Electron Probe Microanalysis in Materials Characterization Karakterizacija materialov z metodami elektronske mikroanalize Z. Samardžija1, M. Čeh, IJS Ljubljana Prejem rokopisa - received: 1996-10-04; sprejem za objavo - accepted for publication: 1996-11-22 Energy dispersive (EDS) and wavelength dispersive (WDS) X-ray spectroscopy as electron probe microanalytical (EPMA) techniques, are used for the determination of chemical composition of solid materials. EDS is suitable for fast qualitative and quantitative analysis. It has however limited sensitivity in quantitation. WDS guantitative analysis, with appropriate reference standards, has better sensitivity and accuracy. WDS procedure is especially suitable for quantitative analysis of elements which overiap in the EDS spectra, in the čase of low elementaI concentrations in the samples, and for analysis of light elements. In general, quantitative analysis with both EDS and WDS is performed on the micro-level where the analyzed volume of the material is about 1 |xrrr. Results of the analyses of selected lead glass samples, BaTi03 - doped ceramics, and ALNICO magnets are presented and discussed. Key words: electron probe microanalysis Energijsko disperzijsko spektroskopijo (EDS) in valovno disperzijsko spektroskopijo (WDS) rentgenskih žarkov smo uporabili kot tehnike elektronske mikroanalize (EPMA) za določanje kemijske sestave materialov. EDS je primerna za hitro kvalitativno in kvantitativno analizo, ima pa omejeno občutljivost pri kvantitativni analizi. WDS kvantitativna analiza z uporabo referenčnih standardov ima višjo občutljivost in natančnost. WDS je posebej primerna za kvantitativno analizo elementov, ki jih ne moremo ločiti v EDS spektrih zaradi prekrivanja, v primeru elementov nizkih koncentracij in za analizo lahkih elementov. Kvantitativno analizo z EDS ali WDS naredimo na področju velikosti nekaj mikronov. Pri tem analiziramo približno 1 \irrr3 materiala. V prispevku poročamo o rezultatih analiz izbranih vzorcev stekla, dopirane BaTiC>3 keramike ter ALNICO magnetov. Ključne besede: elektronska mikroanaliza 1 Introduction Electron probe microanalysis (EPMA) deals with the analysis of characteristic X-rays, emited from the region of the solid material where the electron beam impinges. The analysis yields compositional information of both qualitative and quantitative nature. X-ray spectral meas-urement can be performed by energy dispersive spectros-copy (EDS) with solid state Si(Li) detector or by wav-elentgh dispersive spectroscopy (WDS) with crystal spectrometers. In modern commercial instruments EDS and WDS are integrated with the scanning electron mi-croscope (SEM). This combination allovvs to obtain SEM micrographs of samples which contain topographic and compositional informations and to perform chemical analysis (EDS or WDS) in selected regions by measuring the energy and intensity distribution of the X-rays. EPMA is integrated with modern computers and soft-ware support for ali data acquisition and manipulation, including digital image processing, qualitatitive and quantitative elemental analysis, automatization of the measurements, etc.. The major advantage of EPMA tech-niques is the possibility to carry out chemical analysis on a micrometer scale. The analyzed volume (about 1 |im3) contains small amount of material, typically in order of 10"10 - 10-" g. ' Zoran SAMARDŽIJA. dipl. fizik Inštitut Jožef Štefan 1001 Ljubljana. Jamova 39 Nowdays these techniques are not restrieted to basic scientific research but also common in laboratories de-voted to the development of "high-tech" products, qual-ity control in produetion, failure analysis, enviromental care, etc. In the Ceramics department of the "Jožef Štefan" Institute SEM, EDS, and WDS techniques are used for the characterization of different materials including ceramics, metals, alloys, composites, polymers, thin films, glasses, etc. A short list of the EPMA capabilities is shown in Table 1. This paper reports on the results of applied EPMA methods using examples of analysis of lead glass, BaTiOj based ceramics, and ALNICO magnets. 