binding energy measurements can be important for interpretation of XPS data for many different applications. In recent studies of thin oxidized aluminum layers we have seen a range of phenomena that both confound easy data analysis and provide opportunities to learn critical information about oxide layers. In this paper, we will examine the binding energies obtained from oxidized aluminum formed in aqueous solutions and transferred to vacuum without exposure to air. Data from pure aluminum and copper containing aluminum alloys will be compared with each other and to other data in the literature. Influence of ion and electron bombardment and oxide doping will be presented. A more general examination of some issues related to understanding binding energies from oxides and oxide films will be discussed. A summary of some physical phenomena that occur during XPS measurements of thin insulating layers on conducting or semi-conducting substrates that can complicate charge referencing will be given. Some classic experiments of grounded and ungrounded specimens will be reported for a new generation of instruments with advanced methods of charge control and illustrative data from other systems will be reported.
5:00 PM		AS-MoA-10 Core-level Photoemission of Zirconium and Hafnium Silicates for Use as High Dielectric Oxides R.L. Opila, R.B. Van Dover, G.D. Wilk (Agere Systems) Photoelectron spectroscopy is often used to characterize the composition and bonding in high dielectric constant materials being developed for the next generation of integrated circuits. However, the peak positions vary continuously as a function of composition and thickness for binary oxides. Thus we have studied the core-level photoemission spectra of hafnium silicates and zirconium silicates, sputter deposited on silicon, as a function of oxide thickness. The metal peak shifts by as much as 1 eV to higher energy as the silicate component is increased. The Si 2p peak also shifts to greater binding energy as silicate component is increased. For pure metal oxide, the O 1s shows two components corresponding to the hydroxide and oxide at higher and lower binding energies respectively. As the silicate component is increased a new component appears at even higher binding energy. In addition to all of these peak shifts, for a given composition, a thin layer (less than 10 nm) has lower binding energies than seen for a corresponding thick film. All of these shifts in metal core-level, Si 2p, and O 1s can be related to two phenomena. The first is the relative electron donation of the metal and Si. Since the metal easily donates its valence electrons to oxygen, the O 1s peak is found at lower binding energies. As the Si component is increased, less electron density is donated to the oxygen, shifting the O 1s to greater binding energies. The oxygen then withdraws slightly more electron density from the metal, shifting its core levels to high binding energy. This reasoning can be used to explain all of the peak shifts seen for a given thickness. Thin films have lower binding energy because electrons near the Fermi energy help screen the positively-charged photoemission final state. This effect has also been seen for thin silicon dioxide films grown on silicon.