摘要:

Contrast of scanning ion microscope images compared with scanning electron microscope images for metals† Y. Sakai,a T. Yamada,a T. Suzukia and T. Ichinokawab aJEOL Ltd., 3-1-2 Musashino, Akishima, Tokyo 196, Japan bDepartment of Applied Physics, Waseda University, 3-4-1 Ohkubo, Shinjuku-ku, Tokyo 169, Japan Received 2nd September 1998, Accepted 2nd November 1998 The contrast of secondary electron images in scanning electron microscopy (SEM) is compared with that in scanning ion microscopy (SIM) for metals.The dependence of the secondary electron yield on atomic number in SEM is opposite to that in SIM. The secondary electron yields for electron bombardment increase with atomic number for metals, whereas those for ion bombardment decrease with increasing atomic number. The origin of these phenomena is discussed on the basis of the range profiles of these particles with respect to the surface. yields; hence, negative particle imaging is essentially electron 1 Introduction imaging.The focused ion beam (FIB) has been applied to failure analysis, but more recently FIB has been used either as a 2.2 Sample sample preparation tool or as a stand alone component of an The sample was a parallel strip of metal films of 1 mm thickness integrated failure analysis environment, operating as a secondand 0.1–1 mm width deposited on a clean Si(100)2×1 surface ary ion microscope. In scanning ion microscopy (SIM), ionwith the arrangement of the order of atomic number Z2 as induced secondary electron images are used as a monitor shown in Fig. 1, i.e. Al (Z2=13), Cu(29), Ag(47), and Au image together with secondary ion images, because of the high (79), which have a similar electronic structure. The sample emission yield, high topographic sensitivity, channeling conwas fixed onto the sample holder of the microscope ex situ of trast, and better material contrast. However, systematic studies the sample holder of the deposition chamber.The contami- on the ion-induced secondary electron yield for various targets nation and oxide layers were sputtered by Ar+ ion bombard- to explain the SIM images have not yet been conducted. ment in an ultra-high vacuum to obtain a clean surface and Therefore, detailed experiments on contrast formation of were checked by Auger electron spectroscopy. The contrast of secondary electron images as a function of atomic number Z2 the secondary electron images of the deposited metal films was were carried out under the same experimental conditions.The not changed during observation by SIM with primary beams systematic contrast changes of the secondary electron images of Ga+ and Ar+ ions. with Z2 for scanning electron microscopy (SEM) and SIM were observed by focused electron, Ga+ and Ar+ ion beams. 2.3 Procedure The energy spectra of the true secondary electron emission by electron and Ar+ ion bombardment were measured to confirm The sample was observed by both microscopes to check the the contrast of SEM and SIM images depending on Z2.The contrast of the secondary electron images as a function of the diVerent features between SEM and SIM images are presented atomic number Z2. The contrast changes with Z2 were conand the origin of this diVerence is discussed based on the firmed by the secondary electron spectra measured by the diVerent range profiles of these particles with respect to hemispherical electron energy analyzer of the JAMP-7800F the surface.for incident beams of electrons and Ar+ ions. Since the probe size of the focused Ar+ ion beam was approximately 0.1 mm 2 Experimental 2.1 Apparatus The apparatus used included a scanning Auger electron microscope (JAMP-7800F) with a vacuum pressure of 5×l0-10 Torr installed with a hemispherical electron energy analyzer, a secondary electron detector of a scintillator and photomultiplier system and an Ar+ ion gun with a probe diameter of 0.1 mm and a scanning Ga+ ion microscope (Micrion-9000) with an annular micro-channel plate detector mounted coaxially with the incident ion beam to be biased to detect either positive or negative particles.Secondary electron yields are, in general, far larger than secondary negative ion Fig. 1 The sample is a parallel strip of films of Al, Cu, Ag, and Au †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.deposited on the clean surface of Si(100). J. Anal. At. Spectrom., 1999, 14, 419–421 419and the whole range of view of the image was larger than 5×5 mm2, the secondary electron image obtained by Ar+ ion bombardment was observed by mechanically scanning the specimen by means of step motors in order to avoid a deviation error from the focusing point of the hemispherical analyzer. The primary energies of the electron, Ga+ and Ar+ ion beams were 10, 30, and 3 keV, respectively. 3 Results Fig. 2(a) shows a secondary electron image observed with the JAMP-7800F at an incident electron energy of 10 keV. The brightness increases with atomic number, although Al(13) is brighter than Si(14). Fig. 2(b) shows a secondary electron image observed with the Micrion-9000 with a Ga+ ion beam at 30 keV. The brightness decreases with increasing atomic number unlike the SEM image. Fig. 2(c) is a secondary electron image observed with the JAMP-7800F with an Ar+ ion beam at 3 keV.The brightness decreases with increasing atomic number as was found with the Ga+ ion beam. Such contrast changes in the secondary electron images were quantitatively checked by measurement of the true secondary electron spectra for electron and Ar+ ion bombardment. Fig. 3(a) shows the electron energy spectra measured in the energy range 0–2500 eV for electron beam bombardment at 10 keV. A number of Auger peaks, viz., KLL, LVV, LMM, MNN, MVV, and NVV, were observed depending on the metal.Fig. 3(b) shows the true secondary electron spectra in the energy range 0–300 eV obtained by electron bombardment for samples of Au, Ag, Cu, Al, and Si. In this case, the Fig. 3 Electron emission spectra for electron and Ar+ ion bombardment at 10 and 3 keV, respectively. (a) High energy range for electron bombardment, (b) and (c) low energy range for electron and Ar+ ion bombardment, respectively (K.E.=kinetic energy of secondary electrons).The orders of magnitude of the secondary electron yields agree with those of SEM images. intensities of the sharp Auger peaks are negligible compared with those of the large background owing to true secondary electron emission. The peak height of the true secondary electron spectra increases with atomic number and corresponds to the contrast of the secondary electron images, although an exception between metals and semiconductors is observed. Fig. 3(c) shows the secondary electron spectra obtained by Ar+ ion bombardment at 3 keV by using the sputtering ion gun.The spectra are simple compared with those from electron bombardment, and the Auger electron spectra due to core electron excitations are not observed. This fact is attributed Fig. 2 (a) Secondary electron image taken by the incident electron to the slow velocity of the incident ions compared with those beam at 10 kV. (b) Secondary electron image taken by a Ga+ ion of the orbital electrons in the target atoms.The true secondary beam at 30 kV. (c) Secondary electron image taken by a focused Ar+ electron yields decrease with increasing atomic number, con- ion beam at 3 kV. The atomic number dependence of the secondary electron yield in (b) and (c) is opposite to that in (a). sistent with the result of the SIM image. From these results, 420 J. Anal. At. Spectrom., 1999, 14, 419–421Fig. 4 Dependence of the secondary electron yield on Z2 for electron and Ar+ ion bombardment. The values on the ordinate are in arbitrary units, but the secondary electron yields were quantitatively measured from the secondary electron spectra in the range 0–100 eV.it was found that the atomic number dependence of the Fig. 5 Examples of computed ranges as a function of M2/M1. The secondary electron yield is opposite for SEM and SIM of vertical lines represent the ‘surface’ in the computer model, hence the metals. Fig. 4 shows the quantitative relationship of secondary shaded regions may be interpreted as ‘ion reflection’.Rp is a projected electron yield with atomic number for electron and Ar+ ion range. The range profile approaches the surface with increasing bombardment. In Fig. 4, the yields were measured in the M2/M1. energy range 0–100 eV and are given in arbitrary units. surface with increasing M2/M1. In Fig. 5, Rp is a projection 4 Discussion of the total ion range to the surface normal. The total ion range is a sum of the projected range and the lateral range R The secondary electron yield ce for electron bombardment is (i.e., range straggling).Approximate distributions of the total given by Baragiola et al.1 and is as follows: ion range were provided4 as a function of M2/M1 in units of Rp. The loss of secondary electron emission due to reflected ce= P(x) J P2 0 exp A- x cos h l BAdE dxBe dx (1) ions increases with M2/M1; therefore, the secondary electron yield for ion bombardment decreases with atomic number.On where (dE/dx)e is the electronic stopping power, P(x) is the the other hand, for electron bombardment it should be remem- probability of secondary electron production as a function of bered that the range profile lies far away from the surface and the depth x, J is the average ionization energy, and the the back-scattered electrons contribute eVectively to the sec- exponential term is an attenuation factor for secondary elecondary electron yield. trons from the bulk to the surface.The electronic stopping power is given by the Bethe–Bloch equation and is almost independent of atomic number. J increases with atomic 5 Conclusion number and deviates in a group of the Periodic Table. A It has been found that the atomic number dependence of the cascade process for secondary electron production is included secondary electron yield in ion bombardment is opposite to in P(x); P(x) increases with atomic number due to the eVect that in electron bombardment. This is attributed to the diVer- of back-scattered electrons.The analytical estimation of ce is ent range profiles regarding the surface for electron and ion diYcult, but a Monte-Carlo simulation is the best way to bombardment and to the loss of incident ions by surface reveal the Z2 dependence of the secondary electron yield. reflection which aVects the secondary electron yield as a For ion bombardment, the secondary electron yield ce is function of Z2.expressed also by eqn. (1). In eqn. (1), (dE/dx)e is the electronic stopping power due to inelastic scattering. For ion bombardment, the nuclear stopping power (dE/dx)n due to Acknowledgement elastic scattering should be taken into account. Both values are comparable to the ion energies used here and are several The authors thank Professor J. Kirschner of the Max-Planck tens of eV per A° . The electronic stopping power is proportional Institute for Microstructure Physics (Halle/Saale) for helpful to the secondary electron yield and was calculated by using discussions.the Lindhard equation.2 The calculated values, however, are almost independent of Z2. Therefore, the opposite dependence References on atomic number between electron and ion bombardment should be explained by another factor rather than the 1 R. A. Baragiola, E. V. Alonso and A. Oliva-Florio, Phys. Rev. B, stopping power. 1979, 19, 121. The three-dimensional range distributions in the ion 2 J. Lindhard and M. SharV, Phys. Rev., 1961, 124, 128. 3 H. E. Schiøtt, K. Dan. Vidensk. Selsk. Mat-Fys. Medd., 1965, 35, bombardment were calculated by Schiøtt3 and correspond to No. 9. the implant atom distributions in the target. Fig. 5 shows the 4 Ion Implantation, Sputtering and their Applications, ed. P. D. range profiles computed by Lindhard and SharV2 with respect Townsend, J. C. Kelly and N. E. W. Hartley, Academic Press, to the surface as a function of the mass ratio, M2/M1. In London, 1976, ch. 2. Fig. 5, the vertical lines represent the ‘surface’ in the computer model; hence, the shaded regions may be interpreted as ‘ion Paper 8/06833J reflection’. The peak of the range profile approaches the J. Anal. At. Spectrom., 1999, 14, 419–421 421

ISSN:0267-9477    

DOI:10.1039/a806833j

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

SHORT PAPER Application of carbon profiling using WDX spectrometry in the SEM to failure analysis in carburized gear teeth† Tulin Hidayetoglu Eaton Corporation Innovation Center, 26201 Northwestern Highway, Southfield, Michigan 48037, USA Received 19th October 1998, Accepted 27th January 1998 The depth and carbon levels of carbon profiles are a major concern in carburized components such as transmission gears. This paper describes a case study where carbon levels of carbon profiles were determined using wavelength dispersive X-ray (WDX) spectrometry in the scanning electron microscope (SEM).cation of 5000×. Fig. 1 illustrates the calibration curve for 1 Introduction SAE 86XX gear steel. The nominal composition of SAE 8620 Today, the demand for higher load carrying capacity from (core material ) is presented in Table 1. The case material is truck transmissions is more intense than ever. If heat treated martensitic at the surface, and a mixture of martensitic and and finished properly, carburized steel gears oVer the desired bainitic further away from the surface with increasingly more load carrying capacity.The depth of penetration of carbon bainitic. The core material is ferritic. into a gear tooth is aVected by the carbon potential of the To determine carbon profiles in unknown samples such as atmosphere, the temperature in the carburizing furnace and gear teeth, a cross-sectional plane of the unknown is polished the composition of the gear steel.To achieve high quality down to 1 mm finish. Carbon X-ray counts per second are carburized gears, not only case depth, case microstructure and collected at 25 mm increments up to a depth of 100 mm, then case hardness have to be controlled, but also the finishing at 50 mm increments up to a depth of 500 mm, and at 100 mm processes.1–6 Carbon profiling, defined as measuring carbon up to a depth of 2500 mm on this cross-sectional plane. The gradient in carburized components, is done to determine the X-ray counts are then converted to % carbon using the eYciency of a carburizing furnace, the eYciency of a carburiz- calibration curve for the steel.Finally, the carbon profiles are ing cycle and the accuracy of the CBN (cubic boron nitride) plotted using % carbon as a function of depth from gear tooth grinding process. surface. A more in-depth description of the procedure is given in reference 7. 2 Procedure 3 Application The first step in measuring carbon profiles using wavelength dispersive X-ray spectrometry in the SEM is to construct a The procedure described above was applied to identifying the calibration curve.This is necessary because the X-rays of light root cause of uneven wear around a transmission gear subelements, such as carbon, tend to be absorbed not only by the jected to a dynamometer test. A post-mortem analysis was specimen chamber of a scanning electron microscope but also conducted to determine the eVectiveness of the carburizing by the sample itself, due to their long wavelengths and low process, and the eVects of CBN (cubic boron nitride) grinding energies.A set of control samples with known uniform, on the carburized layer. nominal composition covering the range of carbon content of interest has to be prepared for each given gear steel. Next, carbon X-ray counts per second (cps) on each control sample in the set is determined using a wavelength dispersive X-ray spectrometer in the SEM.Carbon X-ray cps plotted as a function of the carbon content of each control sample then constitutes a calibration curve for that steel. A Leo S-360 (Thornwood, New York, USA) scanning electron microscope with an Oxford/Microspec (Concord, Massachusetts, USA) wavelength dispersive X-ray spectrometer was used in this study. The wavelength dispersive X-ray spectrometer was equipped with an LSM 80 detecting crystal. The scanning electron microscope is fitted with a rotary pump, a turbomolecular pump, as well as an ion pump for the gun unit.The working distance was 25 mm (as recommended by the equipment manufacturers). The analysis was run at an accelerating voltage of 10 kV, a probe current of 300 nA, and a magnifi- Fig. 1 Calibration curve constructed for SAE 86XX series alloys using †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.wavelength dispersive X-ray spectromety in the SEM. J. Anal. At. Spectrom., 1999, 14, 423–424 423Table 1 Nominal chemical composition of SAE 8620 C Mn P(max) S(max) Si Cr Ni Mo 0.18–0.23 0.70–0.90 0.035 0.040 0.15–0.30 0.40–0.60 0.40–0.70 0.15–0.25 4 Discussion Fig. 2 illustrates the carbon profiles measured on both teeth for the entire range of depth from the surface chosen for this study. The two profiles overlap when the profile for tooth #38 is moved about 50 mm in the positive x-direction, as shown in Fig. 3. Given the fact that the measurements were done on the coast side, this diVerence can result either from inadequate heat treatment or CBN (cubic boron nitride) grinding. Since uneven wear was not limited to a block of consecutive gear teeth, but spread around the gear, the possibility of inadequate heat treatment was eliminated. Therefore, the root cause of uneven wear around this gear tooth was identified as inadequate CBN (cubic boron nitride) grinding that removed about 50 mm of extra stock material from some teeth.Excessive removal of the carburized layer left some of the teeth more Fig. 2 Comparison of carbon profiles of tooth numbers 33 and 38 for vulnerable to damage, as demonstrated by the dynamometer the entire range of depth from the surface chosen for this study. test. 5 Conclusion It was concluded that the root cause of uneven and excessive wear around a gear tested in a dynamometer was inadequate CBN (cubic boron nitride) grinding.References 1 G. F. Bastin and H. J. M. Heigligers, in Electron Probe Quantitation, ed. K. F. J. Heinrich and D. E. Newbury, Plenum Press, New York, USA, 1991, pp. 145–161. 2 R. E. Smith, Quality Gear Inspection-Part I, Gear Technology, 1994 (September/October), pp. 32–38. 3 R. F. Kern, Controlling Carburizing for Top Quality Gears, Gear Fig. 3 Comparison of carbon profiles of tooth numbers 33 and 38 Technology, 1993 (March/April ), pp. 16–21. with the profile for tooth no. 38 moved about 50 mm in the positive 4 C. Kakes and J. Wilde, How to Carburize a Finished Gear, Gear x-direction. Technology, 1995 (March/April ) pp. 26–27. 5 N. Bartels and W. R. Stott, Gear Heat Treating in the 90s, Gear Technology, 1995 (March/April ), pp. 18–19. Two gear teeth were selected, one worn excessively (tooth 6 J. Theissen, High Power Transmission with Case-hardened Gears and #38) and one worn minimally (tooth #33). Both teeth were Internal Power Branching, Gear Technology, 1985 (January/ February), pp. 35–41. sectioned, mounted and polished to 1 mm finish. Carbon 7 T. Hidayetoglu, Carbon Profiling Using Wavelength Dispersive profiles were measured on the coast side as described above. X-ray Spectrometry, JSAE Spring Convention Proceedings 984, Measurements were done on the coast side in order to work May 1998, pp. 129–132. with untested, fresh surfaces. The measured carbon profiles are presented in Fig. 2 and 3. Paper 8/08105K 424 J. Anal. At. Spectrom., 1999, 14, 423–424

ISSN:0267-9477    

DOI:10.1039/a808105k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Electron probe X-ray microanalysis of glass–ceramic carbon electrode surfaces modified by copper, cadmium and lead electrochemical co-deposition† Nikolai V. Alov* and Kirill V. Oskolok Department of Analytical Chemistry, Faculty of Chemistry, Lomonosov Moscow State University, Moscow 119899, Russia Received 28th August 1998, Accepted 5th November 1998 Glass–ceramic carbon electrode surfaces modified by electrochemical co-deposition of copper, cadmium and lead from aqueous solutions were studied by electron probe X-ray microanalysis in combination with X-ray photoelectron spectroscopy and X-ray fluorescence analysis. The metal distribution on the disc electrode surface was calculated.The dependence of the element composition of the modified disc electrode surface on the absolute and relative metal concentrations in solution, the electrode diameter (cathodic current density accordingly) and the type of intermetallic interaction was established. The results of the spectroscopic investigation were compared with those predicted by theoretical electrochemistry.On the basis of the spectroscopic data, mechanisms of metal co-deposition are proposed. Hitherto, the electrochemical co-deposition of metals on for- surfaces were measured with a Leybold LHS-10 electron spectrometer. XRF analyses were performed with a eign substrates has been a little-studied section of theoretical electrochemistry. The complicated mechanism of this process SPECTROSCAN X-ray fluorescence spectrometer.The conditions of the spectroscopic study are described in detail has made an adequate theoretical description diYcult.1 The study of such processes has a great practical significance for elsewhere.9 electrochemistry and electroanalytical chemistry. Scanning Theory of EPMA of solid disc electrode surfaces tunneling microscopy has been widely used to study the early stages of individual metal deposition from aqueous solutions The surface of a disc electrode is divided into m concentric with high metal concentrations (10-2–10-1 mol l-1) on nano- zones of analysis.The coordinate of the zone and analysis metre-sized areas of monocrystalline electrode surfaces.2,3 ‘point’ accordingly is determined by the equation: However, this method is unsuitable for the study of the electrochemical deposition and co-deposition of metals on real ri= (2i-1)R 2m (polycrystalline and rough) electrode surfaces from solutions with metal concentrations below 10-3 mol l-1.For this purwhere i is the zone number and R is the electrode radius. pose, the application of X-ray and electron spectroscopic EPMA is performed along k arbitrary radii (m analysis ‘points’ methods with diVerent lateral and depth resolutions is highly per radius). The magnitude of the electron scan is eVective.4–7 The aim of this work was to study glass–ceramic 200×200 mm. The average specific element content in each carbon electrode surfaces modified by electrochemical zone is: co-deposition of copper, cadmium and lead from aqueous solutions by electron probe X-ray microanalysis (EPMA) in combination with X-ray photoelectron spectroscopy (XPS) Ci= .k j=1 Cij k Km i=1 and X-ray fluorescence (XRF) analysis. Experimental where Cij is the result of ‘point’ analysis and j is the radius Sample preparation number. The average specific element content on the electrode surface (‘integral’ analysis) is: Metal deposition was eVected from n×10-4 M solutions of Cu, Cd and Pb nitrates in 0.01 M HNO3 on the surface of Cint=.m i=1 aiCi glass–ceramic carbon (GCC) electrodes (diameter 4 or 10 mm) under solution mixing conditions. GCC is an advanced type where ai is the coeYcient of statistical weight of a zone, which of glassy carbon with excellent electrochemical and mechanical is determined as: properties for anodic stripping voltammetry.8 The electrolysis potential was -1000 mV and the electrolysis time was 300 s.ai= i2-(i-1)2 m2 Km i=1 The deposition technique is described in detail elsewhere.9 Spectroscopic equipment For distributive EPMA, it is necessary to determine the reduced average element content for each analysis zone: The EPMA data were obtained with a CAMEBAX microbeam electron probe X-ray microanalyzer. XPS profiles of electrode Ci,reduced= aiCi Cint Km i=1 †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.Finally, a graph of Ci,reduced versus (r/R) is plotted. J. Anal. At. Spectrom., 1999, 14, 425–428 425uniform than lead distribution. This conclusion is in good Results and discussion agreement with our experimental results (Fig. 1). Metal deposition The nucleation rate is mainly defined by the magnitude of the overpotential:13,14 Copper deposition on a GCC electrode surface results in the formation of a uniform thin film.During lead deposition, J~exp(DGC(g)/kT) (4) cloverleaf-shaped three-dimensional crystallites are formed. The cathodic current density is proportional to the metal Individual cadmium deposition does not proceed to any signucleation rate, at least in the early stages of deposition. The nificant extent. Recently, we showed that lead deposition and primary distribution of the current density (i) on a disc the early stages of copper deposition on GCC electrode electrode surface is expressed by the equation: surfaces proceed by a Volmer–Weber mechanism (three-dimensional island growth) with the prevailing lateral growth of i/iav=0.5(1-r2/R2)-0.5 (5) copper nuclei.7,10 The amount of metal deposited increases where iav is the average current density, R is the electrode with increasing metal concentration in solution and decreasing radius and r is the distance from the electrode center.11 Thus, electrode diameter (an increase in cathodic current density the current density increases from the center to the electrode consequently). According to EPMA data, metal contents edge, which will result in an increase in the amount of metal increase from the center to the electrode edge; copper distrideposited from the center to the electrode edge.This fact is in bution on a GCC disc electrode surface is more uniform than good agreement with our experimental data (Fig. 1). lead distribution (Fig. 1). The reasons for these phenomena can be described as Copper and lead co-deposition follows.A nucleation process is characterized by an energy of critical nucleus formation: During co-deposition the amount of metal on the GCC electrode surface is several times less than that for individual deposition according to XRF and EPMA data. This fact DGC=f (Ds, s, rat, z, g) Ds=s+sS-C-sS-S (1) indicates a change in the process kinetics. Indeed, during copper and lead co-deposition on GCC electrode surfaces a where s is the specific surface free energy of the nucleus, sS-C ‘mechanical mixture’ alloy is formed.More than 90% of the and sS-S are the energy of formation of the ‘substrate–crystal’ copper and lead are deposited independently. However, the and ‘substrate–solution’ interfaces, respectively, rat is the radius mutual influence of the metals is large. Copper suppresses lead of the metal atom, z is the charge of the metal ion and g is co-deposition (Table 1). the overpotential.11 The first four physical parameters are The above data are in good agreement with the electrochemi- defined by the nature of the metal, substrate and solution.cal concept of metal co-deposition.11 Indeed, copper and lead During three-dimensional lead island formation: do not form intermetallic compounds and their mutual solu- DG3~s2Dsrat2/z2g2 (2) bility is slight (no more than 2–4%).15 Hence, deposition of the metals has to proceed independently. As the number of The later stages of copper deposition proceed by the active centers is limited, competitive nucleation proceeds, Frank–van der Merwe mechanism ( layer-by-layer growth):10,12 which causes a change in the crystallization kinetics.Since the DG2~s2rat2/zg (3) overall amount of metal deposited is proportional to the number of nuclei, the amount deposited is smaller for Unlike copper (DsCu<0), lead adhesion to the substrate is co-deposition. In addition, as the fixation of the electrolysis weak (DsPb>0).The radius of the lead atom (0.175 nm) is potential is not an instantaneous process, copper nucleation considerably greater than that of the copper atom (0.128 nm). (E°Cu2+/Cu0=338 mV) starts earlier than lead nucleation Thus, the work (energy) of lead nucleation is greater than that (E°Pb2+/Pb0=-126 mV). This phenomenon, in combination of copper nucleation. Therefore, in the former case the influwith a process of interaction of metallic lead with copper ions ence of factors that decrease this work [magnitude of overpotin solution, results in the suppression of lead deposition in the ential—compare eqns.(2) and (3), substrate roughness, presence of an excess of copper in solution.9 impurity adsorption] is particularly significant. Consequently, Although mutual metal solubility is extremely low, copper distribution on the GCC electrode surface will be more supersaturated solid solutions are generated during co-deposition. Gradually, the system transforms to a thermodynamic equilibrium state.This phenomenon is confirmed by the considerable change in the ratio of copper to lead content on the electrodeposited surface with time according to XPS data. To a first approximation, during co-deposition, nucleation and the early stages of metal crystallite growth can be described by classical deposition mechanisms (see above). During the formation of several monolayers of an electrodeposit on the Table 1 Results of spectroscopic analysis of GCC electrode surface modified by copper and lead co-deposition CM/10-4 mol l-1 (in solution) CCu/CPb (on GCC electrode surface) ‘Integral’ XRF XPS Cu Pb EPMA data data data Fig. 1 Characteristic distribution of reduced metal contents on GCC 3 3 0.48 0.49 0.94 disc electrode surface modified by electrochemical deposition or 5 5 1.1 1.1 0.15 co-deposition of copper and lead from aqueous solution according to 5 3 24.4 25.7 9.9 EPMA data: 1, Pb; 2, Cu; 3, quasi-uniform metal distribution (R= 3 5 0.52 0.43 0.51 electrode radius, r=distance from electrode center). 426 J. Anal. At. Spectrom., 1999, 14, 425–428Fig. 3 General view of distribution of copper to lead content ratio on Fig. 2 General view of distribution of copper to lead content ratio on GCC disc electrode surface after electrochemical metal co-deposition GCC disc electrode surface after electrochemical metal co-deposition on electrodes of large (1) and small (2) diameter with equal metal with excess of copper (1) or lead (2) in solution and fixed electrode diameter according to EPMA data (R=electrode radius, r=distance concentrations in solution according to EPMA data (R=electrode radius, r=distance from electrode center). from electrode center).Table 2 Results of spectroscopic analysis of GCC electrode surface electrode surface, four main processes take place: copper modified by copper and cadmium co-deposition deposition on copper film, lead deposition on lead islands, lead deposition on copper film and copper deposition on lead CM/10-4 mol l-1 (in solution) CCu/CCd (on GCC electrode surface) islands. Evidently, the last two processes are responsible for the diVerence between metal deposition and co-deposition.‘Integral’ XRF XPS The correlation of XPS and XRF data indicates that the rela- Cu Cd EPMA data data data tive lead content on the surface of the electrodeposit 1 1 40 42 53 (2–3 nm) is significantly greater than in the surface layers 3 3 132 125 39 (1–2 mm)7,9,10,12 (Table 1).Theoretically, and by scanning 5 3 888 937 30 tunneling microscopy, it was shown that copper nucleation on 3 5 80 72 16 lead substrates does not proceed by the classical Volmer–Weber mechanism. Three-dimensional islands of copper are incorporated in the substrate and are covered by a lead monolayer.16 Cadmium distribution has a maximum. The higher the relative Lead deposition on copper substrates proceeds by the copper content in solution, the nearer is the distribution Stranski–Krastanov mechanism (monolayer+three-dimen- maximum to the electrode center (Fig. 4). The maximum is sional island growth). Consequently, both processes result in less pronounced for electrodes of small diameter. a relative lead enrichment of the surface of the electrodeposit The reasons for these phenomena can be described as in comparison with the surface layers. follows. During co-deposition, three competitive processes It was established that, during copper and lead proceed on the surface of the electrodeposit.The first process co-deposition, the amount of metal deposited increases from is cadmium adsorption on the copper film surface. The rate the center to the electrode edge; copper distribution is also of this process (vad) is proportional to the cathodic cadmium more uniform (Fig. 1). The copper to lead content ratio current density (iCd2+): decreases from the center to the electrode edge during deposvad( r/R)~iCd2+ (r/R) ition on GCC electrodes of large diameter (10 mm) and with an excess of copper in solution.On GCC electrodes of small diameter (4 mm) and with an excess of lead in solution, the distribution of the metal content ratio along the electrode radius is proportionate (Fig. 2 and 3). Evidently, this phenomenon is due to the reasons described above and to the greater selectivity of metal deposition with the cathodic current densities on electrodes of large diameter.Copper and cadmium co-deposition According to EPMA data, during cadmium deposition the average cadmium content is negligible. During copper and cadmium co-deposition the average cadmium content increases tens or hundreds of times, which is due to the high mutual metal solubility and intermetallic compound formation15 On the other hand, since the standard electrode potentials of copper and cadmium (E°Cd2+/Cd0=-404 mV) are considerably diVerent, copper suppresses cadmium co-deposition much Fig. 4 Distribution of reduced cadmium content on GCC disc more than lead deposition. The correlation of XPS and XRF electrode surface modified by electrochemical co-deposition of copper data indicates that the relative cadmium content on the surface and cadmium from n×10-4M solutions of Cu and Cd nitrates in of the electrodeposit (2–3 nm) is significantly greater than in 0.01 M HNO3 according to EPMA data: 1, Cu (n=5) and Cd (n= the surface layers (1–2 mm) (Table 2).The distribution of the 3); 2, Cu (n=3) and Cd (n=3); 3, Cu (n=3) and Cd (n=5); 4, quasicopper content along the electrode radius is equivalent to the uniform metal distribution (R=electrode radius, r=distance from electrode center). distribution of the metal during individual copper deposition. J. Anal. At. Spectrom., 1999, 14, 425–428 427Conclusions Glass–ceramic carbon disc electrode surfaces modified by electrochemical co-deposition of copper, cadmium and lead from aqueous solutions were studied by modern spectroscopic methods.The eYcient technique of ‘integral’ and distributive EPMA of disc electrode surfaces was proposed. The dependence of the lateral and depth distribution of the metals on the electrode surface on the metal concentrations in solution, cathodic current density and type of intermetallic interaction was established. Tentative mechanisms of metal co-deposition are discussed. It was shown that the results of the spectroscopic investigation are correlated with the conclusions of theoretical electrochemistry. It was also demonstrated that the mutual application of spectroscopic methods with diVerent lateral and depth resolutions to the study of the chemical condition of Fig. 5 General view of distribution of copper to cadmium content solid electrode surfaces is highly eYcient. ratio on GCC disc electrode surface modified by electrochemical metal co-deposition with excess of cadmium in solution (1) and with excess This work was supported by the Russian State Scientific and of copper in solution (2) according to EPMA data (R=electrode Technical Programme ‘Fundamental Spectroscopy’ (Grant radius, r=distance from electrode center).No. 08.02.60). The authors are grateful to Dr. A.I. Kamenev for sample preparation and Dr. K.B. Kalmykov for assistance where r is the distance from the electrode center and R is the with EPMA measurements. electrode radius.According to eqn. (5), the current density distribution results in an increase in the cadmium content References from the center to the electrode edge. The second process is cadmium solubilization on the surface of the electrodeposit 1 Kh. Z. Brainina and E. Ya. Neiman, Tverdofaznye Reaktsii v during interaction with copper ions in solution. The rate of Elektroanaliticheskoi Khimii, Khimiya, Moscow, 1988. this reaction (vsol) can be described as follows: 2 M. Szklarczyk and J.O’M. Bockris, J. Electrochem. Soc., 1990, 137, 452. vsol=kaCd aCu2+ 3 A. J. Bard and F.-R. F. Fan, Anal. Sci. Technol., 1995, 8, 69A. 4 H. Dasarathy, C. Riley and H. D. Coble, J. Electrochem. Soc., aCd~sCd (r/R)~iCd2+ (r/R) 1994, 141, 1773. 5 M. Zhou, N. Myung, X. Chen and K. Rajeshwar, J. Electroanal. aCu2+~iCu2+ (r/R) Chem., 1995, 398, 5. 6 M. Jeske, J. W. Schultze, M. Thonissen and H. Munder, Thin vsol(r/R)~iCd2+(r/R)iCu2+(r/R)# CCu2+ CCd2+ iCd2+2(r/R) Solid Films, 1995, 255, 63. 7 N. V. Alov and K. V. Oskolok, in CANAS ’97, Colloq. Anal. Atomspektrosk., Freiberg, Germany, March 9–14, 1997, ed. C. where k is the reaction rate coeYcient, aCd and aCu2+ are the Vogt, R. Wennrich and G. Werner, Universita�t Leipzig und UFZ thermodynamic activities of the components, sCd is the share Leipzig-Halle, Leipzig, Germany, 1998, p. 437. of the surface of the electrodeposit covered by cadmium, 8 I. P. Viter, A. I. Kamenev, A. A. Sidakov and V. N. Zygan, Zh.iCd2+ and iCu2+ are the cathodic current densities of the metals Anal. Khim., 1994, 49, 1295. and CCd2+ and CCu2+ are the metal concentrations in solution. 9 N. V. Alov, K. B. Kalmykov, A. I. Kamenev, K.A. Kovalskii, K. V. Oskolok and V. K. Runov, Poverkhnost’. Rentgen. Analogous equations can be written for the third process— Sinkhrotron. Neitron. Issled., 1998, 6, 44. cadmium solubilization in the copper film. The second process 10 N. V. Alov, K. B. Kalmykov, A. I. Kamenev, K.V. Oskolok and contributes to a decrease in the cadmium content from the V. K. Runov, Dokl. Akad. Nauk, 1997, 353, 759. center to the electrode edge, whereas the third process contrib- 11 Yu. D. Gamburg, Elektrokhimicheskaya Kristallizatsiya Metallov utes to an increase in the cadmium content in this direction. i Splavov, Yanus-K°, Moscow, 1997. Thus, the resulting cadmium distribution has a maximum; the 12 N. V. Alov and K. V. Oskolok, in ECOSS 17, 17th European Conf. Surf. Sci., Europhys. Conf. Abstr., Enschede, The Netherlands, location of the maximum is a function of the ratio of copper September 16–19, 1997, ed. G. A. M. Kip and H. Wormeester, to cadmium concentration in solution. With an excess of European Phys. Soc., Enschede, The Netherlands, 1997, vol. 21E, cadmium in solution (that is, with a high cathodic cadmium WeP13. current), the cadmium distribution depends mainly on the 13 A. Milchev, Contemp. Phys., 1991, 32, 321. distribution of the cathodic cadmium current density on the 14 W. J. Lorenz and G. Staikov, Surf. Sci., 1995, 335, 32. disc electrode surface. The cadmium content increases from 15 M. Khansen and K. Anderko, Structury Dvoinykh Splavov, Metallurgizdat, Moscow, 1982. the center to the electrode edge (Fig. 4). Thus, with an excess 16 C. Nagl, E. Platzgummer, M. Schmid, P. Varga, S. Speller and of cadmium in solution, the copper to cadmium ratio decreases W. Heiland, Surf. Sci., 1996, 352–354, 540. from the center to the electrode edge and greatly increases with an excess of copper in solution (Fig. 5). Paper 8/06763E 428 J. Anal. At. Spectrom., 1999, 14, 425–4