2 Experimental In general it is required that samples for EPMA analysis are flat, polished and electroconductive1. An ex-ception is the SEM examination of rough surfaces, frac-tures, etehed samples and particles where the information is focused on topography and microstructure. Standard metallographic techniques are usually applied for sample preparation. Non-conductive ceramic and glass samples are coated with a thin film of carbon or metal to prevent charging. Samples for elemental analy-sis (EDS, WDS) were coated with carbon to reduce ab-sorption of the emitted X-rays. Preliminary investigations of the samples by SEM-EDS is usually recommended to specify the microana-lytical problem, as far as possible, with regard to the fol- Table 1: List of SEM-EPMA methods applied for materials characterization Tabela 1: Pregled SEM-EPMA metod za karakterizacijo materialov SEM EDS analysis WDS analysis microstructure. grain size qualitative: quantitative: qualitative: X-ray maps topography element identification standardless qualitative: line profile compositional images with standards quantitative with standards particle size. morphology line profile ZAF. PRZ quantitative: line analysis image analysis X-ray maps corrections ZAF. PROZA corrections phases present in the sample. A micrograph displaying a defect is shown in figure 1. The right side of the micrograph is recorded at a higher magnification showing the presence of four phases, which are marked as A, B, C, and D, respectively. Phases in the region containing defects differ in gray-level contrast. EDS analyses of the matrix M and phases A, B, C, and D were done at 15 keV, 0,5 nA. take off angle 40°, and acquisition tirne 100 s. EDS spectra were quantified using PRZ matrix correc-tion programs2. Both standardless and analysis using ox-ide standards were performed. Calculated quantitative results for unknown oxide compounds are given in Table 2. When analyzing alkaline glass matrices containing Na and K one should take into account possible migra-tion of alkaline ions, induced by the electron beam dur-ing microanalysis. Experimental parameters for EDS spectra acquisition should be carefully determined in or-der to minimize this migration and to obtain reasonable analytical results. The stability of the standards and sample material under the electron beam could often be dif-ferent, and may cause problems in the quantitative analy-sis with standards. These problems are avoided in the standardless procedure. The comparison of quantitative results shows quite satisfactory standardless analysis. In the čase of well-defined EDS spectral peaks, without spectral interference, good analytical precision was achieved with r.m.s. (root mean square) errors for calculated element concentrations up to 5%3. The presence of AI2O3 (phase A) and ZrC>2 (phase B) inclusions in the glass a due to the erosion of furnace refractory material. Phases C and D were formed by the reaction of the aggressive alkaline glass melt with pieces of refractories. These glassy phases differ in the concentrations of Na20, Zr02, PbO, and K2O, whereas the AhOi and SiC>2 content remains practically the same. Table 2: Average results of quantitative EDS analysis of glass and defects (oxides, wt%) Tabela 2: Rezultati EDS kvantitativne analize stekla in defektov (oksidi, mas.%) Na20 AI2O3 Si02 Zr02 PbO K.O PHASE I II I II I II I 11 I II I II matrix M 3,2 3,1 - - 59,7 58,8 - - 28,4 27,9 8,7 10,2 A - - 100 100 - - - - - - - - B - - - - - - 100 100 - - - - C 5,4 5,4 32,1 32,4 33,1 33,2 1,2 1,4 18,0 17,9 10,2 9,7 D 3,4 3,2 32.3 32,4 32.5 33,0 4,2 4.3 7,5 7,4 20,2 19,7 I - standardless analysis, II - analysis with standards 64 lovving application of more sophisticated techniques, such as WDS. A JEOL JXA 840A electron probe mi-croanalyzer equipped with EDS, two WD spectrometers and Tracor Series II X-Ray Microanalysis System was used for overall analysis of the samples. 3 Results and discussion 3.1 Defects in lead glass The aim of work was SEM examination and EDS qualitative and quantitative analysis of defects in lead glass. SEM micrographs of the defects in lead glass were recorded using both secondary (SE) and backscattered (BSE) electrons, emphasizing the Z-contrast of different Figure 1: Combined SE/BSE electron micrograph of the defects in lead glass: M - glass matrix; A, B. C. D: glass defect phases Slika 1: Posnetek SEM (s sekundarnimi in odbitimi elektroni) defektov v svinčevem steklu: M - matrica stekla; A, B. C, D: faze prisotne v področju defekta i j. žL ' j ' * y ' . - v. C,.Si * mm^Mizž-jm IIHH^H Jk + ja»J ^Hnif B 7,4'* ■z. . *jL- ... |^ [jESVm \ * f k7* » ■ wmmmmmsM Figure 2: BSE electron micrograph of La-doped BaTiOs; phase marks: 1:1:4 - BaLa2Ti40i2 lamellae, BT4 - polytitanate BaTi40>). BLT -La-doped BaTiOj grains Slika 2: Slika odbitih elektronov mikrostrukture keramike BaTi03 dopirane