ISSN:0267-9477    

DOI:10.1039/a806763e

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Surface analysis of halide distributions in complex AgX microcrystals by imaging time-of-flight SIMS (TOF-SIMS)† Geert Verlinden,a Renaat Gijbels,*a Ingrid Geuens,b and Rene De Keyzerb aUniversity of Antwerp, Department of Chemistry, Antwerp, Belgium bAgfa-Gevaert N.V., Mortsel, Belgium Received 17th September 1998, Accepted 2nd December 1998 The composition of the surface layer of silver halide microcrystals influences the behaviour of photographic materials as the latent image is formed at specific locations on the surface.In this work, an Ion-Tof IV TOF-SIMS instrument, equipped with a Ga+ liquid metal ion source (LMIS) operating at 25 keV, is used to reveal the diVerent elemental distributions in surface layers with a thickness below 4 nm. Superior lateral resolution (ca. 70 nm, calculated with computer generated line profiles) of the LMIS allows the analysis of individual microcrystals with a size well below 1 mm. The proposed methodology is based on a combination of secondary ion mass spectrometry, imaging and local image depth profiling.After selecting the appropriate measurement conditions, automatic data acquisition is possible. The method is successfully demonstrated for the following array of analytical applications: analysis of increased surface contamination due to storage of emulsions, explanation of sensitometric problems of tabular microcrystals, quick detection of cross contamination of diVerent types of crystals and inhomogeneous halide distributions at microcrystal surfaces.of surface distributions in individual Ag(Br,I ) microcrystals Introduction (dimensions 1–2 mm).1 The lateral resolution of the secondary The surface and internal structures of modern photographic ion images is ca. 300 nm, and ‘shallow’ depth profiling of the silver halide (AgX with X=Cl, Br, I ) microcrystals are inten- upper 50 nm in areas of ca. 0.1 mm2 has been demonstrated. tionally modified to improve various characteristics of the light The depth resolution between two subsequent secondary ion sensitive materials, such as image stability or spectral sensitivity.images was ca. 4 nm. Depending on the application, crystals with diVerent morpho- In newly developed materials, both halide concentrations logies, complex core–shell structures, conversion layers and and layer thicknesses have dropped by a factor of 10–100 (e.g. adsorbed inorganic or organic molecules are designed.from 10 to 0.1% iodide in Ag(Br,I ) and from 50 nm to below A possible structure of a core–shell crystal is shown in 5 nm, respectively) as will be demonstrated by the examples. Fig. 1. A high homogeneity of the intended crystal structure Therefore there is a definite need for an adapted method with over the crystal population is a necessity for obtaining an improved sensitivity, depth resolution and lateral resolution emulsion with reproducible photographic properties. in order to analyse these new materials and in this way support Due to the increasing demands for crystals with thinner the synthesis of more complex structured AgX light sensitive surface layers, shallower halide gradients and lower concen- materials.The new methodology based on time-of-flight sectrations of particular halides, new analytical procedures need ondary ion mass spectrometry (TOF-SIMS) is more suitable to be designed to characterize these complex AgX microcrys- for this type of analysis as a result of the lower primary ion tals.In previous work performed on a Cameca IMS 4f dose, higher sensitivity due to the higher transmission and instrument, we have shown that scanning ion microprobe mass improved lateral resolution (compared with dynamic SIMS). spectrometry using a focused Cs+ ion beam allows the study In general, the method shows the possibilities of TOF-SIMS imaging on individual and (sub)micrometre sized particle †Presented at the Fifteenth International Congress on X-ray Optics populations.and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998. AgBr A B AgBr0.9I0.1 AgBr0.9I0.1 AgBr AgBr0.9I0.1 AgBr AgBr Fig. 1 Schematic representation of the internal structure of core–shell cubic ( left) and tabular (right) silver halide microcrystals. Insets A and B represent a top view and a cross section of a tabular core–shell microcrystal. J. Anal. At. Spectrom., 1999, 14, 429–434 429over the population are within 10% of the figures obtained by Experimental other analytical techniques (SEM-EDX, XRF).Instrumentation When the appropriate measurement conditions (primary pulse length, number of pulses/pixel, number of scans) are All TOF-SIMS experiments were performed on an Ion-Tof IV determined for a given sample, computer controlled automatic TOF-SIMS instrument (Cameca, Courbevoie, France). A data acquisition of subsequent fields is possible using a routine pulsed Ga+ liquid metal ion source operating at 25 keV was developed in our laboratory.used for the primary ion bombardment. The sample surface was bombarded with short primary ion pulses with a typical Sample preparation length varying between 1 and 200 ns depending on the type of analysis (secondary ion mass spectrometry at high mass reso- Tabular and cubic microcrystals were prepared with the double lution or secondary ion imaging). For a more detailed descrip- jet precipitation method10 at Agfa Gevaert N.V.The average tion of the instrument and the principles of SIMS and TOF- lateral dimensions of the crystals were 0.9 mm and 0.7 mm SIMS in particular, the reader is referred to refs. 2–4. Basic respectively. The studied emulsions were diluted with distilled aspects of imaging SIMS are reviewed in refs. 5–7. water and degelled by centrifugation. The dilution was checked with electron microscopy to make sure that the microcrystals Methodology were not coagulated.Then a drop of the dispersion was placed onto a piece of cleaned, polished silicon, dried and transferred The main objective of this work was to develop a method for into the TOF-SIMS instrument. All manipulations and experi- characterizing silver halide microcrystals and, in particular, ments were performed under darkroom conditions. for determining the halide distributions in surface (and subsurface) layers thinner than 4 nm in individual microcrystals. It Results and discussion is also important to be able to analyse in an acceptable time as many crystals as possible in order to obtain statistically In this section, four case studies are presented to illustrate the significant data for the crystal population.The method can be diVerent analytical problems to which the method was successdivided into three main parts as detailed below. fully applied. First, secondary ion mass spectrometry at low magnification (typical size of analysed fields, 100×100 mm2) is used to 1.EVect of sample storage on the surface composition of AgX determine the diVerent elemental distributions at the crystal microcrystals surface over the crystal population (i.e. groups of microcrystals). During this step, short primary ion pulses (<1 ns) are The extreme surface sensitivity of TOF-SIMS demands caution and precise, clean working conditions during sample prep- employed. In this way, high mass resolution spectra (typical values are 2000 on H and 8000 on C2H5) are obtained, which aration and sample storage to avoid surface contamination.Hydrocarbons, gelatine residues and silicone oil from vacuum makes the identification of diVerent peaks at the same nominal mass possible. The primary ion dose density is below 1×109 pumps can contaminate the crystal surface in such a way that accurate analysis of the outer halide distributions becomes ions cm-2, ensuring that the static limit is respected (typical value 1×1013 ions cm-2 for inorganic applications8) and thus impossible.The timing between the synthesis of the emulsion, the sample preparation and the TOF-SIMS experiment is also no significant erosion of the outermost surface takes place. The second and third parts consist of scanning ion imaging. important as the emulsion and the prepared sample must be stored in light-proof containers or cardboard boxes, which In these parts, high lateral resolution (ca. 60–70 nm) is the primary objective. To obtain optimal lateral resolution (small- will increase the surface contamination.In order to evaluate the storage of prepared samples in our est possible beam size), the voltages of the primary ion optics are altered. The primary ion pulse length is raised in the range laboratory, the same sample was analysed with a time interval of 12 months. Although this interval is too long to be of 100–200 ns in order to obtain a suYcient signal to noise ratio in the secondary ion image.This results in higher primary representative, the results clearly indicate the problems that arise due to prolonged storage. Fig. 2 shows scanning second- ion dose densities and unit mass resolution in the mass spectra (typical value ca. 500). ary ion images of the 1H-, 16O-, 32S- and 35Cl- distributions in a selected field acquired under identical experimental con- Overview images are taken at magnifications of 30×30 mm2 with primary ion doses in the region of 1×1012 ions cm-2.In ditions (primary ion type and dose, magnification). The letters/ numbers below each image indicate: the mass of the selected these images, diVerent elemental distributions are visualized at the crystal surface of a group of microcrystals. ion, the maximum intensity per pixel in the image and the total number of secondary ions detected for the selected ion Finally, secondary ion images are recorded at high magnifi- cation (typical size of analysed field, 10×10 mm2) and local in the image.The top row consists of images acquired from a freshly prepared sample, and the bottom row are images from image depth profiling (computer reconstructed) is performed. In this way, diVerent halide intensity ratios for individual the same sample after 12 months of storage. The diVerences between the two series of images are quite microcrystals can be calculated and compared. Primary ion dose densities are significantly higher during this last step and clear.In the hydrogen and oxygen images, there is a clear distinction between the signal coming from the substrate Si are in the region of 1×1014 ions cm-2. Based on the areal density of AgX (which is 1.21×1015 atoms cm-2 for AgBr (oxygen is more enriched on the substrate due to the native oxide layer of Si) and the crystal surface for the newly prepared and 1.29×1015 atoms cm-2 for AgCl) and the experimentally determined sputter yield of 24 for AgBr under Ga+ bombard- sample.This distinction is completely lost for the sample after long storage. The hydrogen and oxygen images show some ment (25 keV; angle of incidence, 45°), it is clear that the static SIMS limit is exceeded and between 1 and 1.5 monolayer intensity variation at the crystal edges caused by topography induced secondary ion yield changes, but there is no clear is removed per recorded secondary ion image. In contrast with double focusing mass spectrometers, the separation between crystal surface and substrate.Fig. 3 shows four subsequent scans of the oxygen distribution of the aged parallel mass detection of the TOF mass analyser allows a direct comparison of distributions of diVerent elements at the sample: the separation between crystal surface and substrate becomes clearer in the last scan. Based on the primary ion same sputtered depth. Due to the nature of the elements under investigation (the halides), the intensity ratios can be readily dose, the contamination layer is estimated to be ca. 2 nm thick (the exact sputter yield of the environmentally induced con- converted into concentration.9 The obtained values averaged 430 J. Anal. At. Spectrom., 1999, 14, 429–434Fig. 2 Secondary ion images of 1H-, 16O-, 32S- and 35Cl- acquired with a time interval of 1 year (top, newly prepared; bottom, after 1 year of storage). The letters/numbers below each image indicate: the mass of the selected ion (e.g. M:1 for 1H-), the maximum intensity per pixel in the image (bottom left) and the total number of secondary ions detected for the selected ion in the image (bottom right).Bright Fig. 4 High resolution TOF-SIMS spectra of masses 16 and 32 of the parts in the secondary ion image correspond to high secondary newly prepared (top row) and aged (bottom row) samples. The signals ion intensity. of S- and O2- at mass 32 are separated. The diVerence in S- intensity in the new and aged samples is clear.tamination layer is not known). Another important point is the definition of the edges of the tabular Ag(Cl,I ) crystals in the 35Cl- image. The top row in Fig. 2 shows an image with well-defined edges from the beginning of the analysis. In the bottom row, the edges of the microcrystals are blurred. Finally, there is evidence of S enrichment at the crystal surface of the stored samples. This is not surprising due to the well-known aYnity between Ag and S (e.g. Ks of Ag2S is 6.69×10-50).The distinction between S adsorption and O2 adsorption can be made on the basis of the following observation. Fig. 4 shows the high mass resolution spectra of masses 16 (O-) and 32 (S- and O2-). At mass 32, the masses of S- and O2- are clearly separated. For the new sample, the O2- is clearly dominant and only a minor S- signal is present. The stored sample, however, shows a prominent S- signal that dominates the O2- signal. For both samples, the O2-/O- ratio is approximately the same (7×10-3 for the new sample and 6×10-3 for the aged sample).Accordingly, the image of the newly prepared sample is representative of the O2- distribution and the image of the aged sample is representative of the S- distribution, in spite of the fact that the images were acquired at lower mass resolution. This observation shows in general the need and importance of the acquisition of secondary ion mass spectra with high mass resolution prior to image analysis. Fig. 5 Mean intensities of 1H- and 13CH- in the surface layer of the The mean intensities of the 1H-, 13CH- and 32S- ions in new and aged samples for subsequent scans. the surface layer for subsequent scans are plotted in Figs. 5 Fig. 3 From left to right: four subsequent scans of the same field of the aged sample showing the 16O- distribution. After scan 4, the separation between the crystal surface (dark areas) and the substrate becomes clearly visible, illustrating the thickness of the contamination surface layer due to storage.J. Anal. At. Spectrom., 1999, 14, 429–434 431Fig. 6 Mean intensity of 32S- in the surface layer of the new and aged samples for subsequent scans. Fig. 8 Secondary ion images of 1H-, 35Cl- and 127I- of sample B. Notice the diVerence in the number of Ga+ ion pulses/pixel to acquire the images (top left corner) from that in Fig. 7. and 6. The values are calculated on the basis of 20 local image depth profiles from individual microcrystals of each sample.The error bar represents the standard deviation of the measureindication of the complete absence of an iodide conversion ment. Again there is a clear diVerence between the values for layer for sample B. these elements obtained for the new sample and the stored The results were confirmed with secondary ion imaging. sample. The mean H and CH intensities diVer between the Fig. 7 and 8 show the 1H-, 35Cl- and 127I- distributions for new and stored samples at the crystal surface by factors of 3.2 a selected field in both samples.The iodide distribution at the and 2.8, respectively. Significantly higher values are found for crystal surface is homogeneous for both samples. However, the stored sample. Fig. 6 confirms the observation of the buildthe experimental conditions necessary to image the iodide are up of an enriched S surface layer. diVerent for samples A and B. In order to obtain a suYcient The images in Fig. 3 and the variations of the intensities as signal-to-noise ratio for sample B, the primary ion dose must a function of the number of acquired scans in Figs. 5 and 6 be increased by a factor of 6.67 (1000 Ga+ ion pulses/pixel also show that, at these higher magnifications and primary instead of 150 under normal conditions). So, despite the low ion doses, the static SIMS limit is exceeded and there is iodide concentration in sample B, TOF-SIMS imaging of these significant erosion of the surface layer.low concentrations in individual submicrometre crystals is still possible as long as the element is not confined in a subnano- 2. Tabular Ag(Cl,I ) microcrystals with sensitometric problems metre thick surface layer so that the intensity can be accumu- In this case study, two emulsions containing Ag(Cl,I ) tabular lated from a certain depth interval. microcrystals with an iodide conversion layer are compared. Finally, to complete the analysis, local image depth profiles The crystals consist theoretically of a 99.57% AgCl+0.43% were constructed from the measurement data to calculate the AgI core (90 vol.%) and a 95% AgCl+5% AgI shell (10 variation of the iodide concentration as a function of sputtered vol.%).Although the theoretical crystal structure and the depth. The results are presented in Fig. 9. The iodide gradient global iodide concentration (determined with XRF) are ident- from the thin conversion layer found in sample A is completely ical for both emulsions, sensitometric tests for one emulsion showed unsatisfactory results.A possible hypothesis to explain this problem was an irregularity during the iodide conversion of the microcrystals. The two crystals are hereafter referred to as sample A (sensitometry satisfactory) and sample B (sensitometry unsatisfactory). Secondary ion mass spectrometry of groups of microcrystals at diVerent positions in both samples immediately indicates a diVerence of about a factor of 20 in the surface iodide concentration.The mean surface iodide concentration for sample A is 10.8±0.2% (five measurements) and corresponds to the thin iodide enriched conversion layer. For sample B a mean concentration of 0.49±0.03% is found. This concentration corresponds very well with the global iodide concen- Fig. 9 Iodide concentration profile as a function of depth for samples A and B (calculated from local image depth profiles). tration determined with XRF (i.e. 0.43%). This gives a strong Fig. 7 Secondary ion images of 1H-, 35Cl- and 127I- of sample A. 432 J. Anal. At. Spectrom., 1999, 14, 429–434Fig. 10 Secondary ion images of 35Cl-, 81Br- and 127I- at low magnification. The inhomogeneous Cl distribution is clearly visible. The diVerence in surface iodide concentration for the Ag(Cl,I ) and Ag(Br,I ) microcrystals is already apparent. absent in sample B. The thickness of the conversion layer in sample A is estimated to be ca. 2 nm.On the basis of these results, it can be concluded that the sensitometric problems in sample B are not caused by other known factors, such as coprecipitation or inhomogeneous distribution of iodide at the surface, but by the complete lack Fig. 12 Three successive scans of two cubic Ag(Cl,I ) crystals [35Cl- of the conversion layer. The problem was later tracked back (top), 127I- (bottom)]. The surfaces of the microcrystal on the left are to a defective automatic inlet system in the precipitation vessel.iodide enriched (especially the edges and corners of the cube), the surfaces of the microcrystal on the right contain no iodide. 3. Detection of cross contamination of two diVerent AgX emulsions The presented methodology can also be used as a quick alternative to check for contamination or possible cross contamination of diVerent emulsions after precipitation. The combination of parallel mass detection (yielding isotope specific images) and the good lateral resolution of the Ga+ gun mean that fields containing ca. 100 individually resolved microcrystals can be scanned and surveyed in 82 s under standard conditions. This oVers a definite time advantage over other microanalytical techniques, such as SEM-EDX, where every individual particle has to be scanned for a number of seconds to obtain an X-ray spectrum with a suYcient signalto- background ratio. Fig. 13 Computer reconstructed line scan across the top face of the To test this possibility, we intentionally contaminated an cubic crystal on the right of Fig. 12 showing the 35Cl- intensity emulsion of Ag(Br,I ) microcrystals with a small volume distribution. fraction of Ag(Cl,I ). A second point of interest was to check whether we could distinguish the diVerence in iodide concentration at the surface of the Ag(Br,I ) and Ag(Cl,I ) crystals Examination of diVerent fields showed an Ag(Cl,I ) crystal to Ag(Br,I ) crystal ratio of 0.1.The counting statistics for the that was observed in preliminary experiments on the separate emulsions. The local concentration was at subpercentage levels iodide image at this magnification are poor, but it can already be seen that the brighter areas of the iodide distribution for the Ag(Br,I ) and ca. 2% for the Ag(Cl,I ) microcrystals. Fig. 10 shows secondary ion images of the 35Cl-, 81Br- and coincide with the Ag(Cl,I ) crystals. At higher magnification (Fig. 11), the diVerence in iodide concentration at the surface 127I- distributions at low magnification.The diVerence in the Cl and Br distributions is clear even at this magnification. becomes clearer. It can also be seen that the accurate visualiz- Fig. 11 Secondary ion images of 35Cl-, 81Br- and 127I- at high magnification. The inhomogeneous Cl distribution is clear. The iodide enrichment at the surface of the Ag(Br,I ) crystals is no longer visible. J. Anal. At. Spectrom., 1999, 14, 429–434 433ation of low iodide concentrations under standard experimen- distributions in surface layers (<4 nm) over the crystal population and in individual silver halide microcrystals.For these tal conditions becomes very diYcult due to a combination of the low concentrations and reduced thickness of the enriched thin layers, the results are clearly superior to those obtained in more dynamic SIMS analysis.15,16 layer in these samples. This observation has also been con- firmed in recent experiments on tabular microcrystals with iodide conversion layers created with varying iodide Acknowledgements concentrations.11 This work was supported by the ‘Flemish Institute for the Encouragement of Scientific and Technological Research in 4.Analysis of inhomogeneous iodide distributions at cubic the Industry’ (IWT), by the Flemish Fund of Scientific microcrystal surfaces Research (FWO) and by the Federal Services for Scientific, Fig. 12 shows three consecutive scans of the 35Cl- and 127I- Technical and Cultural AVairs (DWTC/SSTC) of the Prime distributions in cubic Ag(Cl,I ) microcrystals (global iodide Minister’s OYce (IUAP-IV.Conv. P4/10). concentration, 0.4%; lateral dimension, 0.7 mm). In these images, a three dimensional eVect is visible, created by the References non-identical angles of incidence on the top and side faces of 1 G. Verlinden, G. Janssens, R. Gijbels and P. Van Espen, Anal. the cubic crystal. The diVerences in secondary ion yield make Chem., 1997, 69, 3772.the cubic morphology in the image very clear. Only the crystal 2 A. Benninghoven, B. HagenhoV and E. Niehuis, Anal. Chem., on the left shows significant iodide enrichment. For the top 1993, 65, 630A. surface of this crystal, complementarity between the iodide 3 D. Briggs and M. P. Seah, Practical Surface Analysis: Ion and and chloride distributions can be detected. Higher iodide Neutral Spectroscopy, Wiley, New York, 1990, vol. 2. concentrations are detected at the edges and corners of the 4 A. Benninghoven, F. G. Rudenauer and H. Werner, Secondary Ion Mass Spectrometry: Basic Concepts, Instrumental Aspects, crystal. This observation is no edge eVect or SIMS artifact Applications and Trends, Wiley, New York, 1987. and has already been reported by other authors for bigger 5 R. W. Odom, in Secondary Ion Mass Spectrometry Imaging cubic crystals with higher global iodide concentrations.12,13 It (Practical Spectroscopy Series), ed.M. D. Morris, Marcel can be explained by the diVerence in crystal lattice structure Dekker, New York, 1993, p. 354. between AgI (hexagonal ) and AgCl (face centred cubic). 6 J. Schwieters, H. G. Cramer, T. Heller, U. Ju� rgens, E. Niehuis, Hence the incorporation of AgI in an AgCl lattice during Zehnpfenning and A. Benninghoven, J. Vac. Sci. Technol. A, 1991, 6(9), 2864. precipitation will take place at lattice locations with the most 7 F.G.Ru� denauer, Anal. Chim. Acta, 1994, 297, 197. defects. This makes edges and corners more suitable for the 8 A. Benninghoven, Surf. Sci., 1975, 53, 596. location of AgI in the crystal lattice. 9 G. Verlinden, R. Gijbels and I. Geuens, in Secondary Ion Mass The consecutive images show again that the static limit for Spectrometry SIMS XI, ed. G. Gillen, R. Lareau, J. Bennett and these materials is exceeded under these analysis conditions. F. Stevie, Wiley, Chichester, 1998, p. 995. After three scans, the iodide intensity has almost completely 10 C. R. Berry, in The Theory of the Photographic Process, ed. T. H. James, Macmillan, New York, 1977, p. 88. disappeared from the top and front planes of the cube indicat- 11 G. Verlinden, internal communication 1998. ing only a very thin iodide layer of about 2.5–3 nm. The 12 J. M. Chabala, in Secondary Ion Mass Spectrometry SIMS X, ed. lateral resolution of the images was calculated with a computer A. Benninghoven, B. HagenhoV and H. Werner, Wiley, generated line profile (Fig. 13) and is 70 nm using the 16–84% Chichester, 1997, p. 23. criterion.14 13 R. Steiger, Chimia, 1994, 48, 444. 14 H. Arimoto, T. Akira, E. Miyauchi and H. Hashimoto, Jpn. J. Appl. Phys., 1983, 22, 1780. 15 I. Geuens, R. Gijbels, R. De Keyzer and A. Verbeeck, Proceedings Conclusion of ICPS 94, Rochester, USA, IS&T, Springfield, USA, 1994, p. 27. Four examples demonstrate the practical use of the developed 16 J. M. Chabala, Int. J. Mass Spectrom. Ion Proc., 1995, 143, 191. analytical procedure based on TOF-SIMS. The method is successfully used for the determination of the diVerent halide Paper 8/07276K 434 J. Anal. At. Spectrom., 1999, 14, 429–4

ISSN:0267-9477    

DOI:10.1039/a807276k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Analyses of petrified wood by electron, X-ray and optical microprobes† Andrzej Kuczumow,a Bart Vekemans,b Olivier Schalm,b Walter Dorrine�,b Pierre Chevallier,c Philippe Dillmann,c Chul-Un Ro,d Koen Janssensb and Rene� Van Grieken*b aFaculty of Chemistry, Maria Sk�odowska-Curie University, 20–031 Lublin, Poland bMicro- and Trace Analysis Centre, Department of Chemistry, University of Antwerp (UIA), B-2610 Antwerp, Belgium cLPS, CEN Saclay et LURE, Universite� Paris-Sud, Bat 209D, F-91405 Orsay, France dDepartment of Chemistry, Hallym University, Chuncheon, 200–702, Korea Received 28th August 1998, Accepted 27th November 1998 Samples of petrified wood of diVerent origins were analyzed by the use of the electron microprobe, capillary X-ray fluorescence microprobe, synchrotron capillary X-ray microprobe and optical microscope, applied in a microprobe manner.The main attention was given to the investigation of the ring structure of the petrified wood and the comparison of this with the ring structure of the living trees analyzed by much the same methods.The continuous X-radiation, applied in a microprobe manner, the distribution of the gray-scale representation of the secondary electron intensities and the characteristic X-ray signals, mainly from the light elements, were registered by the use of the electron microprobe method. The X-ray capillary microprobe detected the Rayleigh and Compton signals, scattered from microareas of the samples, and the characteristic X-ray signals, mainly from the heavier elements.In the synchrotron-based capillary microanalytical measurements, one of the most important results was achieved by the microprobe application of scattered synchrotron radiation. The emission and scattering results were supplemented by transmission measurements, where possible. All the methods proved to be complementary in the analysis of such periodic structures as tree rings. Both capillary microprobes were much more eYcient in the detection of heavy elements and penetrated deeper than the traditional electron microprobe.Careful analysis of diVerent signals indicated that some samples of petrified wood in the authors’ possession, composed of silica of variable density, are the chemical negatives of the primordial living wood. This is the first such observation in the literature. MicrodiVraction studies of the samples proved that polycrystalline a-quartz was the main matrix component of all these samples.The elemental analysis of the petrified wood gives important indications about the petrification processes. Comparison of the particular ring structure of the petrified wood with the ring structure of living trees shows great similarities. The widths of rings, density variations and density maxima are easily readable from the microanalysis of petrified wood. These parameters potentially can be exploited for the investigation of the biological, chemical, chronological and climatic information included in the fossilized tissues.image of the ring. Smaller steps allow penetration of the intra- Introduction ring structure, giving information about the seasonal changes Analyses of wood have attracted great interest for centuries, in, e.g., temperature. Previous research with X-ray microprobes and there has been a renaissance recently, coupled with enabled the changes to be traced with an accuracy correspondprogress in modern optical and especially X-ray techniques. It ing to a decade change in the living wood density.27 On the has been reported that the structure of wood preserves much other hand, scanning in small steps makes a spectrum ranimportant information, and the ring structure of wood seems domized and obscure, preventing the extraction of important to be of special interest.The data which can be extracted from general information. In extreme cases, excessively detailed tree rings include dating1–5 (each ring can be identified and scans demand averaging, which obviously contradicts the coupled with the annual increase in the biomass), nutrition earlier decrease in the step size.conditions, information about pollution, detailed climatic data DiVerent kinds of X-ray microprobe methods have been (overall kind of climate,6 temperature and precipitation7,8) applied in research on living wood structure. The first attempts and previous volcanic activity.9 Similar data can be extracted were made by PIXE.28–30 Later, an X-ray capillary microprobe from many natural deposits which are characterized by annual was used.27,31 The electron microprobe has not been applied increase patterns, e.g., coral skeletons,10–13 hard clamshells,14,15 in the field, probably because it demands extensive sample stalactite and stalagmite structures,16–21 ice-core deposits22–24 preparation and a vacuum is essential.While PIXE microanaland marine sediments.25,26 Measurements of the density and ysis allowed the tracing of ring structures by scanning the contents of diVerent trace elements by linear or area mappings, chemical composition are of great significance in the field of the X-ray capillary microprobe introduced a new approach, tree ring research.The fundamental question is how sensitive the use of scattered Rayleigh or Compton radiation for density our probes are towards the changes in composition/density/ determinations. Especially, linear scans of the density were of topographical details and how many steps within the radial interest.One can understand this, since the most important size of the ring can be made to obtain a suYciently detailed information is coupled with one dimension only, namely that radial to the trunk axis. It has been proved that for trees †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.growing in Polish or southern Swedish conditions, the intra- J. Anal. At. Spectrom., 1999, 14, 435–446 435ring density patterns strictly followed the changes in the warm transforms into gel but only after the penetration of liquid into the tissue.36 There have also been other similar reports period temperature distribution. The intra-ring patterns in northern Swedish Picea abies wood had diVerent structures. announcing the easy transformation of the wood into a fossilized-like form.37–39 The conditions existing in such arti- Their density patterns followed only the very short summer’s biological activity, characteristic of the northern regions.The ficial solutions were found to be similar to the conditions existing in places of great volcanic activity. Indeed, the petrifi- dark wood zone was greatly shortened in total ring structure. Thus, X-ray methods allow not only tracing of the long-term cation process at Yellowstone National Park occurs at a rate of 1–4 mm yr-1.40 However, the process as such obeys a very patterns of ring distribution, which has been applied in chronology and climatology for many years, but also tracing of the subtle equilibrium, because probably even very small changes in the environment can provoke the uncontrolled precipitation short-term, sub-year or seasonal changes in rings, strictly correlated with temperature variations.If it is possible to of silica; then nothing is preserved of the primary wood structure. The main disrupting factors include a sudden examine the wood ring structure of recent trees, it would be even more interesting to examine the ring structure as it is increase in heavy metal content, change in pH of the solution and thermal shocks.A recent report by Carroll et al.34 describes preserved in petrified wood. Petrified wood is known in many regions of the world, the precipitation of amorphous silica in the temperature range 60–120 °C in both laboratory and field experiments.It has especially the USA (Arizona, Texas), Argentina, Antarctica, Mongolia32 and the Sahara. It is recognized as an important been established that there are surface defects and surface nucleation centers which control the rate of the process in geological witness of earlier biological activity. Some of the samples are considered to be as old as r 200 million supersaturated and trace element-rich solutions. During the geological scale of time, the primary hydrated gel transforms years, but examples of pieces only a few hundred years old are also known.Petrified trunks from the Sahara and Arizona into an opal-like structure and finally into a polycrystalline quartz (chert) form.37,41 The transformation process takes deserts and Antarctica provide information about the biological life existing in places where nothing is now growing. Major about 40 million years in natural conditions and is the single secondary process widely described.There are many other natural collections of petrified forests are now available in some national parks (USA, Argentina). Cross-sectioned pieces secondary processes leading to the reshaping of the structure and composition of petrified wood, but such transformed of wood are prized for their beauty, especially if saturated with heavy metal oxides. Nevertheless, where the petrified objects are not of particular interest. The aim of this research was to detect if any kind of wood is used for dating, pieces with clear and simple ring patterns and without disorders and inclusions are preferred.information, which is observable by the naked eye on the cross-section of a petrified tree, namely clear remnants of tree The petrification (fossilization) process is a peculiar phenomenon from the biological and geological points of rings, can be observed by instrumental methods and quantified. We selected the tools that should normally be used in research view. In general, the petrified wood is found in the silicified form; however, specimens transformed into calcite,6 pyrite or on fresh tree rings, i.e., the electron microprobe and X-ray capillary microprobe.With suYcient spatial resolution, these charcoal are known also.33 Because the silicified form is far more prevalent and probably better preserved than others, it methods should yield detailed scans of the ring and intra-ring structure.One can then compare the structure patterns with was adopted as an object of our research. Two mechanisms responsible for a silicification process are known. In the first, those of living wood. If they are very similar, then the conclusions drawn for modern living trees might be carefully the tissue underwent decay in a moderately hot environment, rich in minerals. The organic tissue was simply replaced by matched with petrified wood scans. Some modifications to the techniques and methods applied in studies of fresh wood will inorganic, silicon-rich matter.In that kind of process only the sequence of rings is preserved, reproduced in the new material. be suggested to meet the limitations imposed on such a specific kind of material as petrified wood. To our knowledge, linear One can call this kind of process the ‘replacement process’. Dark colored samples analyzed by us seemed to belong to this scans of such samples have been made up to now only with the use of optical microprobes and never with the use of kind of fossil. Another kind of petrification occurs when a solution containing some mineralizing chemicals, such as dis- electron or X-ray microprobes; elemental scans have never been made.solved silica34 and potassium and sodium silicate, penetrates the tissue, not replacing it, but impregnating it. Then not only the general pattern of rings, but also even the patterns of tissues and cells are conserved. This process can be called an Experimental ‘impregnation process’.The remains of this process are clearly observable in some of our beige-colored samples. The natural Samples process often involved a balanced contribution of these two mechanisms. In many cases the trees were covered with ash Several samples of petrified wood were collected for our research. Two of them were from the collection of the Higher from a volcanic eruption35 and conserved in that way. Later, they were covered and saturated with water (e.g., in a lake or School of Farming, Agriculture Academy, Warsaw, Poland.One is beige in color, typical for a silicon oxide-rich composi- spring) and soluble Si compounds, most probably diluted silica and alkali silicates, impregnated the tissue material. The tion (indicated below as S1). The second is dark brown, indicating a greater level of mineralization by saturation with volcanic ash is a probable source of the silicon compounds.The spring waters in regions of great seismic activity heavy metal oxides (S2). The third piece was taken from the natural reserve in Siedliska village, very close to the Polish– (Yellowstone, New Zealand and Kamchatka) leach the silicon compounds from the lava and ash and include considerable Ukrainian border. This piece is also beige, with well-preserved tree rings in the outside parts, with sand-like sediments amounts of soluble silica and silicates. Then the material is transformed from dissolved silica and alkali metal silicates observed inside (S3).Several other samples ( labeled S with further numbers) were obtained from a private collector during into insoluble Si compounds, SiO2 being the most common. It has been discovered that one can imitate the natural the Antwerp Exhibition of Minerals, Minerant, 1998. These were dug out in Hoegaerden, Belgium, and belonged to the process by the impregnation of living wood with a laboratorymade solution, composed of, e.g., sodium or potassium silicate local geological formation called Landeniaan (or more generally Maestrichtiaan) from the end of the Jurassic era, 75 dissolved in spring or volcanic water, rich in some metallic ions and with an addition of citric or malic acid.This solution million years old. The species to which they belonged are 436 J. Anal. At. Spectrom., 1999, 14, 435–446assumed to be Dynoxilion silvestri. All the pieces from Belgium in front of the capillary outlet by a set of four computerresembled visually the beige samples S1 and S3 from Poland.controlled step motors. A linear scan was again the preferred The samples were prepared for analysis in such a way that option; thus only one motor was eVectively working, except the pieces were cut from the mineralized trunk fragments with at the moment of initial positioning of the sample. An Si(Li) a diamond saw. Pieces were cut in a manner to reveal surfaces detector was applied for the detection of the fluorescent and perpendicular to the trunk axis and at the same time radial to scattered radiation and an NaI(Tl ) scintillation detector for the tree rings.Fragments that were suYciently large were cut the registration of the transmitted radiation. The scans were oV and presented to the beam after slight polishing. The made on samples that were as thin as allowed by the sample smaller cut oV pieces were embedded in resin. The surfaces to preparation technique (about 1 mm), in order to avoid an be analyzed were polished with corundum abrasive papers of excess of scattered radiation, the intensity of which is prosteadily decreasing grain size.The final polishing was eVected portional to the thickness of the sample and more than with the use of fine diamond powder rotating targets, with the suYcient for the analytical aims. The brittleness of the material grain size of diamond powder decreasing from 15 to 1 mm.prevented further thinning of the samples. However, the This kind of preparation was suYcient for analysis by capillary thickness of the samples was uniform and under control to X-ray spectrometry. For electron microprobe analysis, the within 10 mm. surface had to be covered with a thin conductive graphite The synchrotron measurements were performed at the 4.45 layer. For microdiVraction measurements, the bulk samples GeV positron synchrotron in HASYLAB, Hamburg. Beamline were polished on one side and the next 0.7 mm thick layer L was used, but the primary beam was attenuated by the was cut oV, with the cut surface parallel to the polished surface.introduction of an 8 mm thick Al sheet. Other restrictions Even after careful preparation, the samples had some defects. were imposed on the final beam size by installing a 10 cm long Especially, the ring boundaries were filled with a brittle mate- straight capillary of 30 mm id. The energy range of the primary rial, which was in many places removed, with openings left.beam was between 1 and 100 keV. The secondary radiation This demanded careful selection of the scan location, great was registered with an HPGe detector. The fluorescent lines care during the analyses and special attention during the data and also the selected channel in the scattered white synchrotron treatment. Unfortunately, the relatively small pieces of our radiation (38.5–53.5 keV) were chosen for the collection of material did not allow the analysis of very long sequences of the data.petrified tree rings (over 40 rings). All the scans, regardless of the method applied, were carried out on the same samples and at locations as close as possible in order to obtain comparable results. However, the diVerent Instrumentation methods of sample preparation and diVerent times and places The samples were analyzed by the use of diVerent micro- of analysis did not allow strictly the same positioning for analytical devices.The first was IMIX (Integrated Micro- diVerent methods. The lengths of the scans were comparable, analyzer for Imaging and X-Ray), produced by Princeton but they could sometimes diVer in the position perpendicular Gamma-Tech. It was a scanning electron microscope with an to the length direction by as much as several hundred energy-dispersive Si(Li) detector. The working voltage for the micrometers. Because the ring structure is not strictly parallel, electron gun was always set on 20 kV.The line profile analysis this sometimes gives a slight shift in the ring width. option was selected to make linear scans through the samples. In general, the electron microprobe can deliver a better This option was mostly dedicated to match a material such as spatial resolution than the X-ray capillary microprobe (even tree rings, in which the most important information is included up to three orders of magnitude). The depth penetration and in the radial profile.The searched field of view was selected the information volumes are totally diVerent (the depth pene- using the image obtained from the secondary electron signals. tration of the material for the electron microprobe is of the The image extracted from the secondary electron signals was order of a few micrometers whereas for the tabletop X-ray preferred, because the topographical orientation on the sample capillary microprobe it reaches 1 cm and for the synchrotron was essential for coupling the chemical information with version it even exceeds 5 cm).The electron microprobe brings surface details. The line scan was performed across the center information more concentrated on the low-Z elements, whereas of the image. The orientation of the image was selected in the two X-ray capillary microprobe versions are much more such a way that the scan proceeded perpendicular to the trunk sensitive for heavy element determinations.Hence, the electron and radial to the ring direction. The radial direction was microprobe brings results that are strictly surface-orientated, followed outwards. The secondary electron image of the whereas the X-ray microprobes give a much deeper insight sample was stored to establish and control the location of into the sample. Still, if the results can easily be compared other measurements to be continued on the same sample. This with the same structure observable in all kinds of scans, this image was also used in our analysis in another way.A graytestifies to the uniform general features of the petrified ring scale level distribution along the line of scan was made. The structure. Brief details of the methods applied are given gray-scale morphology diagram in its raw form was diYcult in Table 1. to analyze but, after smoothing of data points, the internal The microdiVraction measurements were made on an X-ray structure of the scanned area was revealed.All the samples microprobe at LURE (Orsay, Paris). It was installed on the analyzed by the scanning electron microprobe were thick. We D15 beamline at the DCI storage ring. This installation has could not prepare samples thinner than several hundred been described elsewhere.44,45 The 10 keV photons were micrometers and the penetration depth of this kind of extracted from the white synchrotron radiation and, after microprobe was not greater than 5 mm.passing the input diaphragm, focused with the Bragg–Fresnel A tabletop X-ray capillary microprobe was the second multilayer lens to a focal spot 20 mm wide. The diVracted option for the analysis. The device was built in the Department radiation was registered in the transmission mode on a of Chemistry at the University of Antwerp (UIA), Belgium, 2D-CCD ‘imaging plate’ made by Fuji. The image from the and has been described in detail elsewhere.42,43 A Siemens plate was scanned with a Molecular Dynamical scanner and rotating anode Mo tube was used for excitation.The X-ray stored as an image-type data file. After the image integration, beam was squeezed to the size of the outlet capillary diameter, the diVraction 2h pattern was extracted. It can be matched i.e., 15 mm. Since the distance from the outlet to the sample with the library of the diVraction patterns or processed for was relatively large (about 1.5 mm), the eVective beam size was at least double the above value.The sample was driven phase identification. J. Anal. At. Spectrom., 1999, 14, 435–446 437Table 1 Details of the methods applied Microprobe Spatial Steps Penetration Detectable Auxiliary method resolution/mm applied/mm depth/mm elements signals Electron <1 50< 5 Na–Ca Bremsstrahlung Tabletop X-ray (capillary) 15–30 100 Up to 10 000 Ti–Pb Rayleigh-Compton Synchrotron X-ray >30 50 Up to 50 000 Ti–Pb White the structure, which is not a fact.The average Ca concentration Results is highly correlated with those of S and Cl and, to a smaller The measurements by the electron microprobe gave diVerent extent, with that of P [Fig. 1(c)]. This level of correlation results. The linear scans made on sample S2 gave very clear might show that small amounts of Ca in the petrifying solution indications of preservation of the primordial ring but not cell were present as a mixture of calcium sulfate and chloride. structure. The scans made on samples S1 and S3 were much However, in locations with locally elevated Ca contents, the more obscure; however, careful analysis helped to reveal the correlation with Cl no longer holds.Some of these elevated essential details also. In general, the faint brown samples S1 levels of Ca are clearly associated with the presence of S and and S3 turned out to be much less ring ordered and mineralized others with P. Small inclusions are then inclusions of calcium than the dark brown sample S2.Otherwise, the primordial sulfate or phosphate, never together. In contrast, the K tissue structure was preserved only in beige-colored samples. contents are relatively well correlated with the chloride and The electron microprobe analyses revealed the main phosphate concentrations. composition of the samples, which was surprisingly uniform It was impossible to establish, on the basis of electron even on a macroscopic scale. Silicon dioxide was the most microprobe measurements, the real physico-chemical reason prevalent component, with other elements clearly noticeable, for the diVerences between the dark and bright locations on but at low levels: Cl, S, P, K and Ca.The K lines of these six the wood cross-section. They certainly followed the tree ring elements were selected after the initial selection as the separate pattern of living trees, but the chemical composition was measurement channels; one additional channel was added in totally determined in this case by the presence of SiO2.The the position of the intensive bremsstrahlung in the range great diVerences, especially in the bremsstrahlung and Si 2.9–3.0 keV. The latter channel was selected to make the signals, could be related to the variability in the density of the information more oriented to the density variations, thus material. In the same way it was diYcult to decide what kind comparable to those obtained from scattered signals in XRF.of trace heavy metal cations caused intensive coloring of the Statistically significant lines of heavy metals were absent in samples. Maybe microscopic metallic clusters were the source the electron microprobe measurements. This was striking, of the coloration. Alternatively, it was even possible that no because each sample was clearly colored, especially the dark single heavy element was responsible for coloring the samples brown sample S2. Of course, the high bremsstrahlung backand the reason lay rather in the degree of silica hydrolysis or ground restricts the possibility of the detection of an element in structural defects.if it is present in concentrations below a few hundred ppm. The scan direction had earlier been established by the use The X-ray spectra taken for all the samples were very similar, of the secondary electron image. The secondary electron image indicating that the macroscopic composition of the samples was preserved and transformed into the gray-scale image.The was very simple and uniform. Much greater diVerences were gray-scale distribution was established along the scan line. The observed in the optical branch of the spectrum, even with the raw gray-scale morphology data were noisy and only after naked eye. filtering did they reveal the internal structure. This was strictly The results of the electron microprobe measurements on the ring structure of the petrified wood as shown in Fig. 2. sample S2 are shown in Fig. 1. The Si line and bremsstrahlung The degree of qualitative similarity in the total pattern is channel contents follow the ring structure of the wood, showing striking; however, there are also serious diVerences in the the dark and light parts of the rings. The K line intensity intensity of the Si and bremsstrahlung on the one hand and follows the same pattern [Fig. 1(a)]. The level of the correlation the secondary electron signals taken from the electron micro- between the three lines showed that they could be used in an probe measurements on the other (see the right side of Fig. 2). exchangeable way to reveal the essential ring structure of the Maybe the discrepancy in the intensity of these kinds of signals tree. For example, the parameter t in Kendall statistics has is a measure of the surface quality. We should remember the value 0.596, while the probability parameter is very close that the secondary electrons are extremely sensitive to to zero, indicating a significant correlation. The Ca line topographic details.intensities follow a much more complicated pattern [Fig. 1(b)]. The analysis of the typical beige-colored sample S3 by the The average (threshold) contents of this element are correlated use of the electron microprobe did not give such conclusive well with the above-mentioned lines of the K, Si and results (Fig. 3). The count rates for the determination of the bremsstrahlung channels.On the other hand, in many scan diVerent spectral lines diVer from those in the previously positions, the levels of calcium are clearly elevated, which may described case. However, the lines of the main matrix compo- indicate some small inclusions. In this respect the behavior of nents (Si, K and the bremsstrahlung channel, expressing the this element is very similar to that in living trees.31 Even the matrix features of the samples) were much the same as for geometrical diameter of the inclusions (50–120 mm) is similar sample S2.This shows that the diVerences in matrices were to those described in real wood. It is even probable that some not great, with silica once again the main component. More part of the Ca content, just related to the elevated levels, had diVerences were seen at the level of the minor elements: the its origin in the Ca content of the primary wood material and Cl and P signals were on average two times weaker in this was not exchanged during the petrification process, but just case, the Ca signal two and a half times and the S signal four conserved.This opinion can be justified additionally by the times in comparison with the values registered for the S2 fact that inclusions are buried deeply inside relatively uniform sample. The ring structure was not as pronounced as in sample silica zones. This contradicts the supposition that the inclusions S2, even by tracing the lines of the main components. The might be the results of secondary processes.Such potential secondary processes would transform the whole of the rest of signal of Si was much clearer than the others, with the ring 438 J. Anal. At. Spectrom., 1999, 14, 435–446Fig. 1 Electron microprobe scans of the petrified wood sample S2. (a) Correlation between the signals of Si, K and the selected channel of the bremsstrahlung radiation (2.9–3.0 keV); (b) calcium levels, with the inclusions clearly noticeable; (c) association of the elevated Ca levels with the sulfate or phosphate contents.structure very clear, but the intensity of the signal dropped by real drops in the density were in the regions of the dark wood, reaching 30% of the maximum value. This was similar to the more than 20% on going to the region with much more pronounced inclusions. values observed for those rings in sample S2. It is worth noting that in sample S3 the internal splits and brittleness of the The bremsstrahlung signal was somewhat obscure when considered on its own.It was helpful to analyze the diagram material did not aVect the drops in the Si signal. The distribution of the K signal was not as correlated with the of the continuous radiation in parallel to the Si curve [Fig. 3(a)]. The bremsstrahlung signal was not as sensitive as bremsstrahlung and Si signals as in sample S2. Similarly, the Ca signal was rather discorrelated from the main component the Si signal to the changes in the granulation of the sample. This shows that the density of the matrix in the inhomogeneous signals [Fig. 3(b)].Still, sudden increases in the Ca level occurred more frequently in this part of the sample where the regions was much the same as in the rest of the material. The J. Anal. At. Spectrom., 1999, 14, 435–446 439are much more bound to the local changes in density, giving some additional substructures on top of the main ring structure. The additional substructures had a diVerent degree of correlation—sometimes the Rayleigh signal followed the changes in Si signal and, in those places, the variability of the matrix seemed to result from the changes in silica, whereas in other places the subsignals were not correlated.The diVerences in the Rayleigh and Compton spectra can be fully understood taking into account the features of these kinds of scattered radiation.46–49 On average, the ratio of the Rayleigh to the Compton signals was constant through the scan, testifying to the constant mean atomic