z La; oznake faz: 1:1:4 - lamele BaLa2Ti40i2, BT4 -polititanat BaTijO«. BLT - zrna BaTiOs dopirana z La 3.2 La-doped BaTiOs ceramics The microstructure and grain size of a La-doped BaTiCb sample is shown in figure 2. It is important that the size of the analyzed phase is quite large (minimum 5-10 pm of average diameter) in order to avoid the gen-eration of X-rays from adjacent phases. Because of peak overlapping of the most intense Ba, Ti, and La spectral lines in the EDS spectrum, quantitative analysis based on EDS (resolution 150 eV) is not possible. The better reso-lution of the WD spectrometer (5-10 eV) allovvs analysis of the BaLai, TiKai, and LaLai lines without spectral interference. Furthermore, higher X-ray collection effi-ciency and better analytical sensitivity improve the accu-racy of quantitative analysis4. WDS analyses were performed on ten La-doped BaTiCb grains using a PET crystal, at 20 keV, 10 nA, and 40° take off angle. Measured X-ray intensities in the samples were transformed into k-ratios relative to cali-brated X-ray intensities from the BaTiCb and La2Ti207 standards. Quantitative analysis was performed through the ZAF matrix correction procedure, transforming the measured k-ratios into element concentrations. The print-out of WDS-ZAF results for measurement performed on one point of sample 2 is given in Table 3. Table 3: Printout of the results of quantitative WDS-ZAF analysis Tabela 3: Izpis rezultatov kvantitativne WDS-ZAF analize element wt% norm at% oxides wt% norm —---wt% _wt% Ba 47,73 47,77 16,20 BaO 53,30 53,38 T' 19,74 19,76 19,22 Ti02 32,93 32,98 La 11,62 11,63 3,90 La.Oj 13,62 13.64 --20,83 20.84 60.68 total 99,85 100 00 total 99,92 100,00 100.00 *by difference An unnormalized analytical total of 99.85% for oxide wt%, indicates a very good result (totals betvveen 99 in 101% are treated as good analytical results1) of the ap-plied microanalyticaI method. In Table 4 the average results, of measurements of cation concentrations, in three La-doped BaTiCb samples are presented. Oxygen con-tent was calculated by difference, which is the usual ap-proach in the analyses of oxide compounds. Data were calculated to the perovskite AB03 formula taking into account that La3+ substitutes on Ba2+ sites with the for-mation of Ti-site vacancies5. Table 4: WDS results for La-doped BaTiC>3 samples Tabela 4: Rezultati WDS mikroanalize vzorcev La-dopiranega BaTi03 Composition of La-doped BaTi03 phase: Bai.xLa,Tii.„dfVT,V"\«01 Sample tla (at%) La(at%) Ti Y(mol%Tfi) 1 2 3 17,35 ± 0,28 2,53 ± 0,04 19,43 ± 0,25 16,24 + 0,20 3,97 ±0,04 19,58 10,25 14,69 + 0,23 5,84 + 0,09 18,63 + 024 13 20 28 The quoted standard deviations are related to data measured on the various analyzed grains, in the samples, with r.m.s. errors for calculated element concentrations between 1-2%. As an illustration of the achieved analyti-cal precision the La-doped BaTi03 sample 3 can be con-sidered as an example. The starting composition of the sample was expressed by the formula Bao.7i5Lao.285 Tio.928(VTi)""03. The sintered sample was monophase material and the calculated formula based on the results of WDS microanalysis was Bao7i5Lao285Tio907 (VTi)""03. The results allow the determination of solid solubility and investigation of the mode of dopant incorporation in BaTi03. The basic advantages of WDS microanalysis are direct analysis of phases of interest in the chosen samples and quantitative analysis with improved analytical sensitivity and precision. 3.3 ALN1CO magnet s ALNICO magnets were analyzed. SEM examination of a polished cross section of the sample reveals the pres-ence of inclusions in the matrix and defects on the sur-face. The matrix composition was determined by quanti-titative EDS standardless analysis using five different points along the matrix. Results are given in Table 5. Good reproducibility of the quantitative results was ob-tained for major constituents of the matrix (Al, Ni, Co, and Fe) with r.m.s. errors between 1,5% and 3%. Ti, Cu, and Nb are present in lower concentrations consequently increasing r.m.s. error from 7% (Ti) to 15% (Cu), due to poor X-ray counting statistics and possible variation in element concentrations from point to point. Neverthe-less, results affirmed the use of EDS standardless analy-sis as a fast method to determine the elemental composition of the sample. Table 5: Results of EDS quantitative analysis: composition of ALNICO matrix Tabela S: Rezultati kvantitativne analize EDS: sestava matrice