number of the matrix, determined totally by the silica.The discrepancies are very small. Similar to the analysis by the electron microprobe, the levels Fig. 2 Gray-scale distribution of the secondary electron intensity levels of Cl and S were interrelated and they were also correlated (solid line) along the line of the scan from Fig. 1 (in steps equal to with the levels of Ca (not shown here). Especially striking was 12.5 mm)—right scale.The silicon signal is shown for comparison this correlation in the places where the levels of Ca were (dotted line) (steps 50 mm)—left scale. Electron microprobe analysis elevated in a very significant way. of sample S2. The X-ray capillary microprobe allowed the tracing of the signals of some heavy elements present in the samples of petrified wood (Ti, Fe, Cu and Zn in amounts reaching several tens of ppm and Mn, Ni and Pb at concentrations in the ppm range).All of them concentrate very clearly at the sites of the dark wood (minima in Rayleigh, Compton and Si signals). It gave the total metallic concentration at a level of many hundreds of ppm. The increase in the heavy metal contents [Fig. 4(c)] is not followed by an increase in the contents of anion-creating substances [Cl, P and S; Fig. 1(c)]. It obviously would demand much greater amounts of anionic elements than those detected. Hence it can be concluded that heavy metals in the parts of petrified wood imitating the dark wood are mainly in the oxide form or as silicates.Fe, Ni and Zn also have some smaller, irregular and grain-like inclusions in the places corresponding to the light wood in living wood, similar to the behavior of Ca. This is analogous to the results of the analysis of living wood.31,50 Additional measurements were possible by the application of the transmission technique in X-ray capillary microprobe mode.The result is shown in Fig. 4(b). The reversed transmission scan shows a striking similarity to the direct Rayleigh or Compton spectrum.27 The transmitted radiation is inversely proportional to the density of the material; if the reversal of the transmitted signal is equivalent to the scattered signal, then the latter should be directly proportional to the density. The higher intensities of scattered signals and the lower intensities of transmitted signals are coupled in sample S2 with the locations of earlier light ( less dense) real wood.This result is important and strongly confirms our hypothesis that some kinds of petrified wood are the chemical negatives of the Fig. 3 Electron microprobe scan of sample S3. (a) Correlation between Si and selected bremsstrahlung channel signals; (b) calcium signal. natural wood. The synchrotron-based X-ray capillary microprobe was also applied. Owing to the very long optical path, this device was granular structure was pronounced. The correlation between poor in detecting low-Z elements, even Si.The investigation the Ca, S and Cl levels nearly disappeared. A slightly better of heavier elements was much more eYcient, with Zn, Cu, Pb correlation could be observed between Ca and P. Much better and Ni concentrating at the locations of the primordial dark but still far from ideal was the correlation between K and all wood ring boundary [Fig. 5(b)]. In addition, the application three anions. of the wide channel selected from the background synchrotron The measurement of sample S2 by the tabletop X-ray radiation (38.5–53.5 keV) brought interesting results capillary microprobe gave another series of results (Fig. 4). [Fig. 5(a)]. The scan based on this radiation enabled us once The most important among them were the spectra of the again to trace the density structure of the petrified wood, coherently and incoherently scattered tube Mo Ka line and giving full analogy with, e.g., the Rayleigh or Compton scans the lines of some metallic components of the petrified wood.from the tabletop X-ray capillary microprobe. Because of the Rayleigh, Compton [Fig. 4(a)] and Si signals followed the ring strong contribution of the Compton (even multifold) scattered patterns and could be treated in an exchangeable way. Still radiation to the white synchrotron scattered emission, the scan there are some diVerences between the signals: the Compton resembles more the Compton than the Rayleigh scan from the signal seems to be less sensitive to the local changes in the tabletop instrument [Fig. 4(a)]. The spectral resolution of both density of the petrified wood, giving an averaged picture of the rings. The Rayleigh scattered and Si characteristic signals probes was similar (about 30 mm), as were the penetration 440 J. Anal. At. Spectrom., 1999, 14, 435–446Fig. 4 Scans of sample S2 by a tabletop X-ray capillary microprobe. (a) Rayleigh (solid) and Compton (dotted) scans; (b) Si (solid) and reversed transmitted (dotted) signals; (c) heavy metal scans (Pb, Zn, Cu).depths (order of centimeters), and did not add any diVerences identical locations. Careful analysis of Fig. 6 shows the similarities in the Si scans for the samples and the less pronounced to the scans. similarity between the bremsstrahlung and Rayleigh spectra. Comparisons between microprobes They can certainly be treated as exchangeable if the conditions of the sample preparation and analysis were set strictly the We compared the Si signals, the bremsstrahlung channel from same.Still, we can see some specific diVerences, resulting from the electron microprobe, the Rayleigh signal from the tabletop the nature of the sample and the specific features of the and the scattered white signal from the X-ray synchrotron analytical methods used in this study. One can see in the microprobe, measured for the same piece of petrified wood Rayleigh spectrum (and also in the Compton spectrum in S2.We should remember that the samples were prepared in Fig. 4) only one wide ring (second from the right). The diVerent ways (thick sample for EPMA and thin sample for bremsstrahlung and secondary electron spectra show two ring structures at the same location. Silicon spectra, taken both mXRF) and that the scans were carried out at close but not J. Anal. At. Spectrom., 1999, 14, 435–446 441Fig. 5 Synchrotron-based X-ray capillary microprobe scan of sample S2.(a) Application of the selected channel (38.5–53.5 keV) for the scan; (b) heavy element presence in the petrified wood structure. from the electron and from the X-ray microprobes, show two electron and all other electron microprobe signals are much more ‘noisy’ than the spectra from the X-ray capillary micro- rings, much better presented on the electron microprobe scan. This discrepancy can only be explained when we assume that probe.This can be explained from various points of view: for some reason the ring structure in this location is fully (i) the X-ray capillary microprobe certainly has a worse represented only in the shallow surface region, of the order of spatial resolution than the electron microprobe; hence its work the depth of the electron penetration (hence of the order of resembles the averaged electron microprobe scan; 5 mm for 20 keV electrons, used in this work). The probes, (ii) the electron and secondary electron microprobes are which are penetrating very thin layers, can easily detect it. The much more sensitive to all the minor roughnesses on the deeply penetrating X-ray microprobe averages the signals on surface of the sample, resulting both from the sample prep- going from the deep to shallow layers.For photons with an aration and from its brittleness; energy of the order of 20 keV, e.g. the Rayleigh and Compton (iii) owing to its deep penetration, the X-ray capillary photons originating from the Mo Ka radiation, the contrimicroprobe averages the influences on the signal both from bution of the micrometer deep layer is not noticeable in the the shallow and deep roughness.total balance of the detected photons. Their penetration depth in SiO2 is of the order of 8500 mm. One cannot diVerentiate For example, in the electron microprobe-based fluorescent the uppermost layers by those signals. The same holds for the and secondary electron spectra of one ring structure, one can selected channel in the synchrotron background radiation, of see a number of oscillations and jumps, which are hardly or energy between 38.5 and 53.5 keV, the radiation from which not observable in the X-ray capillary spectrum.We can can penetrate the sample by over 50 000 mm. However, by attribute them to the uneven locations on the surface or to the using the Si characteristic radiation, the mentioned additional slightly spongy structure of the material.Everything is averring structure can be detected in the X-ray capillary micro- aged by the robust manner of the data collection in X-ray probe. In this case, the signal is collected from a depth capillary microprobes, both the tabletop and synchrotrondetermined by the Si Ka (1.7 keV) escape length, thus from a based types. depth of about 30 mm. This signal is still better seen in the electron microprobe measurement, where it is collected from MicrodiVraction measurements the depth governed by the electron range.That distance (about 5 mm) is probably still better suited to the real depth of the The microdiVraction measurements of the bright and dark locations on samples S2 and S3 were made using the LURE structure than the Si Ka escape length. The next detail, which can be seen by comparison of spectra facility. As the diVerent previous microprobe measurements suggested, the silica in some forms should be by far the main from diVerent devices, is the diVerent sensitivity of signals to the local changes in the samples.On average, the secondary component of the samples. However, this matrix should vary 442 J. Anal. At. Spectrom., 1999, 14, 435–446Fig. 6 Similarities in: (a) Si scans as made by the electron and tabletop X-ray capillary microprobe; (b) bremsstrahlung scan (EPMA) and Rayleigh scan (tabletop capillary mXRF). Measurements by the electron microprobe made in 50 mm steps, by the X-ray capillary microprobe in 100 mm steps.Sample S2. substantially in density in dark and bright locations. The dark structure, while the coarse-grained silica is present, which leads to loosening of the structure. microdiVraction measurements gave roughly the same image of the main crystallographic component for dark and bright The microdiVraction measurements are in full accordance locations in both the samples [Fig. 7(b)]. This pattern was with previous independent reports.37,41 They seem to suggest identified by comparison with the ASTM Powder DiVraction that in the first phase of the petrification, the amorphous, Files as an a-quartz phase.The diVerence both between dark opal-like structure is created, then subjected to the continuous and beige wood and between dark and bright locations on recrystallization process, leading to the formation of microcryseach sample was in the intensity ratios of the lines of diVerent talline quartz.It can, but does not need to, lead to the loss of diVraction order in the spectra. We interpreted this as resulting the primary tissue structure, if earlier it has been fixed. from the diVerent ratios of the quartz phase to the unidentified, However, it is not clear at what stage of the process the tissue but probably amorphous, silica. It obviously leads to changes pattern of sample S2 was lost. Maybe the greater amount of in the density of samples. The amount of the crystalline quartz mineral components, both of metallic (Ca, Ti, Fe, Cu, Zn and phase should be much greater in the black sample S2.The others) and anionic (S, Cl and P) character, changed (acceler- amorphous silica (opal ) will be present at increased levels in ated?) the conditions of petrification and recrystallization, not the original bright locations. The diVraction patterns were allowing preservation of the original structure since the very obviously ring-like [Fig. 7(a)], suggesting the existence of the beginning. Alternatively, the tissue pattern has been damaged polycrystalline phase each time. Our concept of the structure during passage from the opal-like to the micro a-quartz of the petrified wood is as follows: structure.41 (i) fine-grained a-quartz is the main component; (ii) some amounts of amorphous opal are present in Conclusions locations on the primary light wood, increasing the density of the material; The results of the investigation fully confirmed the hypotheses about the delicate replacement of the organic tissue in wood by (iii) amounts of opal are small in locations of primordial J.Anal. At. Spectrom., 1999, 14, 435–446 443and sulfate are still uniformly included in this solid liquid, easily diVerentiated from the Ca inclusions, the latter certainly of primordial wood origin. However, we cannot suggest that the whole Ca threshold content is of external origin; probably part of it can be accounted for by intrinsic Ca presence in the wood tissue.31 Maybe the diVusion plays some, but limited, role here. The boundaries of the grain-like Ca- and Fe-bearing structures are surprisingly sharp.In general, the structure of the black petrified wood can be treated as a kind of ‘chemical negative’ of the primordial wood structure. Similar analysis of the results concerning the beige-colored samples is not so decisive, because the auxiliary signals from the electron and X-ray capillary microprobes do not supply us with such unanimous indications.Analysis with a petrological microscope did not give any indications of the secondary geological transformations of the petrified wood. Especially, there was no proof of the calcite inclusion in the petrified wood structure. Calcium, if present at elevated levels, was always in gypsum or phosphate form. The contents of all the other elements are small in comparison with SiO2. The amounts of other elements (K, S, P, Cl ) are diVerent in both kinds of samples—in the darkcolored sample S2 2–4 times greater than in the pale brown samples S1 and S3 and in the Belgian samples. It seems that a more pronounced increase in the contents of these elements would disturb the subtle equilibrium in the liquid saturating the primordial wood.Tracing the amounts of Ca seems to be especially instructive in this context. For example, increased amounts of Ca can be observed in those locations in samples S1 and S4 where the granular structure is more pronounced.The granular structure is placed inside the trunk and is Fig. 7 (a) MicrodiVraction pattern as scanned from the imaging plate. surrounded by the well-preserved ring structure. It is not clear The measurement was made for the dark wood location on the black whether the trunk was petrified only in part from the outward wood sample S2; the exciting beam from the D15 beamline was of side, the interior being intact and with organic matter inside 10 keV energy and 20 mm in size; (b) 2h scan extracted by integration obeying the normal decay process.Then the whole structure of the previous image. transformed in something like an SiO2 pipe, with the empty interior filled with additional, granular and sand-like mass the petrifying solution of silica and silicates, most probably a (see samples S1 and S4). These remnants within the internal K-rich one, with a small admixture of calcium chloride and structure of the log could be created in the non-volcano- sulfate.The presence of Na in the primary solution, probable related process of the petrification. Alternatively, maybe, the from the chemical point of view, was not confirmed by the whole structure was transformed into the petrified form with results of our analyses. Clear indications of the second petrifi- the ring structure fully formed, but later the process of rapid cation mechanism, namely by impregnation, have been found, secondary solidification around the Ca inclusions started.It because the remnants of the tissue/cellular structure have been finally damaged the ring structure inside the trunk. preserved in samples S1 and S3, easily detectable with the On average, the degree of additional, non-silica mineral optical microscope. A similar situation has been found in some contents in the dark sample S2 is much more pronounced Belgian samples. The essential structure of all the samples is than in the other samples.In addition, in that sample all the very similar, with SiO2 in an a-quartz form by far the main components can be clearly coupled with each other: locations constituent. On the boundary of rings, a large decrease in the with the inclusions of calcium sulfate are clearly observed and Si signal occurs (up to 30%). Since the additional material can be distinguished from the inclusions composed of calcium filling these places is present in distinguishable but small phosphate.The threshold amounts of Ca and K are strictly amounts (heavy metals up to the level of hundreds of ppm, coupled in sample S2 to the signals of Si, the bremsstrahlung e.g., Fe, Cu, Zn), it looks as if the density of SiO2 became and the scattered signals. This testifies to the uniform dissolu- lower in the location of the dark wood for the black-colored tion of the whole K and threshold Ca contents in silica. This fossils (originally in the fresh wood with a greater density of regularity is not so clear for samples S1 and S3.cellulose). This results from the analysis of the Si, bremsstrah- On the other hand, sample S2 seems to be mechanically lung, transmission and scattered Rayleigh and Compton sigsensitive in the locations of the singular ring boundaries. The nals (a more detailed logical signal analysis is given low-density material characteristic of such locations is brittle elsewhere51). The said petrifying liquid, composed of an and can easily be damaged, especially during the sample aqueous solution of silica and potassium silicate, with a cutting and preparation processes.These locations are relatively threshold amount of calcium chloride and sulfate, was very saturated with heavy metal contents. Thus, the losses in sensitive to the presence of any heavier cations (including Ca) material contents, fractures and even splits additionally and to the internal structure of the cellulose and lignin.The sharpen the ring boundaries. This must be taken into account cellulose in now dark-colored wood samples was substituted and demands special care during the spectral analysis. with the denser Si-bearing material where it was less dense The spectra in the systems bremsstrahlung, Si, K, Cl (for and vice versa. The Na contents, if present at all, were probably sample S2) and the Si or bremsstrahlung signal intensity (for leached from the material at the time of hardening, while some S1 and S3) versus the scan length (or scan step number) are amounts of potassium silicate were preserved.Probably now very similar to the scans of the tree rings in living trees. The it is a solid solution of potassium silicate in solidified SiO2, relevant linear scans were made with the tabletop X-ray composed of fine-grained a-quartz admixed with amorphous opal. In addition, the threshold amounts of calcium chloride capillary microprobe.27,31 The tree ring shapes revealed in the 444 J.Anal. At. Spectrom., 1999, 14, 435–446scans on living wood seemed to belong to two categories: the between the petrified and the last diagram is very pronounced. This conclusion demands checking on a more representative sharply shaped characteristic of the trees growing in the very population of samples than that which was available for this severe conditions of northern Sweden, and mildly shaped study. Especially, the possibility that such sharpening of the characteristic of the wood from trees from southern regions figure is not an artifact resulting intrinsically from the petrifi- of Sweden or from Poland.We extracted the typical shape of cation process must be excluded. Still another possibility is the singular petrified ring (Fig. 8). In Fig. 8(a), the shape of that, in the dark wood analogs of the petrified tissue, the the ring of the petrified wood, as obtained from the bremsstrahbrittleness of the materials gives some mechanical fracture, lung signal, is compared with the shape as revealed from the which itself adds to the normal density diversification, thus Si signal.The similarities are striking (the Kendall t parameter sharpening the boundary. However, in our opinion, the degree is equal to 0.269, the probability parameter for that test is of the correlation between the patterns of petrified and living 0.0878, both testifying to a significant correlation). This proves trees is so meaningful that we can consider the information that both signals can be used in a fully exchangeable way.In preserved in petrified tree rings as speaking a lot about past Fig. 8(b) and (c), the shape of the petrified wood ring is conditions. compared with the typical tree ring shape of the moderate Finding the northern type of petrified wood in Poland is climate wood (b) and northern wood (c). The similarity not unrealistic, for several reasons.For example, the climatic conditions changed periodically on the geological scale of time from the tropical to the polar. In addition, the Scandinavian glaciers, expanding from time to time, brought and left great amounts of mineralogical materials from the northern regions. The study by Francis6 of Antarctic petrified wood confirmed that there were essential climatic changes over tens or hundreds of millions of years. Future prospects As we have shown, the investigation of petrified wood should be an important supplement to recent research aimed at extracting the information included in various periodic structures: tree rings, ice annual layers, sea sediments, corals, stalactites, rings on clamshells.The analysis of tree rings seems to be particularly attractive in some respects: (i) it gives us an insight into the petrification process, and in that sense it should reveal some important details of the old and recent geological processes, especially in the regions of great geothermal activity; in particular, it can be important for the search for ‘cold’ geological processes, where the temperatures are relatively low (below and around 100 °C) and the process goes through an array of subtle equilibrium conditions;34 (ii) it shows a general pattern of tree rings, allowing attribution of the investigated pattern to the conditions of the tree growth on a 1-year long scale (microscale climatic research); (iii) in the case where a much longer sequence of the rings is available, the ring spectra can supply us with information on the climatic variations on a scale of a few hundred years; this kind of information is now intensively sought,52,53 increasing our knowledge of the climate; it would be interesting to know if the medium time-scale of changes has some repeatable patterns on the geological time-scale, going from our recent knowledge of the climate to the past; (iv) for tree rings with remnants of the tissue structure, detailed botanical studies would be possible, possibly leading to the identification of the species and to comparable investigations with the living species; the petrified structure can be easily revealed by optical and X-ray investigations, the latter allowing the addition of the elemental knowledge of the sample; some of the elemental details can be attributed to the primordial structure of the wood; (v) finally, some kind of dendrochronological application is also possible by the use of petrified wood, as has been shown by Kumagai and Fukao35 for the estimation of the volcanic activity in some areas in Japan; of course, any kind of absolute dendrochronological estimation, possible for recent trees,3–5,54,55 is excluded in the case of petrified material. Climatological research involving the information included Fig. 8 Outline of: (a) the typical petrified tree ring shape (sample S2) in petrified wood seems to be very valuable and one of not so as given by the bremsstrahlung or Si signals; (b) fresh conifer wood many possibilities of tracing the micro- and medium time- from the moderate conditions of Poland; (c) fresh conifer wood from the severe conditions of northern Sweden.scale variability of climate in geological epochs older than J. Anal. At. Spectrom., 1999, 14, 435–446 44516 G. M. Thompson, D. N. Lumsden, R. L.Walker and J. A. Carter, several million years. For such distant times, sea sediments, Geochim.Cosmochim. Acta, 1975, 39, 1211. ice cores and corals lose much of their usefulness whereas 17 P. Holden, University of California, Santa Cruz, personal com- petrified wood does not. Of course, we should keep in mind munication, 1998. that the methodology of research should be modified in 18 I. J. Winograd, T. B. Coplen, J. M. Landwehr, A. C. Riggs, comparison with recent dendrological research. In the latter K. R. Ludwig, B. J. Szabo, P. T. Kolesar and K. M. Revesz, Science, 1992, 258, 255.field, the search of long sequences of rings, matching the 19 K. R. Ludwig, K. R. Simmons, B. J. Szabo, I. J. Winograd, J. M. overlapping sequences and scaling by comparison with climato- Landwehr, A. C. Riggs and R. J. HoVman, Science, 1992, 258, logical databases are all possible. Meanwhile, new techniques 284. will be demanded for the application of tree ring knowledge 20 R. L. Edwards, H. Cheng, M. T. Murrell and S. J. Goldstein, for petrified wood. The short- and long-scale internal stan- Science, 1997, 276, 782.dardization of the ring width and amplitude will probably 21 J. M. Vadillo, I. Vadillo, F. Carasco and J. J. Laserna, Fresenius’ J. Anal. Chem., 1998, 361, 119. play a key role in this new research. 22 R. B. Alley, D. A. Meese, C. A. Shuman, A. J. Gow, K. C. Taylor, The additional merit of our approach lies in establishing for P. M. Grootes, J. W. C. White, M. Ram, E. D. Waddington, the first time that at least three major branches of microspectro- P.A. Mayewski and G. A. Zielinski, Nature (London), 1993, 362, scopy, viz., electron microprobe, X-ray capillary microprobe 527. and optical microprobe, may contribute to research in the 23 W. Dansgaard, S. Jjohnsen, H. B. Clausen, D. Dahl-Jensen, field. Moreover, these methods complement each other, giving N. S. Gundestrup, C. U. Hammer, C. S. Hvidberg, J. P. SteVensen, A. E. Sveinbjo� rnsdottir, J. Jouzel and G. Bond, Nature (London), in many places analogous and in other places, more dependent 1993, 364, 218.on the surface conditions, diVerent information. Especially, the 24 G. A. Zielinski and M. S. Germani, J. Archeol. Sci., 1998, 25, 279. auxiliary signals (bremsstrahlung in electron, scattered radi- 25 D. W. Oppo, J. F. McManus and J. L. Cullen, Science, 1998, ation in X-ray microprobes) give similar, density/wood struc- 279, 1335. ture-related information. Elemental information is clearly 26 H. Schulz, U.von Rad and H. Erlenkeuser, Nature (London), complementary—about the light elements from electron 1998, 393, 54. 27 A. Kuczumow, A. Rindby and S. Larsson, X-Ray Spectrom., 1995, and about the heavy elements from X-ray micro- 24, 19. probes. In our opinion, further investigation of the subject by 28 K. Pernesta°l, B. Jonsson and J. E. Ha�llgren, Nucl. Instrum. the application of recent methods, supplemented with isotopic Methods, Sect. B, 1993, 75, 326. investigations and dating, should bring many other 29 S.A. E. Johansson, Analyst, 1992, 117, 259. interesting results. 30 G. Loevestam, E. M. Johansson, S. Johansson and J. Pallon, Ambio, 1990, 19, 87. 31 A. Kuczumow, S. Larsson and A. Rindby, X-Ray Spectrom., 1996, Acknowledgement 25, 147. 32 A. M. Keller and M. S. Hendrix, Palaios, 1997, 12, 282. The authors are deeply grateful to Professor M. Matejak and 33 G. T. Creber and W. G. Chaloner, Palaeogeogr. Palaeoclimatol. Eng. M. Jablonski of the Main School of Farming, Agriculture Palaeoecol., 1985, 52, 35.Academy, Warsaw, and to Eng. Z. Kostrzewski of Zamosc, 34 S. Carroll, E. Mroczek, M. Alai and M. Ebert, Geochim. Poland, for supplying samples of petrified wood. We also Cosmochim. Acta, 1998, 62, 1379. 35 H. Kumagai and Y. Fukao, Geophys. Res. Lett., 1992, 19, 1859. thank Mr. H. Werner, a private collector at Hoevenen, 36 H. Hicks, US Pat., 4 612 050, 1986. Belgium, for the generous donation of samples from his 37 G. Scurfield and E. R. Segnit, Sediment. Geol., 1984, 39, 149. collection. A. Kuczumow acknowledges a research grant from 38 R. W. Drum, Science, 1968, 161, 175. the Belgium OYce for Scientific, Technical and Cultural 39 J. H. Oehler, Geol. Soc. Am. Bull., 1976, 87, 1143. AVairs. 40 A. C. Sigleo, Geochim. Cosmochim. Acta, 1978, 42, 1397. 41 C. Stein, J. Sediment. Petrol., 1982, 52, 1277. 42 K. Janssens, L. Vincze, J. Rubio and F. Adams, J. Anal. At. References Spectrom., 1994, 9, 151. 43 K. Janssens, B. Vekemans, L. Vincze, F. Adams and A. Rindby, 1 Methods of Dendrochronology, ed. E. R. Cook and L. A. Spectrochim. Acta, Part B, 1996, 51, 1661. Kairiukstis, Kluwer, Dordrecht, 1990. 44 P. Chevallier, P. Dhez, F. Legrand, A. Erko, Y. Agafonov, 2 F. H. Schweingruber, Tree Rings Basics and Applications of L. A. Panchenko and A. Yakshin, J. Trace Microprobe Technol., Dendrochronology, Kluwer, Dordrecht, 1988. 1996, 14, 517. 3 P. I. Kuniholm, B. Kromer, S. W. Manning, M. Newton, 45 P. Dillman, P. Populus, P. Chevallier, P. Fluzin, G. Beranger and C. E. Latini and M. J. Bruce, Nature (London), 1996, 381, 780. A. Firsov, J. Trace Microprobe Technol., 1997, 15, 251. 4 C. Renfrew, Nature (London), 1996, 381, 733. 46 G. Andermann and J. W. Kemp, Anal. Chem., 1958, 30, 1306. 5 H. J. Bruins and J. van der Plicht, Nature (London), 1996, 382, 213. 47 H. A. van Sprang and M. H. J. Bekkers, X-Ray Spectrom., 1998, 6 J. E. Francis, Palaentology, 1986, 29, 665. 27, 31. 7 H. C. Fritts, Tree Rings and Climate, Academic Press, London, 48 P. M. Van Dyck and R. E. Van Grieken, Anal. Chem., 1980, 1976. 52, 1859. 8 H. C. Fritts, Reconstructing Large-Scale Climatic Patterns from 49 M. F. Arau` jo, P. Van Espen and R. Van Grieken, X-Ray Tree-Ring Data, University of Arizona Press, Tucson, AZ, 1991. Spectrom., 1990, 19, 29. 9 K.R.BriVa, P. D. Jones, F. H. Schweingruber and T. J. Osborn, 50 S. A. E. Johansson, Endeavour, 1989, 13, 48. Nature (London), 1998, 393, 450. 51 A. Kuczumow, B. Vekemans, M. Claes, O. Schalm, L. Vincze, 10 J. W. Back, J. Recy, F. Taylor, R. L. Edwards and G. Cabioch, W. Dorrine�, K. Gysels and R. Van Grieken, X-Ray Spectrom., Nature (London), 1997, 385, 705. accepted for publication. 11 J. J. Labrecque and P. A. Rosales, Spectrochim. Acta, Part B, 52 G. Bond and R. Lotti, Science, 1995, 267, 1005. 1997, 52, 1645. 53 M. E. Raymo, K. Ganley, S. Carter, D. W. Oppo and 12 W. Beck, Science, 1998, 279, 1003. J. McManus, Nature (London), 1998, 392, 699. 13 M. K. Gagan, L. K. AyliVe, D. Hopley, J. A. Cali, 54 D. W. Stahle, M. K. Clevealand, D. B. Blanton, M. D. Therell G. E. Mortimer, J. Chappell, M. T. McCulloch and M. J. Head, and D. A. Gay, Science, 1998, 280, 564. Science, 1998, 279, 1014. 55 S. L. de Silva and G. A. Zielinski, Nature (London), 1998, 393, 455. 14 I. R. Quitmyer, D. S. Jones and W. S. Arnold, J. Archeol. Sci., 1997, 24, 825. 15 S. R. Hart and J. Blusztajn, Science, 1998, 280, 883. Paper 8/06748A 446 J. Anal. At. Spectrom., 1999, 14, 435&nda