ALNICO Point Al(wt%) Ni(wt%) Co(wt%) Fe(wt7o) Cu(wt%) Ti(wt%) Nb(wt%) 1 8,1 14,9 35,1 33,0 1,9 5,9 1,1 2 8,1 14,5 35,9 33,2 1.4 5,9 0,9 3 7,8 15,7 34,9 33,4 2,0 5,2 1,1 4 7,7 15,0 35,6 32,9 2,2 5,6 1,0 5 8,3 15.8 36,3 32,0 1,5 5,0 1,1 average S.0±0.2 15.2+0.5 35.610,5 32.910.5 1,810,3 5,510.4 1.010.1 Figure 3: Defect near the surface of ALNICO magnet: (A) SEM micrograph, (B) X-ray dot map for Ti, (C) X-ray dot map for O Slika 3: Defekt pri površini ALNICO magneta: (A) posnetek SEM, (B) slika porazdelitve titana, (C) slika porazdelitve kisika Defects were analyzed qualitatively by WDS X-ray mapping. X-ray maps are generally used to show the dis-tribution of a particular element on a selected area of sample. A SEM micrograph and the corresponding X-ray maps, of the defect, near the surface of the magnet are shown in figure 3. Surface defects contain Ti and O and were identified as titanium oxide layer on the sample surface. Oxygen is present only near the surface vvhereas titanium is distributed over a wider area inside the raa-trix, in form of titanium-carbides and/or titanium-carbo- Figure 4: Inclusions in ALNICO matrix: (A) SEM micrograph. (B) X-ray dot map for Ti, (C) X-ray dot map for C Slika 4: Vključki v ALNICO matrici: (A) posnetek SEM, (B) slika porazdelitve titana, (C) slika porazdelitve ogljika nitrides. The presence of C and N was confirmed by WDS mapping. Similar analysis was performed for de-fects-inclusions which are found in the ALNICO matrix (figure 4). X-ray maps reveal the presence of Ti and C in inclusions. These defects were identified as titanium car-bides. Results of EPMA analysis of ALNICO magnets showed a regular composition of the matrix and the presence of defects containing light elements (C, O, N). These defects are identified qualitatively by WDS X-ray maps as a titanium oxide layer, on the sample surface, and titanium carbide inclusions in the matrix. Analysis allows the determination of the defects' origin and conse-quently their elimination with improvements in the pro-duction process. 4 Conclusions MicroanalyticaI methods EDS and WDS in combina-tion with SEM were applied to characterize three differ-ent types of materials: lead glass, ceramics, and alloys. The microstructures of the samples were investigated on a SEM using both secondary electron and backscattered electron imaging. Usually information of elemental com-position is obtained first by EDS qualitative analysis. Chemical composition of the samples or phases in the samples was determined by EDS quantitative analysis with and/or without standards. WDS quantitative analy-sis with appropriate reference standards allows to meas-ure the elemental concentration with better sensitivity and accuracy. WDS procedure is especially suitable for the quantitative analysis of elements which overlap in the EDS spectra, in the čase of low elemental concentrations in the samples, and for light element analysis. The relative accuracy of EDS quantitative analysis achieved is between 3-10% vvhereas in the WDS quanti-tative analysis accuracy between 1-2% is obtained. EDS analysis has a limited sensitivity (detection limit about 0.1 wt%) and limited quantification in situations of com-plex materials with spectral interference (overlap) and ultralight elements. However, combination of SEM-EDS regarding to its flexibility, ease of operation, and data in-terpretation is widely used. WDS technique is time con-suming but more sensitive and accurate (detection limit about 0.01 wt%) quantitative procedure suitable for vvide range of microanalytical problems. Acknowledgements This publication is based on work partially sponsored by the U.S. - Slovene Science and Technology Joint Fund in cooperation with the Ministry of Science and Technology of Slovenia under Project Number US-SLO 95/6-05. 5 References ' J. I. Goldstein et al., Scanning Electron Microscopy and X-Ray Mi-croanalysis, Plenum Press, New York, 1992 2 G. F. Bastin, H. J. M. Heijligers, F. J. J. van Loo, A further improve-ment in the Gaussian <)>(pz) approach for matrix correction in quanti-tative electron probe microanalysis, Scanning, 8, 1986, 45-67 3 J. L. Pouchou, Standardless X-Ray Analysis of Bulk Specimens, Mik-rochimica Acta, 114/115, 1994, 33-52 4J. S. B. Reed, Electron Microprobe Analysis, Cambridge University Press, 1975 5 D. Makovec, Z. Samardžija, U. Delalut and D. Kolar, Defect Structure and Phase Relations of Highly La-Doped BaTi03, Journal of American Ceramic Society, 78, 1995, 8, 2193-2197