ISSN:0267-9477    

DOI:10.1039/a806748a

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Electron probe microanalysis of irradiated nuclear fuel: an overview† Clive Walker The European Commission, Joint Research Centre, Institute for Transuranium Elements, P.O. Box 2340, D-76125 Karlsruhe, Germany Received 28th August 1998, Accepted 29th October 1998 This paper focuses on the features that set apart the electron probe microanalysis (EPMA) of highly radioactive material from standard EPMA. In addition, it describes the diYculties encountered in the analysis of irradiated nuclear fuel and explains how certain problems have been solved or at least overcome.The paper then presents specific examples showing how EPMA is currently utilized in nuclear fuel research. It is shown that, despite the use of incorrect standards for certain elements, problems arising from X-ray line interference and uncertainties in the matrix correction, EPMA is still able to deliver basic data on chemical composition which are of fundamental importance in understanding the in-pile behaviour of nuclear fuel. 1 Introduction 2 Handling samples of nuclear fuel Irradiated nuclear fuel emits a-rays and high energy b- and There has always been a synergetic relationship between electron probe microanalysis (EPMA) and the nuclear power c-rays. As a result, specimen preparation (cutting, grinding and polishing) must be carried out in a hot cell using manipu- industry. Both were born in the 1950s, and in the 1960s EPMA and the civil use of nuclear power developed side by side.This lators and the microprobe must be shielded with heavy metal. In addition, an extra step, decontamination, is introduced in was no coincidence. The small volume of material analysed and the ease of quantification made EPMA the ideal analytical the specimen preparation step to avoid carry-over of radioactive particles to the clean areas of the laboratory as well as tool with which to study the chemical behaviour of fission products in nuclear fuel. Today, EPMA is helping to maintain the specimen chamber and electron column of the machine.Decontamination involves light ion etching of the specimen to the high safety standards of the nuclear industry by providing fundamental information about the behaviour of nuclear fuel remove the fission product Cs smeared over the surface during polishing and repeated washing of the specimen in ultra- under extreme irradiation conditions and about the performance of new fuels designed for the in-pile incineration of sonically agitated absolute ethanol.This removes, along with any remaining Cs, fuel particles loosely attached to the nuclear waste. EPMA of irradiated nuclear fuel is not straightforward and specimen and specimen holder. The heavy metal shielding on the microprobe falls into two diVers in a number of ways from standard EPMA. First, the machine must be shielded with heavy metal to protect the categories: biological shielding to protect the operator from radiation exposure and detector shielding to reduce the X-ray operator from radiation exposure, and detectors and preamplifiers must be shielded or located in out-of-the-way places background generated by b- and c-rays. Fig. 1 shows the shielded electron probe analyser at the European Institute for in order to reduce the radiation background. Second, it involves the determination of the ultra-heavy elements U, Np, Transuranium Elements. The machine, which is unique, combines the high performance crystal spectrometers of the Pu, Am and Cm.For these elements, the M X-ray lines are used (the critical excitation potential, Ec, is 2–5 kV compared Cameca MS46 analyser with the three lens electron optical column of the CAMEBAX analyser. Biological shielding has with 16–19 kV for the L-lines). The use of the M-lines in EPMA is plagued by a number of problems. These include been added to the specimen carriage and to the original MS46 chassis of the machine.This shielding, which consists of plates X-ray line interference, uncertainties about the values of the mass absorption coeYcients and a lack of knowledge about of lead and denal (W 95 wt%, Cu or Fe 5 wt%; density 19 g cm-3), enables the examination of fuel specimens with the fluorescence yields. This paper focuses on the features that set apart the EPMA activities of 75 GBq to be carried out with full protection. To reduce the level of radiation impinging on the X-ray of highly radioactive material from standard EPMA.In addition, it describes the diYculties encountered in the analysis detectors, plates and blocks of denal up to 13 cm thick have been placed between the specimen chamber and the spec- of irradiated nuclear fuel and explains how certain problems have been solved or at least overcome. The paper then presents trometer chambers of the machine. In addition, the prospecific examples showing how EPMA is currently utilized in portional counters have been fitted with lead collimators nuclear fuel research.It is shown that, despite the use of aligned on the active area of the diVracting crystals. With the incorrect standards for certain elements, problems arising spectrometer used for light element analysis, the window from X-ray line interference and uncertainties in the matrix assembly supplied by Cameca has been replaced by a collicorrection, EPMA is still able to deliver basic data on chemical mated one of denal.Further, magnets (field strength 0.5 kG) composition which are of fundamental importance in have been installed before the mylar windows to repel b-rays understanding the in-pile behaviour of nuclear fuel. and back-scattered electrons. Fig. 2 shows the influence of the shielding and supplementary pulse height discrimination on the background level obtained when working with a penta- †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.erythritol (PET) crystal. The data are for a fuel specimen J. Anal. At. Spectrom., 1999, 14, 447–454 447Fig. 3 A specimen of UO2 nuclear fuel prepared for EPMA. The fuel is highly porous and contains many cracks. Fig. 1 View of the shielded electron microprobe at the Institute for Transuranium Elements. (v) Oxide fuels based on mixtures of UO2, NpO2 , PuO2 and AmO2 designed for the transmutation of Am and Np.5 After irradiation not only are these materials highly radioactive, but they are also highly inhomogeneous.Fuels that have seen high reactor powers or have been irradiated to high burn-up are generally highly porous. They also contain many microscopic and sub-microscopic fission product inclusions, the most common of which are metallic inclusions containing Mo, Tc, Ru, Rh and Pd. Fig. 3 shows the appearance of a UO2 fuel specimen taken for EPMA. The fuel had been submitted to a transient test in which the reactor power was rapidly raised (in a matter of hours) to an unusually high level of 42 kW m-1.It can be seen that the fuel is not only porous but also contains many cracks. Hence, care is needed in positioning the electron beam. At Karlsruhe the specimen current image (absorbed electron image) is used to position the electron beam with an accuracy of better than 5 mm. 4 EPMA procedures for nuclear fuel The EPMA of nuclear fuel falls within four distinct categories.(i) Measurement of the radial distribution of solid, gaseous and volatile fission products contained in the fuel matrix in Fig. 2 EVect of shielding and pulse height discrimination on the concentrations smaller than 1 wt% (mainly Nd, Xe and Cs). radiation background when working with a PET crystal. The c dose (ii) Determination of the composition of the metallic fission rate from the fuel specimen at 30 cm distance was 1.5 mSv h-1. product inclusions containing Mo, Tc, Rh, Ru and Pd.(iii) Analysis of the radial distributions of the actinide elements U, Np, Pu and Am. In some fuels these elements are delivering a c dose rate of 1.5 mSv h-1 at 30 cm distance. It can be seen that the attenuation factor for the shielding is major constituents with concentrations ranging from 20 to 90 wt%, but in irradiated UO2 fuel the concentrations of Pu about 25 and that the use of electronic discrimination decreases the background further by an additional factor of 6.(generally <1 wt%), Np and Am (0.05 wt% max.) are small. (iv) Determination of the composition of oxide material in the fuel–cladding gap (mainly ZrO2 in light water reactor fuel 3 Materials analysed rods and Cr2O3 in fast breeder reactor fuel rods). The analysis conditions used for some of the most abundant In recent years five types of nuclear fuel have been examined in the EPMA laboratory at the Institute for Transuranium fission products and for the actinide elements in irradiated nuclear fuel at the Institute for Transuranium Elements are Elements.(i) Conventional UO2 fuel irradiated under steady-state and summarised in Table 1. The La1 characteristic lines are used for the fission products and the Ma1 lines for the actinides. transient conditions.1,2 (ii) UO2 fuel doped with Gd as a neutron poison. Where there is line interference the beta line is used instead of the alpha line. A high electron acceleration potential (E0) and (iii) MOX fuel with a duplex structure consisting of (U, Pu)O2 agglomerates in a UO2 matrix.3 a high beam current are used for the analysis of fission products dissolved in the fuel matrix.This ensures good (iv) U–Pu–Zr alloy fuel containing small additions of Np, Am and Cm and the rare earths Nd, Ce and Gd.4 counting statistics, a low detection limit and that the depth of 448 J. Anal. At. Spectrom., 1999, 14, 447–454Table 1 Analysis conditions for the most abundant fission products (i.e., total peak count duration 5 min, total background count and for the actinide elements in irradiated nuclear fuel at the Institute duration nearly 7 min).Under these conditions, the confidence for Transuranium Elementsa limit on the measured concentrations at a significance level of 99% is about 5% relative at 0.5 wt% and 10–20% relative at X-ray DiVracting E0/ Beam 0.05 wt%. The limit of detection lies in the range 250–750 ppm. Element line crystal kV current/nA Standard Fission products— 4.2 Determination of fission gas xenon Zr La1 PET 25 250 Zr Mo La1 PET 15 100 Mo The determination of Xe is complicated by the fact that part Tc La1 PET 15 100 Tc of the gas retained in the fuel matrix is contained in small Ru La1 PET 15 100 Ru bubbles, the size and distribution of which aVect the emitted Rh La1 PET 15 100 Rh X-ray intensity.7 To take this into account, a special correction Pd Lb1 PET 15 100 Pd procedure has been developed that considers the probability Te Lb1 LiF 25 250 Te I La1 LiF 25 250 CsI of an electron/gas bubble interaction at a given depth z, Xe La1 Quartz 1011 25 250 Sb dPn(z), the depth distribution of X-ray production, w(rz), and Cs Lb1 LiF 25 250 CsI the influence of gas density, dg.8 Integration is first made over Ba La1 Quartz 1011 25 250 BaCO3 the bubble diameter, 2r, and then over the electron penetration La La1 LiF 25 250 La depth, zmax.For randomly dispersed gas bubbles the emitted Ce La1 (Lb1) LiF 25 250 Ce X-ray intensity, I, is found by summing the X-ray contributions Nd La1 LiF 25 250 Nd Gd La1 LIF 25 250 Gd from the bubbles in each size class.Thus, Fuel components— U Ma1 Quartz 1011 20 (25) 100 UO2 Np Ma1 Quartz 1011 20 (25) 100 Np I=. 2 n=0 Pzmax z=0 P2r m=0 dPn(z)·w(rz)·Ar(x)·d(dgm) Pu Mb Quartz 1011 20 (25) 100 PuO2 Am Mb Quartz 1011 20 (25) 100 PuO2 Cm Mb Quartz 1011 20 (25) 100 PuO2 where Ar(x) is the correction factor for X-ray absorption, n aThe Lb1 line of Ce is used in the presence of Ba.An acceleration is the spatial configuration index for each size class and m is potential of 25 kV is used when the actinides are analysed together the depth of the emitting gas layer within the bubble. with the fission products Xe, Cs and Nd. The correction model predicts that when the bubble size exceeds about 10 nm, the X-ray intensity falls.7 For example, the X-ray intensity emitted from bubbles with a radius of electron penetration is suYcient to avoid surface eVects. For 100 nm is 50% of that obtained from an atomic dispersion of the metallic fission product inclusions, 15 kV is used to restrict gas.This is the finding when the Xe concentration is fixed at electron spread. As for standards, the pure metal is used 1.14×10-4 mol cm-3 and the gas pressure is assumed to equal whenever possible, even for the rare earths. (Nuclear labora- 2c/r. Under these conditions, the volume fraction of bubbles tories are equipped with glove boxes with N2 atmospheres.(swelling) is approximately 0.2%. The gas concentration and Hence, the conservation of metals which are highly oxidising volume fraction of bubbles are typical for fast breeder reactor does not present a problem.) Compound standards are used fuel at a burn-up of 1% fissions per initial metal atoms for Cs, I, Ba, U and Pu. Xenon, Am and Cm constitute special (FIMA). cases as far as the standards used are concerned.An Sb standard is used in the determination of Xe and a PuO2 5 DiYculties encountered in the analysis of nuclear standard is used for the determination of Am and Cm (see section 5.1). fuel Concerning the choice of diVracting crystal for the analysis 5.1 Problems with standards of the actinides, it can be seen from Table 1 that a quartz 1011 crystal with a d-spacing of 0.3343 nm is used for these elements Xenon. A suitable standard for Xe has so far not been at the Institute for Transuranium Elements.Modern auto- identified. To overcome this problem, an Sb standard is used mated microprobes, however, such as the Cameca SX 50 and for Xe and a correction factor is applied to the intensity of SX 100 machines are not equipped with this crystal. This is the Sb La1 line. The correction factor is derived by interpolatbecause the X-ray lines obtained with the 1011 crystal are so ing the intensity for pure solid Xe from the intensities for narrow that it is diYcult to find the peak position with adjacent elements in the Periodic Table.The calibration graph automated spectrometers. Consequently, a PET diVracting for 15 kV is shown in Fig. 4. It can be seen that the intensity crystal is employed for the analysis of the actinide elements of the La1 line increases linearly with atomic number in the when the microprobe is automated. Although the PET crystal range of interest and that the X-ray count rate for Xe is about has a much higher sensitivity than the quartz 1011 crystal, its 1.3 times higher than that obtained from Sb.spectral resolution is lower.6 Hence, better separation of the Pu Ma1 and U Mb lines is obtained with a quartz 1011 crystal. Caesium. A CsI standard is used for the determination of Cs. Like other alkali metals, Cs is highly mobile around the 4.1 Analysis of fission products in the fuel matrix point of impact of the electron probe. Caesium iodide has a low thermal conductivity and hence under the harsh conditions The procedure adopted for the analysis of fission products in the fuel matrix follows the rules for trace analysis which used for fission product analysis (25 kV and 250 nA) the temperature in the vicinity of the beam rises by at least 100 °C.prescribe that the net peak count from the standard and the counting time for the background should be as large as As a result, the intensity of the Cs La1 line falls continuously when the counting time is prolonged beyond 20 s.To obtain practically possible to ensure a low detection limit. A high peak-to-background ratio is achieved by the use of a high a stable X-ray signal, the standard is therefore analysed at a beam current of 50 nA and the counting time is restricted to electron acceleration voltage of 25 kV and a high beam current of 250 nA. As to the counting time, fission product concen- 10 s. The standard net peak count rate measured under these conditions is then increased by a factor of 5 to obtain the trations are calculated from k-values that are the average of six peak and eight background determinations of 50 s duration X-ray intensity at a beam current of 250 nA.J. Anal. At. Spectrom., 1999, 14, 447–454 449for U, Np, Pu and Am obtained by Kleykamp9 are shown in Fig. 6 and the wavelengths and relative intensities (r.i.) of the strongest lines are listed in Table 2, which is also taken from Kleykamp’s paper. It is evident that the U Mb line interferes with the Pu Ma1 line and that the Np Mb line interferes with the Am Ma1 line.The data for Cm are not contained in Fig. 6 and Table 2, but, according to Kleykamp,10 the wavelength of the Cm Ma1 line is 350.7 pm and that of the Cm Mb line is 331.7 pm. Thus, the U Mc2 and Pu Mb lines overlap with the Cm Ma1 line. As indicated above, the U Mc2 line interferes with the Pu Mb line used for the determination of Pu. Although the Mc2 line is a relatively minor one (r.i.is 2), this interference has to be considered in situations where the concentration of U is high and the concentration of Pu low (e.g., in the determination of Pu in UO2 fuel ). In this case the following procedure is adopted which can also be employed for other interferences. Fig. 4 Relative intensity of the La1 X-ray line related to the atomic The spectrometer is positioned on the Pu Mb line and the number of the elements from In to Nd. The X-ray intensity measured X-ray intensity obtained from UO2 at this setting recorded. on Sb is taken as unity.E0=15 kV; beam current, 50 nA; quartz 1011 At 25 kV with a quartz 1011 crystal, about 100 C s-1 mA-1 is diVracting crystal. registered which corresponds to roughly 0.8 wt% Pu. The true percentage of Pu at each point analysed is then derived by Americium and curium. Owing to problems in handling these subtracting from the measured Pu concentration the amount highly radiotoxic substances, a PuO2 standard is used for the obtained when 0.8 wt% is multiplied by the mass fraction of determination of Am and Cm.Fig. 5 shows the variation in UO2 present. the emitted intensity of the Ma1 and Mb lines from UO2, NpO2, PuO2, AmO2 and CmO2 at 20 kV when working with a quartz crystal. It can be seen that, whereas the intensity of 5.3 Uncertainties in the matrix correction the Ma1 line increases linearly with the atomic number, Z, the intensity of the Mb line initially drops on going from U to The conventional matrix correction used to convert the Np and then increases slightly.The intensities of the Ma1 and measured k-ratios to mass concentrations has two important Mb lines from AmO2 and CmO2 may be read from the shortcomings as far as the analysis of nuclear fuel is concerned. extrapolated lines. It is found that the intensity of the Am Mb (i) The mass absorption coeYcients (MACs) for the M-lines line is 5% higher and the intensity of the Cm Mb line is 8% of Np, Pu, Am and Cm and the MACs for the absorption of higher than the intensity of the Pu Mb line.Hence, for the the fission product L-lines in these elements are suspect. determination of Am and Cm at 20 kV, the Mb intensity from (ii) The correction does not consider K–M, L–M or M–M the PuO2 standard is increased by these percentages. characteristic fluorescence. The reason for the inflection in the Mb intensity curve at These deficiencies do not generally produce gross errors in Np is not known.It is, however, apparently not related to the EPMA results, although they are a source of uncertainty. either absorption in the counter gas or absorption in the At the Institute for Transuranium Elements the matrix beryllium window of the spectrometer. correction is made with the QUAD2 program of Farthing et al.,11 which is based on the Quadrilateral Model of Scott 5.2 M-line interference and Love.12 In this program, the MACs are calculated from the most recent algorithms of Heinrich13 and extended from X-Ray line interference is fairly common amongst the M-lines Z=92 to 96 using the line energies recommended by of the actinides and can be troublesome when the specimen Kleykamp.9,10 Heinrich recognises that the MACs provided contains two or more actinide elements.M-line X-ray spectra by these algorithms are unsafe if the line lies between -5 and +20 eV from an absorption edge or if the energy of the line is less than the energy of the Mv edge of the absorber.MACs for the Ma1 lines of the most frequently encountered actinide elements from tables compiled by Farthing and Walker14 are given in Table 3. Since the M lines are not widely used in EPMA, K–M, L–M and M–M fluorescence were disregarded when the correction for characteristic fluorescence was formulated. Fluorescence excitation of the M lines, however, is generally considered to be insignificant. This assumption is supported by the values of the M-shell fluorescence yield (vM), which are reported to be small compared with the K and LIII fluorescence yields.15 Moreover, in the EPMA of irradiated fuel the failure to consider M-line fluorescence generally carries no practical disadvantage.There appear to be two reasons for this. First, because the fission products (the main emitters of L X-rays in the fuel ) are usually present in low concentrations (<1 wt%). Second, because E0 is only slightly higher than Ec for the actinide L lines [at 25 kV, E0/Ec varies from 1.3 (for Cm) to 1.5 (for U)].Fig. 5 Variation in the emitted intensity of the Ma1 and Mb lines Table 4 shows EPMA results for U–Pu–Zr alloy fuel from UO2, NpO2, PuO2, AmO2 and CmO2. E0=20 kV; beam current, 100 nA; quartz 1011 diVracting crystal. containing small amounts of Np, Am and Cm. It can be seen 450 J. Anal. At. Spectrom., 1999, 14, 447–454Fig. 6 M-line spectra for U, Np, Pu and Am (after Kleykamp9).Table 2 Wavelength, l (pm), and the relative intensity, r.i. (%), for whereas the correction factors for Am and Cm approach 2.0. the M lines of U, Np, Pu and Am (after Kleykamp9)a The diVerence in the magnitude of the correction factors can be traced to the X-ray absorption correction which was U Np Pu Am considerably larger for Am and Cm. Moreover, the low concentrations measured for these elements appear to be a X-ray line l r.i. l r.i. l r.i. l r.i. consequence of the formation of AmxCmy inclusions rather MIINIII 275.4 1 — <1 258.1 <1 — <1 than of any deficiency in the correction program.MIINIV 281.9 2 272.3 1 264.1 1 255.5 3 MINII 290.7 1 — <1 — <1 — <1 MIIOIV,V 295.1 5 286.2 2 278.3 3 270.4 3 6 Some current applications MIINI 333.1 2 322.2 2 312.7 3 303.0 6 MIIINV (c1) 347.9 13 338.7 3 329.5 6 320.6 7 6.1 Determination of the local burn-up NIVOII 352.0 2 342.5 <1 333.7 1 324.9 1 MIIINIV (c2) 357.6 1 — <1 336.2 <1 — <1 Burn-up is defined as the cumulative output of heat from a MIVNVI (b) 371.6 180 360.8 72 351.0 67 341.3 64 nuclear fuel and is directly proportional to the fraction of MVNVII (a1) 391.0 100 380.0 100 370.1 100 360.2 100 heavy metal atoms fissioned.Burn-up is commonly expressed MIIINI 433.0 0.5 421.5 0.2 412.2 0.2 402.2 0.2 in gigawatt days per tonne of uranium. The fission product MVNIII (j1) 494.3 0.9 480.4 0.3 467.2 0.2 454.3 0.2 MVINII (j2) 504.9 0.6 491.3 0.2 478.4 0.2 466.2 0.2 Nd can be used as an indicator of the local burn-up because it is immobile in the fuel.(It remains in the lattice site where aThe relative intensities of the lines are machine-specific. it comes to rest following fission.) Consequently, the local concentration of Nd increases almost linearly with the local burn-up as shown in Fig. 7. that, despite the absence of a correction for fluorescence excitation of the Ma1 and Mb lines, the QUAD2 program Fig. 8 shows the radial distribution of Nd in a UO2 fuel with a cross-section average burn-up of 77.5 GWd/tU. The delivers concentrations that are reasonably precise. It can also be seen that the measured k-ratios sum to 87.95 and that after Nd concentration profile indicates that in the body of the fuel the local burn-up is slightly less than the average value, correction the concentrations obtained sum to 100.53 wt%. Remarkably, the ZAF (atomic number, absorption, fluores- whereas at the fuel pellet surface the local burn-up is around 2.5 times higher than the average burn-up.Before EPMA was cence) correction factors for U, Np and Pu are close to 1.0, J. Anal. At. Spectrom., 1999, 14, 447–454 451Table 3 Mass absorption coeYcients for the Ma1 lines of selected actinide elements (after Farthing and Walker14) Emitter Absorber Z Th U Np Pu Am Cm Be 4 19.92 16.57 15.10 13.92 12.69 11.75 B 5 45.06 37.67 34.42 31.80 29.08 26.97 C 6 87.64 73.79 67.67 62.71 57.55 53.54 N 7 143.07 120.68 110.75 102.70 94.32 78.79 O 8 216.54 183.08 168.20 156.11 143.52 133.69 Na 11 522.31 455.62 411.19 383.06 353.59 330.46 Al 13 814.34 699.12 647.09 604.43 559.58 524.27 Kr 36 1377.83 1190.52 1105.54 1035.68 962.04 903.90 Sr 37 1605.80 1390.34 1292.40 1211.78 1126.69 1059.44 Y 39 1737.51 1505.93 1400.56 1313.78 1222.13 1149.64 Zr 40 1852.96 1607.65 1459.93 1403.87 1306.57 1229.58 Mo 42 2089.73 1816.81 1692.28 1589.54 1408.85 1394.74 Tc 43 1903.38 2069.78 1799.21 1690.73 1575.90 1484.88 Ru 44 1995.85 1738.70 1889.47 1776.33 1656.50 1561.47 Rh 45 438.26 1843.84 1720.06 1617.77 1759.18 1658.94 Pd 46 469.13 405.68 1351.48 1686.48 1574.25 1485.15 Te 52 698.33 603.42 560.45 525.17 488.02 458.72 Xe 54 810.55 700.83 651.08 610.20 567.12 533.12 Cs 55 872.00 754.29 700.88 656.96 610.67 574.12 Ba 56 916.88 793.51 737.49 691.40 642.80 604.41 La 57 983.50 850.78 790.91 741.64 689.65 648.57 Ce 58 1053.14 912.50 848.52 795.84 740.23 696.26 Nd 60 1187.63 1030.36 958.70 899.63 837.21 787.82 Gd 64 1413.09 1231.49 1148.25 1079.41 1006.42 948.49 W 74 2045.91 1798.84 1648.47 1589.32 1487.85 1406.86 Au 79 2118.51 2027.92 1878.08 1755.64 1761.56 1667.13 Bi 83 2039.89 1753.72 2044.66 1912.16 1773.17 1663.96 Th 90 785.20 683.76 637.49 1238.14 1212.76 1780.32 U 92 828.41 720.83 671.74 631.25 588.42 554.50 Np 93 856.93 745.43 694.54 652.56 608.15 572.97 Pu 94 850.62 739.77 689.16 647.41 603.23 568.23 Am 95 864.17 751.40 669.91 657.42 612.46 576.84 Cm 96 849.98 738.96 688.25 646.39 602.10 567.01 Table 4 EPMA results for a U–Pu–Zr alloy fuel containing small amounts of Np, Am and Cma ZAF Specified X-ray k correction Concentration concentration Element line (%) factor (wt%) (wt%) U Ma1 60.5 1.094 66.2 66.0 Pu Mb 20.3 1.047 21.3 19.0 Zr La1 5.4 1.809 9.8 10.0 Np Ma1 0.7 1.063 1.8 3.0 Am Mb 0.9 1.418 1.3 1.6 Cm Mb 0.1 1.809 0.2 0.4 Total 87.9 100.5 aE0=20 kV; beam current=100 nA.Matrix correction program, QUAD2 (ref. 11). used for local burn-up analysis these facts were not fully Fig. 7 Local concentration of Nd in UO2 nuclear fuel as a function appreciated. of the local burn-up. The broken line represents the predicted relationship. 6.2 Fission gas analysis The noble gas Xe is one of the most abundant fission products radial distribution of the fission gas Xe in UO2 fuel with the aim of obtaining a better understanding of the mechanisms formed in nuclear fuel. Under normal steady-state irradiation conditions the gas atoms are evenly dispersed in the UO2 controlling gas release.16 Fig. 10 shows the radial distribution of Xe in a nuclear fuel after irradiation under normal steady- lattice and consequently cause few problems. If the fuel temperature is raised, however, say in response to a reactor state conditions and following a reactor power excursion which raised the temperature at the fuel centre by 500–600 °C. Little power excursion, the Xe atoms become highly mobile.When this happens part of the gas precipitates in the fuel as bubbles or no gas was released from the fuel irradiated under normal irradiation conditions and, as a result, the Xe concentration (Fig. 9) and part is released to the free space in the fuel rod. The result is a deterioration in fuel performance. profile follows the radial power profile in the fuel. However, a substantial amount of gas has been released from the UO2 EPMA is being used to measure the local concentration and 452 J.Anal. At. Spectrom., 1999, 14, 447–454Fig. 8 EPMA point analysis results for the radial distribution of Nd in a section of UO2 nuclear fuel with an average burn-up of 77.5 GWd/tU. The local burn-up increases sharply at the fuel surface owing to the fission of Pu created by neutron capture. Fig. 10 EPMA point analysis results for the radial distribution of Xe in UO2 fuel. Under normal irradiation conditions no gas is released. During a reactor power excursion a substantial amount of gas may be released in the central region of the fuel.Fig. 9 Scanning electron micrograph of UO2 nuclear fuel showing fission gas bubbles on grain faces following a reactor power excursion. grains in the central region of the fuel that had experienced the reactor power excursion. In this case, the microprobe profile exhibits a sharp fall to very low Xe concentrations at a distance of 4 mm from the fuel centre. 6.3 Characterisation of irradiated MOX fuel A number of countries have stockpiles of Pu from the reprocessing of light water reactor fuel.One way to reduce these stores of separated Pu is to mix a few per cent with UO2 producing a mixed oxide (MOX) fuel and to burn this in a light water reactor. So far, the MOX fuel examined at Karlsruhe has consisted of MOX agglomerates less than 100 mm in size and containing up to 30 wt% Pu irregularly dispersed in a matrix of natural UO2.17,18 This fuel exhibits diVerent irradiation characteristics to conventional UO2 fuel since the fissile material and hence the burn-up is concentrated in the agglomerates.At Karlsruhe, EPMA is being used to obtain information about the state of the MOX agglomerates following irradiation, with particular emphasis on what happens to the Pu. Fig. 11 Electron absorption micrograph and Pu X-ray map showing It is also being used to identify the mechanisms by which the distribution of Pu agglomerates on the surface region of a MOX fission gas is released from the MOX agglomerates and reaches fuel.The agglomerates are characterised by clusters of small pores the fuel rod free volume. Fig. 11 shows the appearance and that form as a consequence of the high local burn-up in the agglomerates (>200 GWd/tPu). distribution of the agglomerates in the surface region of a J. Anal. At. Spectrom., 1999, 14, 447–454 453the local burn-up. It can be seen that the threshold burn-up for recrystallisation lies in the range 60–75 GWd/tU and that the transformation of the fuel microstructure is complete between 100 and 120 GWd/tU.When the microstructure is fully transformed, the concentration of Xe in the UO2 matrix does not exceed 0.3 mass-%. 7 Concluding remarks EPMA has provided the nuclear industry with important fundamental data on the in-pile behaviour of nuclear fuel. For example, EPMA supplied the first data on the local retention of Xe required to model fission gas release during a reactor power excursion, and has since provided a crucial insight into the mechanisms involved.It has also recently delivered some of the first results on the transmutation of Np and Am in a fast reactor, which will enable a decision to be made on whether transmutation is a viable option for the long-term Fig. 12 Scanning electron micrograph showing recrystallised UO2 storage of nuclear waste. Presently, one of the important tasks grains at the surface of a conventional light water reactor fuel with of EPMA at the Institute for Transuranium Elements is to an average burn-up of 45 GWd/tM (M=U+Pu).provide basic data on local burn-up, Pu production and fission gas release for the validation of the fuel performance code TRANSURANUS which is being used by fuel vendors, nuclear licensing authorities and electrical power utilities throughout Europe. References 1 R. Manzel and M. Coquerelle, in Proc. Int. Topical Mtg.on Light Water Reactor Fuel Performance, ANS, LaGrange Park, IL, USA, 1997, p. 463. 2 C. T. Walker, C. Bagger and M. Morgensen, J. Nucl. Mater., 1996, 240, 32. 3 C. T. Walker, W. Goll and T. Matsumura, J. Nucl. Mater., 1996, 228, 8. 4 M. Kurata, T. Inoue and C. Sari, J. Nucl. Mater., 1994, 208, 144. 5 C. T. Walker and G. Nicolaou, J. Nucl. Mater., 1995, 218, 129. 6 C. T. Walker, in Proc. EMAS ‘98: Electron Probe Microanalysis Today—Practical Aspects, ed. X. Llovet, C.Merlet and F. Salvat, University of Barcelona Press, Spain, 1998, p. 313. Fig. 13 Local concentration of Xe in a UO2 nuclear fuel as a function 7 C. T. Walker, J. Trace Microprobe Tech., 1997, 15, 419. of the local burn-up. The Xe concentration falls sharply at 8 C. Ronchi and C. T. Walker, J. Phys. D: Appl. Phys., 1980, 13, 60–75 GWd/tU as a result of recrystallisation. The broken line 2175. represents the predicted relationship without release. 9 H. Kleykamp, Z. Naturforsch., Teil A, 1981, 36, 1388. 10 H. Kleykamp, Z. Naturforsch., Teil A, 1986, 41, 681. 11 I. Farthing, G. Love, V. D. Scott and C. T. Walker, Mikrochim. MOX fuel. The white areas on the Pu X-ray map mark the Acta, 1992, Suppl. 12, 117. locations of the Pu-rich agglomerates and the white spots in 12 V. D. Scott and G. Love, X-ray Spectrom., 1992, 21, 27. the electron absorption micrograph are pores. It can be seen 13 K. F. J. Heinrich, in Proc. 11th Int. Congress on X-ray Optics from Fig. 11 that the MOX agglomerates are highly porous and Microanalysis, ed. J. D. Brown and A. H. Packwood, University of Western Ontario Press, London, Canada, 1987, after irradiation. The change in microstructure is a direct p. 67. consequence of the very high local burn-up in the agglomerates, 14 I. R. Farthing and C. T. Walker, Heinrich’s Mass Absorption which may reach 200 GWd/tM (M=U+Pu) and more. CoeYcients for the K, L and M Lines, The European Commission, Institute for Transuranium Elements, Karlsruhe, Report 6.4 Determination of the extent of surface recrystallisation in K0290140, 1990. high burn-up UO2 fuel 15 R. W. Fink, R. C. Jopson, H. Mark and C. D. Swift, Rev. Mod. Phys., 1966, 28, 513. When the average burn-up of UO2 fuel reaches about 16 C. T. Walker, P. Knappik and M. Mogensen, J. Nucl. Mater., 45 GWd/tU, the UO2 grains at the fuel rim begin to recrystal- 1988, 160, 10. lise (Fig. 12). The driving force for this is the stored energy 17 C. T. Walker, M. Coquerelle, W. Goll and R. Manzel, Nucl. Eng. Des., 1991, 131, 1. resulting from radiation damage at low temperature 18 C. T. Walker, W. Goll and T. Matsumura, J. Nucl. Mater., 1996, (T<1000 °C).19,20 When recrystallisation occurs, nearly all the 228, 8. fission gas is swept out of the original grains. The radial extent 19 C. T. Walker, Concerning the Microstructure Changes that Occur of recrystallisation and the threshold burn-up for the process at the Surface of UO2 Pellets on Irradiation to High Burn-up, The can therefore be estimated from the distance over which Xe European Commission, Institute for Transuranium Elements, depletion occurs. In Fig. 13, Xe concentrations measured in Karlsruhe, Report K0291141, 1991. 20 K. Nogita and K. Une, J. Nucl. Mater., 1995, 226, 302. the cold outer region of the fuel are plotted as a function of Paper 8/06761I 454 J. Anal. At. Spectrom., 1999, 14, 447–454

ISSN:0267-9477    

DOI:10.1039/a806761i

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

A survey of novel X-ray analysis methods† Invited Lecture Jun Kawai Department of Materials Science and Engineering, Kyoto University, Sakyo-ku, Kyoto 606–8501, Japan Received 28th August 1998, Accepted 2nd November 1998 Five novel X-ray analysis methods recently developed in the author’s research group are surveyed: extended X-ray emission fine structure (EXEFS), depth-selective X-ray absorption fine structure (XAFS), X-ray fluorescence holography (XFH), X-ray Raman and total reflection X-ray photoelectron spectroscopy (TRXPS).Some of these methods are completely new and others are modified existing methods, and all of them are easily applicable to practical chemical state analysis. 1. Introduction X-rays were discovered about a century ago and many X-ray analysis methods have been discovered since then, such as X-ray powder diVraction, X-ray fluorescence analysis, X-ray photoelectron spectroscopy, extended X-ray absorption fine structure and X-ray tomography.While it can be said from the viewpoint of analytical chemistry that the 20th century is the age of X-ray analysis, it is also said that X-ray methods are now obsolete and that there is nothing new in X-ray analysis. However, in the last decade, at least four thirdgeneration synchrotron radiation facilities, which provide bright and sharp X-ray beams, have come into use, namely APS near Chicago, IL, USA, ESRF in Grenoble, France, SPring-8 in Nishi-Harima, Japan, and ALS in Berkley, CA, USA.One of the most important applications of the thirdgeneration synchrotron radiation facilities is X-ray micro- or nano-beam analysis. The spatial resolution is now at best a Fig. 1 Representative SiC Si Ka X-ray fluorescence spectrum. few tens of nanometers. X-ray analysis methods are now in full bloom using these facilities. In our research group, at least five novel methods of X-ray analysis have been developed in trace elemental analysis. Another prominent interference within the last 5 years: (i) extended X-ray emission fine peak is the radiative Auger eVect (RAE) satellites.1 The RAE structure (EXEFS), (ii) depth-selective X-ray absorption fine satellites begin at 1615 eV and stretch to lower energy. This structure (XAFS), (iii) X-ray fluorescence holography RAE is emitted by the 2pA1s electric dipole transition associ- (XFH), (iv) X-ray Raman and (v) total reflection X-ray ated with another 2p electron excitation.Therefore, in the photoelectron spectroscopy (TRXPS).Method (i) is possible final state of the RAE X-ray emission, two 2p holes are left. without using synchrotron radiation whereas while methods This electron movement is similar to the K–LL Auger electron (ii)–(v) make good use of some characteristics of synchrotron transition. Therefore, these satellites have been called the radiation. In this paper the above five methods are outlined. K–LL RAE satellites. However, the physical process of RAE is completely diVerent from the Auger transition.Whereas the 2. EXEFS Auger transition takes place due to the Coulomb scattering of electrons, the RAE takes place due to the sudden change in Fig. 1 shows a typical X-ray fluorescence spectrum of SiC in atomic potential before and after the 2pA1s electric dipole the vicinity of Si Ka characteristic X-ray lines. This spectrum transition. Whereas no selection rule is found in the Auger was measured using a conventional wavelength-dispersive transition, the RAE transition is strictly ruled by the angular X-ray fluorescence spectrometer for elemental analysis.The momentum selection rule. Therefore, we call this RAE fine small peak at 1647 eV is the X-ray Raman scattering peak. structure the extended X-ray emission fine structure (EXEFS). An Si Ka X-ray emitted from an Si atom is scattered by a 2p Do readers think that the EXEFS in Fig. 1 resembles the electron in another Si atom, and loses its energy.Therefore, real extended X-ray absorption fine structure (EXAFS)? The the energy diVerence between the Si Ka1,2 (1737 eV) and the first idea that occurred to the present author was this question.2 Raman peak is close to the 2p electron binding energy Although the extended X-ray absorption fine structure3 is now (100 eV). This Raman peak is one of the interference peaks one of the most powerful tools for the local structural analysis of condensed systems, it is still very diYcult without using a †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.synchrotron radiation facility. However, if the above question J. Anal. At. Spectrom., 1999, 14, 455–459 455Fig. 2 Aluminum metal X-ray fluorescence spectrum. Reproduced, with kind permission, from ref. 4. is answered positively, then EXAFS is possible using a conven- Fig. 4 Comparison between EPMA and the true EXAFS spectra. tional X-ray fluorescence spectrometer.More than 10 000 wave- Reproduced, with kind permission, from ref. 7. length-dispersive X-ray fluorescence (WDXRF) spectrometers for elemental analysis are now used throughout the world. We measured metallic Al Ka EXEFS4 and found that the 3. Depth-selective XAFS oscillating fine structure in the X-ray fluorescence spectrum (Fig. 2) resembles the true EXAFS. The Fourier transform of Fig. 5 illustrates typical methods of measuring X-ray absorption this oscillation yields four peaks, as shown in Fig. 3, which spectra using a synchrotron radiation facility. X-rays from an successfully correspond to the first, second, third, and fourth electron storage ring are monochromated by a double-crystal nearest neighbor atomic distances in aluminum metal. The error monochromator and are then incident on a specimen. When in distance was within 8 pm. We could reproduce the AlKO the X-rays are absorbed strongly by the sample, then the bond distance in aluminum oxides using this method.5 Tanuma fluorescent X-rays are strongly emitted and the photoelectrons and Nishio6 pointed out that similar EXAFS-like analysis was are also strongly emitted.The former method is called the possible using an EPMA (electron probe X-ray microanalyzer). X-ray fluorescence yield (XFY) method and is sensitive to the They obtained EXAFS-like spectra of Al and Al2O3 of 1 mm bulk. The latter method is called the total electron yield (TEY) diameter area using an EPMA.The number of wavelength- method and is sensitive to surfaces shallower than 100 A° . We dispersive EPMA instruments now in use is 3500 throughout can measure the X-ray absorption spectra using a proportional the world, with which one can obtain EXAFS-like spectra of counter and an electric current meter; we can obtain bulk and 1 mm diameter area. Fig. 4 shows the EPMA spectrum of MgO surface information simultaneously.15 in the vicinity of Mg Ka characteristic lines.7 It took 3 h to Information on the chemical state of sulfur in aerosol is obtain this EPMA spectrum.The electron beam size was 40 mm important from the viewpoint of environmental science, in diameter. A true EXAFS spectrum measured at a synchrotron order to study the origins of the acid rain. For this purpose, radiation facility is also shown in Fig. 4, which was measured we have developed a chemical state analysis method for sulfur within a few minutes. It is remarkable that EPMA spectrum in fly ash using X-ray absorption spectra by measuring TEY reproduces well the true EXAFS spectrum.and XFY simultaneously.16,17 The chemical state is determined We have succeeded in measuring EXAFS-like spectra using by the chemical shift of the absorption peak or edge. the present method for Na in NaCl,8 Al in AlN9 and various We found that the surface of a ZnS powder was SO42- Si compounds.10–14 Si and SiO2 X-ray absorption near edge whereas the bulk was S2- using the present method.The structure (XANES) spectra were measured within 25 min using surface was oxidized in the present case. We also measured fly a Rh anode X-ray tube with a power of 40 kV and 70 mA.11,12 ash, which was produced by the burning of Austrarian brown In summary, using the EXEFS method, we can measure coal obtained from various mines. The fly ash samples fraction- EXAFS-like spectra in the laboratory without using synchro- ated according to their density and diameter were measured. tron radiation.The time required for measuring one spectrum The fly ash, the particle size of which ranged from subis a few hours and sometimes less than 1 h. The area analyzed micrometer to 150 mm, was mostly composed of SO42-, can be as small as 1 mm diameter. because it was burned, but deep within a particle S2- existed, Proportional counter Fluorescent X-rays Monochromator SR X-ray A e– e– Fig. 3 Fourier transform of oscillating fine structure in Fig. 2. Fig. 5 Schematic illustration of X-ray absorption measurement using TEY and XFY methods. Reproduced, with kind permission, from ref. 4. 456 J. Anal. At. Spectrom., 1999, 14, 455–459Fig. 7 Top, Young’s two-slit experiment; bottom, X-ray fluorescence holography. Fig. 6 Measured Si X-ray absorption spectra of aerosol collected near Sakurajima Island on June 1997. (a) Second, (b) third and (c) fourth stage of high volume air sampler.Current, TEY method; XRF, XFY method. even though the coal was burned at temperatures as high as 1773 K. Fig. 6 shows Si K X-ray absorption spectra of aerosol collected near Sakura-jima Island, a volcano in Japan. The aerosol was collected with a four-stage high volume air sampler Fig. 8 Schematic illustration of experimental set-up for X-ray and the particle size was classified into large, medium and small, fluorescence holography. ranging from sub-micrometer to a few tens of micrometers.The two spectra for each particle size were measured by the TEY and XFY methods simultaneously. ‘Current’ spectra are surface spectra and ‘XRF’ spectra are bulk spectra. The aerosol was composed of at least two chemical form of Si. The bulk and surface Si had diVerent chemical states. In summary, we can obtain information on the chemical state of fine particles using diVerent quantum yield methods for X-ray absorption spectra. The present method does not require any kind of sample preparation; the sample powder is simply put on an aluminum foil which is connected to an electric current meter.Hence we can measure large numbers of samples in a limited beam time in a synchrotron radiation facility. 4. XFH The top part of Fig. 7 shows the famous Young’s experiment. Light waves transmitted after the two-slit wall interfere with each other and a diVraction pattern is formed on the screen. Supposing the two holes are replaced by two balls and the wall is removed, as is shown at the bottom of Fig. 7, the light wave, or fluorescent X-rays emitted from the left ball (i.e., atom), is scattered by the two balls but most of the X-rays are transmitted through the two balls without any interactions. This is the schematic illustration of X-ray fluorescence in a crystal. The X-ray fluorescence is strong, but the scattered X-rays are weak. These three waves interfere with each other Fig. 9 X-Ray fluorescence hologram of SrTiO3.Reproduced, with and some kind of modulation pattern is observable on the kind permission, from ref. 21. Dots are the raw data for h=45° and screen. This is X-ray fluorescence holography (XFH).18,19 The solid lines are Savitzky–Golay smoothed data. This figure shows the diVraction (Fig. 7, top) ignores the initial wave, but the importance of the Savitzky–Golay smoothing to obtain a symmetrical hologram from noisy data. hologram (Fig. 7, bottom) has information on the initial wave.J. Anal. At. Spectrom., 1999, 14, 455–459 457Raman peak intensity had a strong correlation with the cluster size. When the milling time was longer, the cluster in the BN powder became smaller. The initial cluster size was estimated to be 100 nm but it became 10 nm after 10 h of milling. We found that this peak intensity was useful for determining the particle size of the analyte. When the particle size became smaller, the number of dangling bonds or number of atoms at the edge of clusters became larger.The electron density of the non-bonding orbital, which is the dangling bond itself, was situated at 192 eV. This is the reason why the Raman peak at 192 eV became stronger when the particle size decreased. The intensity of the peak at 192 eV is sensitive to the chemical environment, hence we can observe the spectral change due to the change in chemical environment using the 192 eV resonance peak.25 The Raman peak is strong only for boron, transition metal compounds and rare-earth compounds.Hence the present method is applicable to the characterization of these kinds of compounds. 6. TRXPS X-ray photoelectron spectroscopy of a flat surface excited by total reflection X-rays (TRXPS) is surface sensitive and the spectra measured by this method have a low background, which was simulated numerically by Kawai et al.26 A schematic illustration of the experimental set-up using a synchrotron radiation facility is shown in Fig. 11. The incident X-ray Fig. 10 Resonant X-ray Raman spectra of h-BN. Reproduced, with kind permission, from ref. 24. energy usually ranges between 1500 and 3000 eV, because when the X-ray energy is so low, the critical angle of X-ray total reflection is as large as a few degrees. Hence the experiment is not diYcult. This means that the hologram has the phase information which is lost in the diVraction experiment. TRXPS spectra of flat surfaces27–29 and multilayer samples30–32 have been reported.By the measurement of Tegze and Feigel20 first reported that the Fourier transform of the angular dependent intensity of the measured Sr K X-ray TRXPS, i.e., measurement of the incident angle and energy dependence of XPS spectra, we could determine the thickness fluorescence from SrTiO3 reproduced well the atomic resolution crystal structure. It took Tegze and Feigel20 3 months to obtain of the surface layer, which was composed of the same element but with a diVerent oxidation number.TRXPS is suitable for one hologram of SrTiO3 using a conventional X-ray tube. We measured SrTiO3 using the synchrotron radiation bending determining the chemical information in a surface shallower than 5 nm. We could determine the island structure of evapor- magnet beam line at the Photon Factory, KEK, as illustrated in Fig. 8, and obtained the Sr Ka intensity modulation shown ated films compared with model calculations of the angle dependence of the XPS signal.31 in Fig. 9.21 Only 9 h were required to obtain the data plotted in Fig. 9. The signal-to-noise ratio in our experiment was, Fig. 12 shows the XPS of 5 nm copper phthalocyanine film evaporated on an Si wafer28 measured above and below the however, bad (scattered dots in Fig. 9 for h=45°) and thus numerical smoothing, such as Savitzky–Golay method, was critical angle of X-ray total reflection. The glancing angle diVerence was only 0.9° between Fig. 12(a) and (b). If the essential. The smoothed data are shown by solid lines in Fig. 9. We also obtained a 0.02% Zn hologram in a GaAs wafer, and we could determine the atomic site of Zn in a GaAs crystal from the Fourier transform of the hologram.22 This measurement was performed at the SPring-8 undulator beam line within 13 h. Thus the atomic resolution X-ray fluorescence hologram is now obtainable within a few hours for dilute samples. The goal of XFH is to detect low atomic number elements within a few minutes.If we use the resonance at the absorption edge, we could discriminate the chemical states of the same element by selective excitation. Multiple energy X-ray holography19 can shorten the measuring time, but another breakthrough, such as a high power X-ray source, is needed for further development. 5. X-ray Raman Fig. 10 shows representative X-ray Raman scattering spectra, where mechanically milled hexagonal boron nitride was measured. 23,24 The peaks at 182 and 170 eV are ordinary Ka X-ray fluorescence peaks of boron. The peak at 192 eV is the resonant X-ray Raman scattering peak.Incident X-rays whose energy was 192 eV excited the 1s electron of boron into the lowest Stepping motor Sample holder Fluorescent screen Monochromatized incident X-ray Concentric hemispherical electron analyzer nA Direct X-ray Reflected X-ray Sample E 75° e unoccupied orbital. The excited electron fell into the 1s orbital Fig. 11 Schematic illustration of experimental set-up for total reflection XPS. Reproduced, with kind permission, from ref. 33. again, and then a 192 eV X-ray photon was emitted. The 458 J. Anal. At. Spectrom., 1999, 14, 455–459aerosol project. Part of this research was performed with a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture, Japan, and Iketani Science and Technology Foundation. References 1 T.A° berg, Phys. Rev. A, 1971, 4, 1735. 2 J. Kawai, T.Nakajima, T. Inoue, H. Adachi, M. Yamaguchi, K. Maeda and S. Yabuki, Analyst, 1994, 119, 601. 3 D. E. Sayers, E. A. Stern and F. W. Lytle, Phys. Rev. Lett., 1971, 27, 1204. 4 K. Hayashi, J. Kawai and Y. Awakura, Spectrochim. Acta, Part B, 1997, 52, 2169. 5 J. Kawai, K. Hayashi and Y. Awakura, J. Phys. Soc. Jpn. 1997, 66, 3337. 6 S. Tanuma and M. Nishio, Spectrochim. Acta, Part B, 1998, 53, 505. 7 J. Kawai, K. Hayashi, H. Takahashi and Y. Kitajima, J. Synchrotron Radiat., 1996, 6, in the press. 8 J. Kawai, K. Hayashi and S. Tanuma, Analyst, 1998, 123, 617. 9 J. Kawai and K. Hayashi, J. Electron Spectrosc. Relat. Phenom., 1998, 92, 243. 10 J. Kawai, K. Hayashi, K. Okuda and A. Nisawa, Chem. Lett., 1998, 245. 11 J. Kawai, K. Hayashi, K. Okuda and A. Nisawa, Rigaku-Denki J., 1998, 29(1), 23. 12 J. Kawai, K. Hayashi, K. Okuda and A. Nisawa, Rigaku J., 1998, 15(2), 33. 13 J. Kawai and H. Takahashi, Spectrochim. Acta, Part B, 1999, 54(1), in the press. 14 J.Kawai, H. Takahashi and R. Shimizu, J. Trace Microprobe Fig. 12 Measured XPS spectra of 50 A° copper phthalocyanine film on Technol., 1999, 17(1), in the press. Si wafer. Reproduced with kind permission from ref. 28. The incident 15 J. Kawai, H. Adachi, S. Hayakawa, S. Y. Zhen, K. Kobayashi, X-ray glancing angle is (a) 0.45° larger and (b) 0.45° smaller than the Y. Gohshi, K. Maeda and Y. Kitajima, Spectrochim. Acta, Part critical angle. B, 1994, 49, 739. 16 J. Kawai, S. Hayakawa, S.Zheng, Y. Kitajima, H. Adachi, Y. Gohshi, F. Esaka and K. Furuya, Physica B (Amsterdam), incident X-rays were not totally reflected (a), then the penetra- 1995, 208–209, 237. tion depth of X-rays was >102 nm. Hence photoelectrons 17 J. Kawai, S. Hayakawa, F. Esaka, S. Zheng, Y. Kitajima, emitted from deeper locations lost their kinetic energy and a K. Maeda, H. Adachi, Y. Gohshi and K. Furuya, Anal. Chem., 1995, 24, 1526. high background was observable. Additionally, the strong 18 C.S. Fadley and P. M. Len, Nature (London), 1996, 380, 27. substrate Si Auger signal is overlapped by the surface copper 19 P. M. Len, C. S. Fadley and G. Materik, in X-Ray and Inner-Shell peak. On the other hand, if the incident X-rays were totally Processes, 17th International Conference, Hamburg, 1996, ed. R. L. reflected, the penetration depth of the incident X-rays was Johnson, H. Schmidt-Bo� cking and B. Sonntag, AIP Conference <0.5 nm. Hence the XPS signal from the top surface was Proceedings No. 389, American Institute of Physics, New York, 1997, p. 295. selectively measured. 20 M. Tegze and G. Feigel, Nature (London), 1996, 380, 49. We could also determine the inelastic electron mean free 21 J. Kawai, K. Hayashi, T. Yamamoto, S. Hayakawa and path, which is an important parameter in quantitative analysis Y. Gohshi, Anal. Sci., 1998, 14, 903. using XPS, from the background reduction factor when using 22 K. Hayashi, T. Yamamoto, J.Kawai, M. Suzuki, S. Goto, total reflection X-rays.33 The inelastic mean free path deter- S. Hayakawa, K. Sakurai and Y. Gohshi, Anal. Sci., 1998, 14, 987. 23 J. Kawai, S. Tadokoro, Y. Muramatsu, S. Kashiwai, H. Kohzuki, mined by this method was half that used in the database. M. Motoyama, H. Kato and H. Adachi, Physica B (Amsterdam), In summary, we can perform the surface chemical state 1995, 208–209, 251. analysis, thin film characterization (thickness and chemical 24 Y.Muramatsu, M. Oshima, J. Kawai, S. Tadokoro, H. Adachi, composition), island structure analysis and inelastic electron A. Agui, S. Shin, H. Kato, H. Kohzuki and M. Motoyama, Phys. mean free path determination using TRXPS. Rev. Lett., 1996, 76, 3846. 25 J. Kawai, Y. Muramatsu, A. Agui, S. Shin and H. Kato, Spectrochim. Acta, Part B, 1997, 52, 593. 7. Conclusion 26 J. Kawai, M. Takami, M. Fujinami, Y. Hashiguchi, S. Hayakawa and Y. Gohshi, Spectrochim. Acta, Part B, 1992, 47, 983.Although X-ray analysis methods are old and seemed to be 27 J. Kawai, S. Hayakawa, Y. Kitajima, K. Maeda and Y. Gohshi, obsolete, they are now showing a resurgence using new J. Electron Spectrosc. Relat. Phenom., 1995, 76, 313. 28 J. Kawai, S. Kawato, K. Hayashi, T. Horiuchi, K. Matsushige synchrotron radiation facilities. Within our small research and Y. Kitajima, Appl. Phys. Lett., 1995, 67, 3889. group, we have developed at least five novel methods of X-ray 29 J. Kawai, S. Hayakawa, Y. Kitajima and Y. Gohshi, Anal. Sci., analysis in the last 5 years. In addition to the five methods 1995, 11, 519. outlined above, we discovered a new X-ray generation 30 K. Hayashi, S. Kawato, T. Horiuchi, K. Matsushige, Y. Kitajima method.34,35 X-Ray analysis is now one of the most progressive and J. Kawai, Appl. Phys. Lett., 1996, 68, 1921. 31 J. Kawai, H. Amano, K. Hayashi, T. Horiuchi, K. Matsushige areas in analytical chemistry. We must therefore keep in touch and Y. Kitajima, Spectrochim. Acta, Part B, 1997, 52, 873. with new X-ray technologies and breakthroughs which are 32 J. Kawai, K. Hayashi, H. Amano, H. Takenaka and Y. Kitajima, utilized in the field of X-ray analysis, such as an X-ray laser, J. Electron Spectrosc. Relat. Phenom., 1998, 88–91, 787. X-ray lens and X-ray detector methods. 33 J. Kawai, H. Adachi, Y. Gohshi, Y. Kitajima, K. Maeda, S. Hayakawa and Y. Gohshi, Anal. Sci., 1997, 13, 797. 34 J. Kawai, K. Maeda, N. Sakauchi and I. Konishi, Spectrochim. Acknowledgements Acta, Part B, 1995, 50, L1. 35 J. Kawai, K. Maeda, N. Sakauchi and I. Konishi, Nucl. Instrum. The author thanks all his co-authors of the papers cited in the Methods Phys. Res., Sect. B, 1996, 109/110, 206. reference list for their suggestions, hard work and good ideas. He also thanks Professor S. Tohno for cooperating in the Paper 8/06767H J. Anal. At. Spectrom., 1999, 14, 455–459 4

ISSN:0267-9477    

DOI:10.1039/a806767h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

The use of Mn La line chemical eVects in X-ray analysis to probe sample homogeneity† Krystyna Lawniczak-Jablonska,* Jakub Kachniarz, Zoya Spolnik,‡ Joanna Libera, Elzbieta Dynowska, Andrzej Nadolny and Janusz Sadowski Institute of Physics, Polish Academy of Sciences, Al. Lotnikow 32/46, 02 668 Warsaw, Poland Received 23rd September 1998, Accepted 2nd December 1998 The content of Mn in the semimagnetic semiconductors Cd1-xMnxTe and Sn1-xMnxTe bulk crystals and epitaxially grown layers was examined by X-ray spectrometry.The content of Mn in the investigated crystals and the homogeneity of samples were determined using electron probe microanalysis (EPMA). Bulk crystals and epitaxial layers were found to be homogeneous. In spite of this, the shapes of the Mn La valence band line from a bulk crystal and a layer with the same high content of Mn were found to be very diVerent. In this case, the content of Mn estimated from EPMA also diVered significantly from that estimated from magnetic measurements.The detailed analysis of Mn La lines and X-ray diVraction revealed the existence of a second MnTe phase in the epitaxial layer. The best match to the experimental Mn La line was given by spectral fitting with about 50% of Mn atoms bonded in a binary compound and 50% in a ternary compound. Therefore, in the case of epitaxially grown layers, where the diVerent phases can be homogeneously distributed in the sample, the analysis of the valence lineshape can help to identify the existence of diVerent compounds.layered semiconducting ferromagnetic–diamagnetic epitaxial Introduction structures. Based on the known magnetic and electronic In X-ray analysis, the intensity of the characteristic core lines properties of bulk crystals of SnMnTe and SnTe, one can of elements is the basis for the quantitative analysis of element expect to realize a model of a low-dimensional magnetic system contents in multi-component materials.The homogeneous with controlled magnetic and electronic properties. Therefore, distribution of the characteristic X-ray line intensities of of particular importance are the control of the Mn content elements is ascribed to the homogeneity of the compound. and the homogeneity of the produced layers. This is generally true in the case of bulk materials, but the situation can diVer for materials grown using epitaxial technol- Experimental ogy. The crystal structure of the support and the conditions of the technological process can impose the homogeneous Bulk crystals of Cd1-xMnxTe and of Sn1-xMnxTe were grown growth of layers of diVerent compounds and result in the by the modified Bridgman technique.The 0.2–2 mm thick existence of several phases with diVerent composition homo- Sn1-xMnxTe layers were produced by molecular beam epitaxy geneously distributed along the layer. In the typical active (MBE) on cleaved (111) BaFe2 substrates with a 0.01–1 mm region of analysis (w # 1 mm) and with a depth of about 2 mm thick SnTe buVer layer.The substrate temperature was 350 °C. at an excitation energy of 20 keV electrons, the total signal is The growth rate was about 0.4–0.8 mm h-1. The process of given by several epitaxial layers and in some cases also by the crystal growth was monitored in situ by reflection high energy substrate. In such a case, wavelength-dispersive analysis of the electron diVraction (RHEED). To modify the growth convalence lineshape can help to identify the existence of diVerent ditions and to change the number of carriers in the produced compounds in the analyzed volume.The chemical shift and layer, a third Te eVusion cell was used in addition to the SnTe the shape of the valence line provide a fingerprint of the and Mn eVusion cells. compound and can also be used for compound identification.1,2 The chemical composition and the crystal structure of such In this investigation, we analyzed the chemical composition, grown layers were examined by X-ray fluorescence (XRF) homogeneity and chemical bonds of bulk and epitaxially grown analysis, electron probe microanalysis (EPMA) and X-ray layers of Mn doped SnTe and CdTe crystals.Bulk crystals of diVraction. The XRF analysis averaged the concentration of these diluted magnetic semiconductors are known to exhibit a Mn from an area of about 10 mm2. EPMA was performed variety of magnetic properties.Depending on the Mn content using Jeol (Tokyo, Japan) JSM-50A equipment and the energy and on the carrier concentration, they can have ferromagnetic, dispersive (EDS) mode of a microanalyzer with an Si5Li spin glass or paramagnetic properties. The ferromagnetic Curie detector to measure the L lines of Cd, Sn and Te and the K temperature can be varied by changing the concentration of line of Mn and a standard ZAF4 correction procedure for the carriers governed by the number of native defects, i.e., metal evaluation of element content.The energy of the electron vacancies.3–5 The detailed knowledge and control of the con- beam was 20 keV and the current was 5×10-9 A. The La ditions of the growth process allow the production of new valence emission line of Mn was measured using the same electron microprobe combined with a wavelength-dispersive (WDS) modified Johann spectrometer, with RAP (rubidium †Presented at the Fifteenth International Congress on X-ray Optics acid phthalate) and TAP (thallium acid phthalate) analyzer and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998.crystals and a flow proportional counter with an energy ‡Present address: Department of Chemistry, University of Antwerp (UIA), Universiteitsplein 1, 2610 Wilrijk, Antwerp, Belgium. resolution of the order of 2×10-4. The Mn La lines from J. Anal. At. Spectrom., 1999, 14, 461–464 461Fig. 2 Mn La emission spectra from elemental Mn (solid circles) and Cd1-xMnxTe with x=0.52 (solid, thick line), x=0.60 (open squares) and x=0.67 (solid thin line); the maximum error bar of the measured points is indicated.and magnetic susceptibility provided an independent estimate of the Mn content in materials. There was satisfactory agreement between the Mn concentration determined from magnetic susceptibility and from EPR data (for details, see ref. 6). For the layers with x<0.06, EPMA also yielded similar results.In layers with x>0.06 the Mn content determined from EPMA was roughly double the Mn2+ ion content detected in magnetic measurements. Fig. 1 Distribution of Sn (triangles), Te (circles) and Mn (squares) Knowing that the shape of the valence band is very characcontents along the sample’s diagonal as determined by EPMA: (a) for teristic for the type of chemical compounds, we decided to a layer with a low content of Mn and (b) for a layer with a high measure the shape of the Mn La valence line in layered and content of Mn (note the diVerent scales).some other reference samples. In Fig. 2, the La emission lines of Mn in elemental Mn and CdMnTe crystals were measured at an excitation energy of well characterized bulk crystals of Cd1-xMnxTe with diVerent 24 keV using the RAP crystal (Roland circle radius 140 mm). contents of Mn (x=0.52, 0.60, 0.67) are presented and the The measurements of Mn La,b lines for SnMnTe crystals were maximum statistical error is indicated.There is a clear diVerperformed in the WDS mode using 7 keV electron excitation ence in the energy position of the lines and in the shape of the and the TAP crystal as an analyzer. The time for recording lines caused by the diVerent chemical bonding of the Mn spectra was adjusted to achieve reasonable count statistics. In atoms in Mn metal and in Cd1-xMnxTe. Comparing the the case of low signals the spectra were recorded several times shapes of the lines for ternary crystals with diVerent contents and summarized.The maximum statistical error is indicated in Fig. 2. X-ray diVraction patterns were collected by means of a conventional powder diVractometer using filtered Fe Ka radiation. Results and discussion The standard XRF and EDS microanalysis which are routinely used to monitor the composition and homogeneity of layers produced in MBE growth processes did not indicate any inclusions or precipitates of foreign phases and compounds.The composition profiles along the sample’s diagonal (Fig. 1) indicated overall good chemical homogeneity of the layers produced. The amount of Mn detected in the investigated layers determined by XRF analysis and EPMA was the same within the limits of calibration errors. The maximum content of Mn detected by EPMA was close to 10 at.-%. The distribution of Mn through the samples was found to be homogeneous both for low contents of Mn [Fig. 1(a)], and for higher contents [Fig. 1(b)] (note the change in the scale for Mn]. Studies of transport and magnetic properties were then performed on the characterized samples of SnMnTe.6 The magnetic properties were investigated by magnetic susceptibility and electron paramagnetic resonance (EPR) measurements. Fig. 3 Spin–orbit doublet of Mn La,b emission spectra from elemental Mn (squares), MnTe (circles) and bulk Sn0.88Mn0.12Te (triangles). The detailed analysis of the intensity of EPR resonance lines 462 J. Anal.At. Spectrom., 1999, 14, 461–464Fig. 5 Sum of the MnTe spectrum and the bulk Sn0.88Mn0.12Te Fig. 4 Mn La emission spectra from (a) Sn1-xMnxTe layers with x= spectrum (dotted line) and the spectrum of the x=0.12 layer (solid 0.12 (solid, thick line), x=0.06 (solid squares) and bulk crystal with line): (a) with weights of 0.4 MnTe and 0.6 Sn0.88Mn0.12Te and (b) x=0.12 (dotted line) and (b) Sn1-xMnxTe layer with x=0.12 (solid, with weights of 0.5 MnTe and 0.5 Sn0.88Mn0.12Te.thick line), bulk crystal with x=0.12 (dotted line) and MnTe (solid circles). is closer to that of the layer with x=0.06 than to 0.12. This gave direct evidence that the x=0.12 layer is not a single of Mn, the existence of three structures in the spectra, marked A, B and C, can be seen. This suggests that the 3d valence phase layer. In Fig. 4(b), the line from MnTe together with the lines from bulk and layer crystals with x=0.12 are electrons gather in the three sub-bands.With an increase in the Mn content, the sub-bands C and B shift in the direction presented. One can see that the low energy side of the layer spectrum matches fairly well the MnTe valence band spectrum. of higher energy (closer to the Fermi level ) and for x=0.6 an additional structure D appears. Therefore, the center of gravity This suggests the existence of an MnTe compound in the investigated layer. In order to estimate how many Mn atoms of the valence band also shifts to higher energy.The bottom of the valence band is located close to 634 eV and, together are bonded in the binary MnTe compound and in the ternary SnMnTe compound, we added these two spectra with a with sub-band A, does not change much. The electrons located at the bottom do not participate in the bonding to the same diVerent weight and compared the result with the spectrum from the layer. In Fig. 5 the two best matches are shown. extent as those located close to the Fermi level.Therefore, the shape of the Mn La emission line is very characteristic for the From this matching it can be estimated that 40–50% of Mn atoms are bonded in the MnTe compound. The rest of the type of Mn bonding and for the content of Mn in the compound. Mn atoms are bonded in the SnMnTe compound. The atoms of Mn seem not to be metallically bonded in the investigated In Fig. 3, the spin–orbit La,b doublets of Mn from the elemental Mn, MnTe and bulk Sn0.88Mn0.12Te crystal are layer, because the maximum of the La line in the metal is shifted more than 2 eV in the direction of low energy and collected.Substantial changes are seen not only in the La line but also in the Lb line and in the intensity ratio of these two should produce an additional structure in the low energy tail of the spectrum. lines. The additional peak seen at an energy of 627 eV in the SnMnTe spectrum is the sixth order of the Sn Lb3 core line, Analysis of the X-ray diVraction patterns confirmed the above findings.The diVraction pattern for a layer with x= which appears owing to the lack of a high energy line filter in the analyzer. In the further discussion we will concentrate only 0.06 is shown in Fig. 6(a). The dominant features are due to rock salt (111)-oriented SnMnTe crystal with the minor contri- on the Mn La line between 630 and 645 eV. In Fig. 4(a), the Mn La spectra from the bulk crystal with bution due to rock salt (100)-oriented crystal phase.In Fig. 6(b) a similar diVraction pattern is shown for a layer x=0.12 and two layers with x=0.06 and 0.12 are presented. One can see the clear diVerence between the shapes of the with x=0.12. One can clearly see the presence of a second crystal phase, which was identified as hexagonal MnTe. The valence line for bulk and layer material with the same content of Mn (x=0.12). The shape of the line from the bulk crystal Mn atoms bonded in the MnTe phase (NiAs type B13, Mn J.Anal. At. Spectrom., 1999, 14, 461–464 463Conclusions The analysis of the shape and energy position of the Mn La valence lines in ternary alloys of Cd1-xMnxTe and Sn1-xMnxTe bulk crystals and epitaxially grown layers was performed using the EDS and WDS modes of a Jeol JSM-50A microanalyzer. The content of Mn and the nature of the samples estimated using EDS were found to be homogeneous in the bulk crystals and epitaxial layers. However, the shapes of the valence lines measured by WDS in the bulk crystal of Sn0.88Mn0.12Te and layer with the same content of Mn diVered significantly.The detailed analysis of the lineshape revealed the existence of a second MnTe phase in the epitaxial layer. The best match to the experimental line gave a fitting with about 50% of Mn atoms bonded in a binary MnTe compound. The existence of the second phase was additionally confirmed by X-ray diVraction measurements. These investigations proved that in the case of epitaxially grown layers, where the diVerent phases can be homogeneously distributed in the sample, the analysis of the valence lineshape can help to identify the existence of diVerent compounds. This work was supported in part by Grant No. 2P03B 101 14 of the State Committee for Scientific Research (Poland). References 1 Z. Spolnik and K. Lawniczak-Jablonska, Adv. X-Ray Anal., 1997, 39, 831. 2 K. Lawniczak-Jablonska, J. Inoue, T. Tohyama and M. T. Czyzyk, Phys. Rev. B, 1994, 49, 14165. 3 W. J. M. de Jonge, T. Story, H. J. M. Swagten and P. J. T. Eggenkamp, Europhys. Lett., 1992, 17, 631. 4 T. Story, Acta Phys. Pol., A, 1997, 91, 173. 5 P. £azarczyk, T. Story, M. Arciszewska and R. R. Ga�a�zka, J. Magn. Magn. Mater., 1997, 169, 151. 6 A. J. Nadolny, J. Sadowski, T. Story, W. Dobrowolski, M. Fig. 6 X-ray diVraction patterns from (a) layer with x=0.06 and (b) Arciszewska, K. S� wia�tek, J. Kachniarz and J. Adamczewska, Acta layer with x=0.12. Phys. Pol. A, 1998, 94, 449. octahedrally coordinated by Te atoms) are not magnetically Paper 8/07431C active and this explains the observed diVerences between the content of Mn atoms resulting from magnetic and X-ray analysis. 464 J. Anal. At. Spectrom., 1999, 14, 461&n

ISSN:0267-9477    

DOI:10.1039/a807431c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

A new EPMA technique for determination of rare earth elements with the use of automated peak-overlap and modelled background corrections† Irene P. Laputina,a Vadim A. Batyrevb and Anton I. Yakusheva aRussian Academy of Sciences, Institute of Geology of Ore Deposits (IGEM) Staromonetny per. 35, 109017 Moscow, Russia bFederal Scientific Center ‘TCNIITMASH’, Moscow, Russia Received 2nd September 1998, Accepted 9th September 1998 Analysis of rare earth elements (REEs) plays an important part in geology and geochemistry for solving fundamental problems of ore formation, e.g., formation of gold and platinum deposits, as well as in ecology, in searching for materials suitable for burial of highly active wastes, where REEs are used in investigations as actinide imitators.However, electron microprobe analysis (EPMA) of REEs faces a series of problems related to the complex character of their X-ray spectra. For solving these problems, we suggest the use of an improved version of the program developed by us for the analysis of platinum group elements (PGEs), in which a computational method of background determination and an automatic correction for interfering lines are employed before applying the standard ZAF correction procedure.The program enables one to solve the problem of background determination, to simplify significantly REE analyses, to reduce random errors, to improve the correctness and speed of microprobe determinations, and to lower the limit of detection (LOD) of the elements analysed.When analysing thermally stable REE-bearing minerals, the LOD for most REEs lies in the concentration range 200–600 ppm. These values are obtained at an accelerating voltage of 20 kV, with a probe current of 100 nA and a counting time of 50 s at the peaks of analytical lines. For elements with multiple interference of analytical lines, such as Gd, Tm, Eu and Lu, the LOD estimated from the standard formula determines virtually the concentration level that might be achieved only in the case of correct measurements of overlap coeYcients of all interfering lines.The use of a computational method for background determinations allows the required analysis time to be reduced by nearly a factor of two. We present in this paper the values of overlap coeYcients and assess the reproducibility of results of the analysis of monazite, xenotime, apatite, and zirconolite. The basic principles of our software can be used successfully with various types of microprobe analysers including modern instruments.quence, EPMA remains to date one of the major methods for Introduction the study of mineral composition in zirconolite-rich ceramics Analysis of REEs is of great importance in geology and and the element distribution among minerals. Whilst EPMA geochemistry for solving fundamental problems of ore forma- is a well elaborated method for microprobe analysis of most tion, as well as in ecology.The study of the composition and minerals, the quantitative determination of REEs in minerals distribution of REEs, F and Cl in accessory minerals (apatite, is by no means a routine procedure because it poses specific monazite, xenotime, allanite, titanite, etc.) is commonly used analytical problems to researchers. Two main problems arise for the determination of the sources of ore substances, tempera- in EPMA of REE minerals due to a great number of closely ture and fluid regimes, e.g., in the formation of gold and spaced X-ray lines related to the excitation of the L series of platinum deposits.1,2 REEs: (1) it is quite diYcult to choose the correct spectrometer In the last two decades REEs have been actively used in positions for background measurements among a multitude studies as actinide imitators in connection with the problem of closely spaced peaks in the REE energy spectrum; (2) a of creating modern materials for the immobilization of high commonly encountered interference of analytical lines with level waste (HLW) components.The isomorphic properties of various X-ray lines of REEs and other elements present in the minerals in zirconolite-rich ceramic materials have been studied given mineral ( Ti, Mn, Fe, Ba, Ta) hinders the measurement most carefully because zirconolite has a good isomorphic of the true intensities of some elements. capability for keeping many of the HLW components in its A great number of works have been devoted to the composition, especially those from actinide fractions.3 These microprobe determination of REEs with the use of wavelengthmaterials also possess the highest chemical stability and dura- dispersive spectrometers (WDSs).8–10 Some researchers bility in comparison to other materials oVered for these employed in analysis the empirical correlation coeYcients (a purposes.4,5 Such materials are produced by the method of factors),11 while others used the method of correction with the subsolidus hot pressing (Australia)6 and by the induction aid of calculated overlap coeYcients for the main X-ray smelting technique (Russia).7 The size of crystals varies from lines.12,13 Detailed investigation of the REEs’ overlap several micrometres to some tens of micrometres; as a conse- coeYcients, careful analysis of the background measurement sites and the assessment of the influence of the overlap coeYcient value have recently been made by Reed and †Presented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24–27, 1998. Buckley.14 However, in all these methods the problem of J.Anal. At. Spectrom., 1999, 14, 465–469 465background measurements remained in essence unsolved. The used the most intense La lines for all elements and a LiF 200 crystal analyser possessing much higher resolution than a PET main goal of our investigation is to solve the problem of background determination and to provide an automatic correc- 002 crystal.The computer program we developed provides the possibility of simultaneous analysis of up to 25 elements with tion of overlapping lines, i.e. to simplify REE analysis, to reduce the number of random errors, and to improve the high or low concentrations. As standards, we used the synthesized compounds, separately for each REE (Table 1). correctness and speed of REE determinations.For these purposes we improved a computer program developed by us for the analysis of platinoid minerals in which the compu- Background modelling tational method of background determination and an auto- Recently we showed the universal character of formulae matic correction for interfering lines of REEs and matrix (1)15 and (2)16 for a broad range of wavelengths from 0.10573 elements are used prior to the use of the standard ZAF to 0.96709 nm (As Kb–As La lines) and average atomic correction procedure.A similar approach has been previously numbers Z of minerals (Z = 14–83). The approach is based employed for the analysis of trace amounts of elements in on the equation silicate minerals15 and for the analysis of platinoid minerals and traces of precious metals in sulfides and sulfarsenides.16 Bl=Tl (Zsp m(Z,E) fl/Zst m(Z,E) kl) (1) where Bl is the modelled background response of the lth Experimental wavelength for the specimen; Tl is the background response of the lth wavelength for pure standard; kl and fl are correction All studies were carried out with a CAMEBAXMICROBEAM (Moscow, Russia) electron probe analyser factors (accounting for absorption and backscattering) for the standard and the specimen, respectively; Zst and Zsp are the equipped with three WD X-ray spectrometers (one tilted and two vertical ) and a PDP 11/23 computer.The operating mean atomic numbers of the standard and the specimen, respectively; Z=S(CiZi), where Ci is the concentration of the voltage was fixed at 20 kV.To perform the REE analysis we Table 1 Interference observed in microprobe analysis of REEs (main peaks) Elements Overlap (%) Analyseda Overlapping Overlap standards This paper Ref. 17 Ref. 12 Ref. 11 Ref. 14 Ce Ba BaSO4 6.8 0.7b Pr La LaF3, LaP5O14 12.0 8.9 12.4 12.72 Sm Sm2O3xAl2O3 0.4 Nd Ce CeF3, CeP5O14 1.2 0.4 0.9 Cs CsAlSi2O6 0.4 0.43 Sm Ce CeF3, CeP5O14 1.9 1.65 0.2 1.9 0.32 Nd NdF3 0.3 Eu Pr PrF3 15.5 14.9 21.7 20.33 14.0 Nd NdF3 2.3 1.27 1.2 0.96 La LaF3, LaP5O14 0.1 0.13 Mn MnCO3 0.5 0.51 0.63 Gd Ce CeF3, CeP5O14 7.6 6.9 9.5 11.79 3.9 Nd NdF3 1.3 0.19 La LaF3, LaP5O14 1.6 0.15 1.2 2.8 0.56 Tb Sm Sm2O3xAl2O3 0.9 0.75 0.91 1.2 Pr PrF3 0.2 1.1 Ce CeF3, CeP5O14 0.1 0.08 0.37 Dy Mn MnCO3 33.6 31.4 37.0 Th ThO2, ThSiO4 1.0 1.34 Eu NaEuGeO4 3.3 0.81 0.1 5.9 Fe Fe3O4 0.3 0.3 1.3 Ho Gdc NaGdGeO4 41.4 41.0 42.1 Nd NdF3 0.1 Lu LuPO4 0.2 Er Tb TbPO4 7.5 7.0 9.2 3.0 Fe Fe3O4 0.4 0.35 0.52 Dy Dy2O3x1.75Al2O3 0.1 Tm Sm Sm2O3xAl2O3 10.1 8.69 9.3 8.66 8.1 Ta Ta met. 2.5 2.48 2.7 Dy Dy2O3x1.75Al2O3 0.9 1.09 1.5 0.42 1.1 Tb TbPO4 0.4 0.37 Gd NaGdGeO4 0.5 1.32 0.07 Yb Tb TbPO4 0.5 0.44 0.3 3.66 Dy Dy2O3x1.75Al2O3 0.8 0.42 0.2 0.26 Sm Sm2O3xAl2O3 0.3 0.28 0.89 Ho HoOOH 0.4 2.9 Eu NaEuGeO4 0.1 3.1 Lu Ho HoOOH 11.0 6.5 9.58 8.9 Dy Dy2O3x1.75Al2O3 8.6 4.48 8.98 7.9 Er Er2SiO5 0.3 0.27 Gd NaGdGeO4 0.1 0.11 Cs Ce CeF3, CeP5O14 1.5 Yb Yb2SiO5 1.8 Y Lad LaF3, LaP5O14 1.5 aLa lines.bWe assume that there is a misprint in Reed’s paper. cIn [10], overlap=67%. dMonochromator PET. 466 J. Anal. At. Spectrom., 1999, 14, 465–469ith element in the specimen; m is the coeYcient of deviation measurements, separately for each element, which were located frequently at much greater distances from the peak of the of the Bl(Z) dependence from the linear law described by the Kramer’s expression. It follows from the data obtained that analytical line than those recommended by the standard programs supplied by commercial firms (up to 2000 steps, with the Z-dependence of background intensity agrees with Kramer’s expression for mean atomic numbers Z<50.Some one step equal to sinh×105). Such a procedure is necessary because the eVect of a strong interfering line is still perceptible deviation from this dependence observed for the samples with Z>50 is particularly pronounced for longer wavelengths even at a distance as large as 800 steps from its peak.Then we measured the overlap coeYcient using the program,16 the (l>0.5 nm). Using the least-squares fitting procedure, we approximated the m values for two wavelength regions, correctness of which for the overlap removal was checked by l<0.5 nm and l>0.5 nm, by the following expressions: us previously on a variety of platinum–iridium alloys (overlap with the Pt Ma peak, 10 determinations) and silver–palladium alloys (overlap with Ag La peak, 30 determinations). The m=1.0092+0.000046 Z-0.0000036 Z2 for l<0.5 nm, m=0.9953+0.0008 Z-0.000015 Z2 for l>0.5 nm.counting time in all measurements was 50 s. The correctness of the measured overlap coeYcients for REEs was then verified (2) experimentally with various probe currents on the same syn- To obtain the calculated background value, the background thetic specimens which are used by us as standards. The intensities are preliminarily measured at the wavelengths of all overlap coeYcients for REE peaks obtained empirically are the analytical lines on a single specimen that hereafter is listed in Table 1 (the absolute values of these factors depend referred to as the background standard. As such a standard, on the resolving power of a particular crystal and vary slightly it is possible to use a pure element or a compound having no for diVerent spectrometers).A total of 58 overlap coeYcients emission lines in the region of the analytical lines of the was introduced in the program to analyse REE-bearing minerals elements to be analysed.To check the applicability of the and zirconolite-rich ceramic materials, 37 being destined for above formulae for the calculation of background intensities REEs only and the remaining coeYcients for 15 attendant for REEs, we performed determinations of background inten- elements. As follows from the table, the analytical lines of sities for standard REE specimens available at our disposal various REEs may overlap to a diVerent extent with one or a (Table 1, 19 standards), using as background standards three few lines of one or more elements, depending on the proximity diVerent minerals with diVerent mean Z producing no interfer- on the energy scale of the analysed and interfering peaks.The ing lines in the range of analytical REE lines. These were table also contains some values of overlap factors obtained by rutile TiO2, cassiterite SnO2, and thorium oxide ThO2.In rare other researchers. A comparison of overlap coeYcients cases, when we dealt with overlaps of X-ray radiation emitted obtained by various authors with microanalysers of diVerent by the background standard with the analytical line of one of types shows reasonable agreement at high overlaps (with the the analysed elements, another specimen, without interference exception of Ho La–Gd overlap studied by Fialin et al.10). in the given region, was used as a background standard. For Poorer agreement was found, however, for lower overlap example, when using SnO2 as a background standard, the coeYcients.The reason may be the diVerent quality of focusing background for the Ca Ka line was measured on TiO2; when the crystal monochromators; the presence of impurities of having ThO2 as a background standard, metallic Ir was used interfering elements in standards, which are now taken into to measure the background for Th Ma and Dy La lines, while account in measurements of overlap coeYcients; and the the background for P Ka and Ca Ka was measured on diVerent sites of background measurements.The eVect of the metallic Pd. choice of these sites on the line interference for various To reduce statistical error in the background measurements, combinations of REEs was clearly demonstrated by Reed and we optimized the measurement conditions with respect to the Buckley.14 The measured values of overlap coeYcients are counting time and probe current.The least diVerence between introduced as a file in a subroutine providing the preparation measured and predicted background intensities is achieved of results of measurements and ZAF corrections. The autowhen the specimen with the highest mean Z is used as a matic correction of measured intensities for interference perbackground standard, which is natural since this choice mits one to introduce corrected, true intensities for all analysed improves the precision of measurement of the standard back- elements in the final calculation of concentrations.ground intensity. Comparing the background intensities measured at the peak of the analytical line of the standard without interferences with that calculated for the same wavelength Software shows good agreement. The absolute background intensity for For REE analyses we used an improved version of the program the given wavelength in a wide range of atomic numbers Z is which has been employed earlier by us for analysis of REE- calculated with an accuracy of±2–8% (with ThO2 background bearing minerals.15 The basis of the program for the simul- standard), but for the specimens with mean Z in the range taneous quantitative determination of 25 elements with the 35–50, typical of REE-bearing minerals, the accuracy of the aid of a CAMEBAX-MICROBEAM microanalyser is the calculated background values for all wavelengths is better determination of the background intensity [eqn.(1) with the than ±4%. m values from eqn. (2)], using the peak-overlap correction When using this program with the calculation of background introduced according to the previously published formula. intensity, there is no need to search for spectrometer positions Since the ‘shift from the line peak’ is sometimes observed in for background measurements; this enables us to solve complithis instrument, the program allows for the adjustment of cated REE analysis problems and reduces the analysis time by crystals in any spectrometer in the process of analysis, without almost a factor of two.interrupting the measurement cycle. The adjustment of peaks is accomplished with a suitable standard having no interference Peak-overlap corrections lines in the analysed spectral region. The computer program contains all possible overlap coeYcients for the group of When measuring the overlap coeYcients, it is of crucial elements to be investigated. The concentrations of major and importance to correctly measure the background intensity.For trace elements are measured simultaneously. Then, all overlaps this purpose we recorded the L spectra of all REEs from the of interfering lines are subtracted from the measured intensities specimens, each containing only a single REE. These spectra were used to choose spectrometer positions for background (which requires several iterations) and the background inten- J. Anal.At. Spectrom., 1999, 14, 465–469 467Table 2 Electron microprobe analysis (wt%, oxides) of REE minerals using procedure described in the text (La lines for REEs) Monazite-(Ce)b Xenotime Zirconolite Apatitea Centreb Rim Average Average Standard Average Standard Single Average Standard Analyte of 7 of 15 deviation of 6 deviation measurement Analyte of 7 deviation P2O5 42.3 30.9 0.97 37.2 1.1 33.9 Mg 0.24 0.006 CaO 53.0 0.34 0.03 0.18 0.03 3.0 Si 0.26 0.001 Fe2O3 0.12 0.18 0.01 0.14 0.03 0.39 Ca 9.11 0.222 Y2O3 <0.015 <0.024 39.0 0.59 20.0 Sr 0.32 0.062 La2O3 0.42 24.9 0.36 <0.032 9.35 Ti 14.4 0.171 Ce2O3 0.60 31.9 0.46 <0.041 13.7 Mn 0.37 0.001 Pr2O3 0.09 1.94 0.14 <0.040 0.11 Fe 6.29 0.112 Nd2O3 0.22 4.8 0.24 0.39 0.04 3.92 Zr 22.9 0.339 Sm2O3 0.09 <0.044 0.58 0.05 1.01 Nb 8.15 0.204 Eu2O3 <0.032 <0.043 0.44 0.06 0.13 Ce 0.43 0.037 Gd2O3 <0.034 <0.044 2.04 0.21 2.13 Pr 0.15 0.036 Tb2O3 <0.037 <0.049 0.61 0.07 0.52 Nd 0.61 0.034 Dy2O3 <0.034 <0.045 6.14 0.34 4.03 Ta 0.61 0.098 Ho2O3 <0.042 <0.056 1.35 0.11 0.54 Th 5.83 0.251 Er2O3 <0.033 <0.045 4.68 0.19 2.54 U 1.45 0.14 Tm2O3 <0.044 <0.058 0.70 0.08 0.29 O 28.5 0.015 Yb2O3 <0.035 <0.047 5.1 0.40 1.96 Total 99.9 1.03 Lu2O3 <0.048 <0.064 0.68 0.09 0.18 ThO2 <0.034 4.8 0.18 0.51 0.31 F 2.32 Total 99.00 99.10 98.96 98.01 aWithout Na, Mg, Al, Si.Analysis: monazite-Ce from Cherry Mountain, Ural, Russia; apatite and xenotime from Kola Peninsula, Russia.bCrystal-chemistry formulae of monazite: REE0.983P1.010O4.007; of xenotime (centre): REE0.958P1.030O4.012. sity is calculated for each element. The completely corrected (7.04%), Tm La1–Sm Lc1 (8.7%) and Sm La1–Ce Lb2 (1.09%), or Eu La1–Pr Lb2 (16.7%) and Pr La1–La Lb1 (7.17%), as relative intensities are introduced in the ZAF correction procedure. The results obtained for major components make it well as the elements with multiple line interference, e.g.the interference of the Gd La line with X-ray lines of Ce, La and possible to monitor the instrumental drift and long-term stability of the specimen in the measurement process. The Nd; interference of the Eu La line with lines of Pr, Nd, La and Mn; the interference of the Tm La line with lines of Sm, limit of detection was determined in accord with the 3s criterion taking into account the ZAF correction factors for Dy, Tb and Gd; and the interference of the Lu La line with lines of Ho, Dy, Er and Gd.In these cases, the overlap factor all analytical lines in the specimen. must be determined most carefully because it influences both the analysis correctness and the real detection limit. The LOD Results calculated from the standard formula determines in this situation only the concentration level achievable for correctly We have analysed a few REE-bearing minerals using the program described above: monazite with the general formula measured overlap factors.(Ce,La,Nd,Th)PO4 containing mostly light REEs up to 70% REE2O3; two xenotimes of the general formula YPO4 Table 3 Analysis of monazite and xenotime using lines free from containing chiefly heavy REEs; RE-rich fluorapatite significant overlaps for REE (as comparison) Ca5(PO4)3(F,OH,Cl); and (Ce,Nd)-bearing zirconolite of the formula CaZrTi2O7. Monazite Xenotime centre The average values of the element concentrations in these minerals and root-mean-square (r.m.s.) deviations are presented Analyte Lines Average of 3 Average of 3 in Table 2.The r.m.s. points obtained were low compared to P2O5 Ka 30.24 36.20 the high stability of the probe current and the uniform REE CaO Ka 0.33 0.98 distribution in the analysed minerals. For comparison we Fe2O3 Ka 0.18 0.15 present in Table 3 the results obtained for these minerals with Y2O3 La 0 39.20 the use of X-ray lines free from significant interferences.The La2O3 La 24.70 <0.032 concentrations of REEs obtained for diVerent lines with Ce2O3 La 31.8 <0.041 Pr2O3 Lb 1.92 <0.040 diVerent programs are close to each other. Nd2O3 La 4.92 0.37 The detection limit in EPMA usually depends on the Sm2O3 Lb <0.044 0.59 accelerating voltage, probe current, and time of measuring the Eu2O3 Lb <0.043 0.45 peak and background intensities, as well as on the mineral Gd2O3 Lb <0.044 2.02 stability at high currents. In the case of REE analysis, the Tb2O3 La <0.049 0.59 LOD value is also aVected by the occurrence of spectral Dy2O3 La <0.045 6.04 Ho2O3 Lb <0.056 1.25 interference and by the necessity of introducing overlap correc- Er2O3 La <0.045 4.73 tions.In this situation, the detection limit and standard Tm2O3 La <0.058 0.66 deviations are worse and depend on the ratio of the concen- Yb2O3 La <0.047 4.98 trations of interfering elements and on the correctness of Lu2O3 La <0.064 0.67 overlap factor determination. This concerns to a greater extent ThO2 Ma 4.79 0.52 the elements for which successive line overlaps are observed, Total 98.89 99.39 for example, Ho La1–Gd Lb1 (42.9%) and Gd La1–Ce Lc1 468 J.Anal. At. Spectrom., 1999, 14, 465–469Table 4 Limits of detection of REEs (wt%) calculated for some rare a modified form of the Kramer’s expression; (3) simultaneous earth phosphatesa measurements of major, minor and trace elements (up to 25), which allow the instrumental drift and specimen stability to Element, line REE-Apatite Monazite-(Ce) Xenotime-(Y) be monitored in the course of measurements; (4) optimization of operating parameters (such as the probe current and Y La 0.015 0.024 0.021 La La 0.027 0.034 0.032 counting times).The program provides REE analysis in ther- Ce La 0.033 0.042 0.041 mally stable minerals with a detection limit in the concentration Pr La 0.032 0.042 0.040 range 200–600 ppm at 20 kV, a probe current of 100 nA, and Nd La 0.023 0.030 0.029 a counting time of 50 s.The total analysis time for 25 elements Sm La 0.025 0.044 0.032 is 10–15 min. Eu La 0.032 0.043 0.039 The basic principles of our software can be used successfully Gd La 0.034 0.044 0.042 Tb La 0.037 0.049 0.045 with various types of microprobe analyser, including modern Dy La 0.034 0.045 0.041 instruments. Ho La 0.042 0.056 0.052 Er La 0.033 0.045 0.041 Tm La 0.044 0.058 0.053 Acknowledgements Yb La 0.035 0.047 0.043 Lu La 0.048 0.064 0.059 The authors would like to thank the referees and editors for Th Ma 0.034 0.044 0.044 their helpful comments. The authors are also grateful to E.I.a3s limits of detection are based on 50 s count time on peak, gun Semenov and M.D. Dorfman (Fersman Mineralogical operating potential of 20 kV and 100 nA probe current, and Museum, Russian Academy of Sciences), as well as to A. background modelling; for LiF 200 crystal, except Y La and Th Khomyakov (Institute of Mineralogy and Geology of Rare Ma (PET). Elements), for a series of minerals supplied for this study.They also express special gratitude to A.I. Kozlenkov for carefully reading the manuscript of this paper and improving its English For thermally stable minerals (phosphates, silicates and style. This study was supported in part by the Russian oxides), the detection limit was determined at 20 kV, with a Foundation for Basic Research (project no. 98-05-65017).probe current of 100 nA and a counting time of 50 s at the peak of the analytical line. The time of analysis of one mineral containing 25 elements (with the use of 3 WDSs) was about References 12–15 min. For almost all REEs, the detection limits are close 1 A. E. Boudreau and I. S. McCallum, Contrib. Mineral. Petrol., to each other, being as low as 0.02–0.06 wt.% (200–600 ppm) 1989, 102, 138. (Table 4), which is comparable with the best reported results17 2 I. P. Laputina and T.L. Grokhovskaya, in International obtained with a CAMECA SX-50 microprobe analyser at Symposium Devoted to the Centenary of Late Academic 25 kV. A. G. Betekhtin, Moscow, 8–10 April, 1997, ed. N. P. Laverov, N. Thermally unstable minerals (carbonates, hydrated minerals, S. Bortnikov, A. D.Genkin, V. A. Kovalenker and Yu. G. Safonov, Na- and Cs-containing glasses) were analysed with a probe IGEM RAS, Moscow, 1997, Abstract, pp. 225–226 (in Russian). 3 S. E. Kesson, W. G.Sinclair and A. E. Ringwood, Nucl. Chem. current of 15 nA, a counting time of 10 s, and with an extended Waste Manag., 1983, 4, 259. 25 mm probe. As a result, the LOD was impaired. Some 4 W. G. Sinclair and A. E. Ringwood, Geochem. J., 1981, 15, 229. phosphates, in particular, apatite containing F, Cl and Na, 5 G. R. Lumpkin, K. L. Smith and M. G. Blackford, in Proc. Mater. as well as carbonates with low REE concentrations, were Res. Soc. Symp., Kyoto, Japan, October 23–27 1994, ed.analysed with the use of a complex procedure. First, under T. Mukurami and R. C. Ewing, Mater. Res. Soc., Pittsburgh, PA, standard analysis conditions (20 kV, 20 nA, 10 s) we carried 1995, vol. 353, pp. 855–862. 6 E. R. Vance, B. D. Begg, R. A. Day and C. J. Ball, Ibid, 767. out an overall analysis of minerals for all elements, including 7 N. P. Laverov, B. I. Omel’yanenko, S. V. Yudintsev and F, Cl, Na and major REEs (Ce, La and Y). Then, the current B. S. Nikonov, Geol. Ore Dep., 1997, 39, 179. and analysis times were increased and, along with major 8 D. G.W. Smith and S. J. B. Reed, X-Ray Spectrom., 1981, 10, 198. elements, trace amounts of REEs were determined with the 9 J. J. Donavan, D. A. Snyder and M. L. Rivers, Microbeam Anal., maximum possible sensitivity. This allowed us to ensure that 1993, 2, 23. volatile elements were determined correctly and were not 10 M. Fialin, M. Outrequin and P. F. Staub, Eur. J. Mineral., 1997, 9, 965. missed in the process of analysis. 11 R. Amli and W. L. GriYn, Am. Mineral., 1975, 60, 599. 12 P. L. Roeder, Can. Mineral., 1985, 23, 263. Conclusion 13 G. Remond, Ph. Coutures, C. Gilles and D. Massiot, Scan. Microsc., 1989, 3–4, 1059. The program described in this paper substantially improves 14 S. J. B. Reed and A. Buckley, Mineral. Mag., 1998, 62(1), 1. the quality of analysis and simplifies the process of analysis of 15 C. Merlet and J. L. Bodinier, Chem. Geol., 1990, 83, 55. 16 I. P. Laputina and V. A. Batyrev, Mikrochim. Acta, 1998, REEs in minerals by the EPMA technique (with WDSs), [Suppl ] 15, 247. increases the correctness and quickness of analysis, and 17 C. T. Williams, in Rare Earth Minerals, Chemistry, Origin and Ore improves the limit of detection. This is achieved by (1) the Deposits, ed. A. P. Jones, F.Wall and C. T. Williams, Chapman & automatic peak-overlap correction with the aim to obtain true Hall, 1996, ch. 13, pp. 335–342. k-ratios before applying the standard ZAF matrix corrections; (2) the exact modelling of background intensity by means of Paper 8/06828C J. Anal. At. Spectrom., 1999, 14, 465–469 469

ISSN:0267-9477    

DOI:10.1039/a806828c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

INTER-LABORATORY NOTE Chemical inhomogeneity of silicates treated by plasma spraying¢Ó Blahoslav J. Kolman,* Karel Neufuss, Jan Ilavsky¢¥, Jir¢§©¥¢¥ Dubsky¢¥ and Pavel Chra¢¥ska Institute of Plasma Physics, Academy of Sciences of the Czech Republic, Prague, Czech Republic Received 28th August 1998, Accepted 3rd February 1999 The properties of thermally sprayed deposits are directly influenced by the homogeneity of their chemical composition as well as by their microstructure. The presented work deals with the evaluation and quantification of the chemical homogeneity of silicates, i.e., basalt, cordierite, garnets, mullite, steatite and wollastonite.The deposits were manufactured by a water-stabilized plasma spray system WSPA PAL160. A number of point chemical analyses were performed for each deposit on a polished cross-section by X-ray microanalysis and the results were statistically evaluated. For some of the selected materials, the homogeneity of deposits was compared with the homogeneity of appropriate feedstock powder used for spraying.The microstructure of the cross-sections was observed by back-scattered electron imaging. Electron probe X-ray microanalysis (EDS-SEM) results were combined with X-ray diVraction measurements and showed the presence of various crystalline and amorphous phases in the deposits. The mean concentration, its standard deviation, the minimum and maximum compound content as well as Weibull statistics were used to characterize the chemical homogeneity of the deposits.Generally, the materials showed changes in homogeneity after deposition. This was most notable in steatite and wollastonite. spray techniques. Out of many previously tried materials, Introduction e.g., basalt, cordierite (2MgO¡�2Al2O3¡�5SiO2), garnets Plasma spraying is a rapid heating/rapid cooling and [3(Fe,Mg)O¡�Al2O3¡�3SiO2], mullite (3Al2O3¡�2SiO2), steatite solidification process which often results in the formation of (MgSiO3), wollastonite (CaSiO3) and zircon (ZrSiO4),3,7.10 metastable phases in deposits.1 Furthermore, since the process the first six were selected for this work.is far from being in equilibrium, during spraying the feedstock material can be overheated in some parts and not molten in Experimental techniques other parts. Therefore, changes from the original phase and chemical composition are regularly observed.2 Deposit manufacturing The properties of the deposits are directly related to their Feedstock was prepared by crushing the industrial minerals chemical and phase composition.3 Therefore, our ability to into suitable size range (30.70 mm) and deposition by a water- reliably characterize and quantify them directly influences their stabilized plasma spray system PAL 160.11 This is a high- engineering applicability.Techniques currently used for chemithroughput system suitable for the manufacturing of deposits cal and phase composition characterization include wet chemion large area engineering components or for the formation of cal analysis, electron probe X-ray microanalysis (EDS-SEM), free-standing parts.Production of free-standing large compo- X-ray diVraction and other routinely applied methods.4 nents is especially useful as novel method of rapid prototyping. Results are usually expressed in per cent. by mass as the average chemical composition. However, such characterization is insuYcient, as the chemical homogeneity is also important.Microscopy and microanalysis More complex means of characterization of the homogeneity SEM and EDS-SEM raw data were obtained from cross have been developed and applied in the thermal spray field, sections. Standard metallographic methods for sample prep- combining the power of EDS-SEM5 and statistical methods. aration, i.e., epoxy resin embedding, sectioning/cutting, grind- A challenge to the statistical methods is to meaningfully ing, polishing and carbon evaporation, were applied before describe the average content of any particular element (or analysis.compound) as well as the homogeneity. A scanning electron microscope (SEM) CamScan 4DV This communication presents examples of feedstock and (CamScan, Waterbeach, Cambridgeshire, United Kingdom) deposit characterization and various methods of presentation with an acceleration voltage of 20 kV was utilized for structural of chemical homogeneity of thermally sprayed deposits studies in secondary (SE) and backscattered electron (BE) obtained at the Institute of Plasma Physics.mode. An energy dispersive X-ray microanalyzer AN 10000 (Link Analytical, High Wycombe, United Kingdom) for Materials elements from Na to Fe, with the ZAF-4/FLS correction, was used for point analysis. A cobalt standard was used for A number of silicates are widely used by the ¡®classical¡� ceramic calibration of the pulse processor and X-ray background was industry.6 Most of them can also easily be sprayed by thermal eliminated by digital filtering. From 20 to 30 points were analyzed on each sample with 100 s measurement time.The ¢ÓPresented at the Fifteenth International Congress on X-ray Optics and Microanalysis (ICXOM), Antwerp, Belgium, August 24.27, 1998. analyzed grains of feedstock were randomly selected. The J. Anal. At. Spectrom., 1999, 14, 471.473 471Table 1 Example of standard deviations of basalt sample Oxide Na2O MgO Al2 O3 SiO2 K2O CaO TiO2 Cr2O3 MnO FeO Mean (a) 2.3 12.0 12.9 46.0 0.9 12.0 2.5 0.1 0.2 11.2 Standard deviation (s) 0.2 0.3 00.3 0.4 0.1 0.2 0.2 0.1 0.1 0.3 measurement points in the deposits cross sections were aligned in two-by-five point grids with distances between two adjacent points of 100 mm. Up to three such grids were randomly placed in the deposit cross section.Care was taken to ensure that no measurement point lies in the voids in the cross sections.The probe size (diameter of the area generating FeKa radiation) was about 3 mm. Ten main elements were determined. Typical minerals and pure metals were used as standards for X-ray Ka line profiles: Na, jadeite; Mg, MgO; Al, alumina; Si, quartz; K, leucite; Ca, wollastonite; Ti, TiO; Mn, Cr and Fe, metallic. The elements were converted to oxides after oxygen was added from the stoichiometric calculation, and the results were normalized to 100%. Examples of standard deviations [wt%] for 20 analyses of a homogeneous basalt-deposit sample are given in Table 1.These data document statistical error of the measurements. The phase composition was determined by the X-ray diVractometer D 500 (Siemens, Karlsruhe, Germany) using Fig. 1 SEM micrograph of wollastonite deposit. Scale bar is 100 mm. Cu radiation. and the distributions can be characterized independently. Also Statistical methods note that to avoid distortion of the results by extreme values A number of complementary statistical methods are routinely (i.e., by impurities or contamination), two ( largest and applied in the materials science laboratory at the Institute.smallest) points of each distribution are usually not counted. The overall chemical composition is described by average content (a) of an element (oxide) and its variation by the Results standard deviation (s) of this element (compound). Further similar parameters which describe the distribution of elements Two minerals, wollastonite and steatite, illustrating diVerent are relative standard deviation (sr): eVects of interaction with plasma, were selected as examples of the procedure currently used.Results for other materials sr= s a ×100 [wt%] will be noted in the discussion. Micrographs of the two selected materials are shown in Fig. 1 and Fig 2. Both micrographs are and the relative diVerence between minimum (min) and taken in BE mode. maximum (max) element (oxide) content (rd): The results of the quantitative EDS-SEM analysis for wollastonite with all the statistical parameters are listed in Table 2.Similar results for steatite are shown in Table 3. rd= max-min a ×100 [wt%] The homogeneity coeYcient (H), basically a parameter similar Discussion to the segregation coeYcient, was introduced in ref. 5 as: Materials under invan be divided by the number of major components into three groups: two-component (mull- H= max-min Óa However, these parameters are not suYcient, especially if many phases are present.Their suitability depends on the overall amount of characterized element and the sensitivity of deposit properties to any particular element content. It has been found very useful to apply Weibull statistics.12 The use of Weibull statistics can assess the extremes, can visualize the presence of several phases and independently assess the homogeneity of each of them separately. The Weibull distribution: F(x)=1-e-(x-xu)m xb introduces a reliability parameter (m), which describes the distribution of observed values.The symbol F denotes the probability that any particular measurement will have a value lower than x; xu and xb are parameters of the distribution. If the distribution is narrow, the parameter m is large. For wide distributions m is small. This parameter has simple graphical representation in Weibull plots,12 where m relates to the slope of line fitted on the data points.If two distinct distributions Fig. 2 SEM micrograph of steatite deposit with typical pores. Scale bar is 100 mm. are found in the data set, the two parameters m can be found 472 J. Anal. At. Spectrom., 1999, 14, 471–473Table 2 Final results of wollastonite feedstock and deposit be the cause of the above-described phase change from CaSiO3 in the feedstock to Ca2SiO4 in the deposit. No. of analyses: 30 Similarly, a decrease in SiO2 content was observed in all the Wollastonite other investigated materials, regardless of the number of a (s) min max sr rd H m original components.This decrease was most important in Feedstock SiO2 51.6 (0. 6) 47.9 52.0 1.2 7.9 0.6 280 mullite and some garnets while in cordierite it was nearly CaO 47.7 (1. 0) 43.1 48.3 2.1 10.9 0.7 330 negligible. The loss of SiO2 content was on average 3 wt%. Deposit SiO2 47.9 (2. 8) 42.4 51.6 5.8 19.2 1.3 20 The observed changes in the microstructure homogeneity CaO 51.4 (2.7) 47.6 57.1 5.2 18.5 1.3 17 diVer widely even for similar materials and seem to vary depending on the point-of-origin of the raw material, manufacturing method, as well as on the spray parameters.Table 3 Final results of steatite feedstock and deposit This underlines the necessity for reliable and routine characterization of the deposits in engineering practice. No. of analyses: 30 Steatite Conclusions a (s) min max sr rd H m The various silicate starting materials, whether pure phases or Feedstock MgO 27.4 (6.1) 12.3 37.0 22.4 90.3 4.7 7 mixtures of minerals, do not behave in the same way during Al2O3 5.7 (3.2) 0.8 13.2 56.7 220.0 5.2 2 thermal spraying.Some of them become less homogeneous SiO2 63.1 (2.1) 59.5 67.2 3.4 12.2 1.0 50 Deposit MgO 29.0 (3.8) 13.9 34.2 13.1 69.7 3.8 18 (wollastonite), and the chemical variation may be observed Al2O3 6.4 (5.5) 1.7 35.1 85.3 520.6 13.2 8 even in the micrographs. Others can become more homo- SiO2 61.0 (3.4) 47.5 66.5 5.6 31.2 2.4 32 geneous, e.g., steatite.The materials may form amorphous phases, sometimes mixtures of amorphous and crystalline phases. A number of statistical parameters and data representations ite, wollastonite), three-component (cordierite, steatite), and have been discussed. Some of them are widely used, some of multi-component (basalt, garnets). Only raw data for twothem are less common. Each shows a diVerent view of the and three-component materials can be successfully processed.deposit microstructure and allows its characterization. A future Multi-component materials give very complicated sets of data task is to find the smallest set of the parameters which will and may need a diVerent approach. allow reasonably complete homogeneity characterization with The two materials presented here highlight changes in a clear and direct relationship to deposit properties. homogeneity created by the thermal spraying process.Wollastonite is a material which is very homogeneous as feedstock material, note very large m values and low s values Acknowledgment in Table 2, but is significantly less homogeneous in the deposit. This work was partially supported by grants GACR The reason for this behavior is in the phase composition. 104/96/1353, GACR 106/97/S008, AVK 1010601 and GACR X-ray diVraction has shown that the wollastonite feedstock is 106/97/0775. composed of a-CaSiO3 phase, while the deposit is composed of b-Ca2SiO4 and an amorphous phase.Steatite, on the other hand, shows an increase in the References homogeneity for MgO and Al2O3 contents and a decrease in 1 H. Herman, Sci. Am., 1988, 259, 112. homogeneity for SiO2. This can be again observed in the 2 P. Chra�ska, J. Dubsky�, B. Kolman, J. Ilavsky� and J. Forman, statistical parameters in Table 3. While the steatite feedstock J. Thermal Spray Technol., 1992, 1, 301. consists mainly of a MgSiO3 phase (as proved by XRD) the 3 R.McPherson, Thin Solid Films, 1981, 83, 297. 4 B. Kolman, J. Forman, J. Dubsky� and P. Chra�ska, Thermal Spray deposit is fully amorphous, although traces of unmelted Coatings: Research, Design and Applications, Proceedings of the crystallites from the feedstock have been identified by XRD. 5th National Thermal Spray Conference NTSC’93, Anaheim, CA, As an example of diVerent capabilities of above-mentioned 1993, ed. C. C. Berndt and T. F. Bernecki, ASM International, statistical parameters we can present the case of Al2O3 in Materials Park, OH, USA, 1993, p. 353. steatite (Table 3). The content of this compound varies from 5 B. Kolman, J. Forman, J. Dubsky� and P. Chra�ska, Mikrochim. about 1% to 35%. The description of homogeneity by s does Acta, 1994, 114/115, 335. 6 L. H. van Vlack, Physical Ceramics for Engineers, Addison-Wesley not reflect this spread, unlike the sr, rd and H, which all reflect Publishing Company, Inc., USA, 1964. this inhomogeneity significantly better.Similarly, m shows low 7 N. N. Ault, J. Am. Ceram. Soc., 1957, 40, 69. values (2 and 8), indicating low homogeneity of Al2O3 distri- 8 E. Wang and D. Wang, Ceramics, Adding the Value, ed. bution in this particular material. When diVerent statistical M. J. Banister, CSIRO, Melbourne, Australia, 1992, vol. 1, p. 359. parameters are compared, rd and H are those most sensitive 9 K. Neufuss, B. Kolman, J. Dubsky� and P. Chra�ska, Thermal to local extremes (one point in the measured set) while sr and Spray: A United Forum for Scientific and Technological Advances, Proceedings of the United Thermal Spray Conference 1997, m are more related to overall spread.Furthermore, while there Indianapolis, IN, USA, 1997, ed. C. C. Berndt, ASM International, is always only one sr for the sample studied, we can obtain Materials Park, Ohio, USA, 1997, p. 477. more m values in cases when phases with diVerent chemical 10 K. Neufuss, J. Ilavsky�, J. Dubsky�, B. Kolman and P. Chra�ska, composition are present. It is the subject of further work to Proceedings of the United Thermal Spray Conference 1999, find the most appropriate parameter that will be best linked Dusseldorf, Germany, submitted for publication. to variations in deposit properties. 11 P. Chra�ska and M. Hrabovsky�, Thermal Spray: International Advances in Coatings Technology, Proceedings of the International Both materials exhibit some loss of SiO213 after spraying, Thermal Spray Conference 1992, Orlando, Florida, USA, ed. which seems to be common for most of the silicate ceramic C. C. Berndt, ASM International, Materials Park, OH, USA, materials. This is probably related to preferential evaporation 1992, p. 81. of SiO2 during spraying, when feedstock particles are injected 12 W.Weibull, J. Appl. Mech., 1951, 293. into the stream of very hot plasma (several thousand K). 13 R. J. Ackermann and R. J. Thorn, Progress in Ceramic Science, Various data bases suggest diVerences in the vapor tensions of Pergamon Press, Oxford, UK, 1961, vol. 1, 63. silicon oxides and other compounds in the silicates used in this work. The observed decrease of SiO2 in wollastonite could Paper 8/06764C J. Anal. At. Spectrom.,

ISSN:0267-9477    

DOI:10.1039/a806764c

出版商:RSC    

年代:1999

数据来源: RSC