EDX Microanalysis in TEM

Transcription

1 EDX Microanalysis in TEM a) Review (brush-up) Generation and detection of X-rays, SDD detectors b) Quantification EDX in SEM, Interaction volume AF matrix corrections EDX in TEM (Cliff-Lorimer) c) Examples 1 X-ray generation: Inelastic scattering of electrons at atoms E electron_in > E electron_out SE inner shell ionization Characteristic X-ray emission Continuum X- ray production (Bremsstrahlun g, Synchrotron) SE, BSE, EELS 2

2 Forbidden transitions! quantum mechanics: conservation of angular momentum 3 Efficiency of X-ray generation Relative efficiency of X-ray and Auger emission vs. atomic number for K lines Ionization cross-section vs. overvoltage U=Eo/Eedge (electron in -> X-ray out) Light elements Auger Spectroscopy Heavy elements EDS SEM TEM -> Cu-K 8.1kV, HT 15kV U = 15/8.1 = 1.85 Light element atoms return to fundamental state mainly by Auger emission. For that reason, their K-lines are weak. In addition their low energy makes them easily absorbed. To ionized the incident electron MUST have an energy larger than the core shell level U>1. To be efficient, it should have about twice the edge energy U>2. 4

3 Right: Si(Li) detector Cooled down to liquid nitrogen temperature modern silicon drift (SDD) detector: no LN cooling required 5 X-Ray energy conversion to electrical charges: 3.8eV / electron-hole pair in average electronic noise+ imperfect charge collection: 130 ev resolution / Mn Ka line Detector acts like a diode: at room temperature the leak current for 1000V would be too high! The FET produces less noise if cooled! Li migration at room temperature! ->Detector cooling by L-N 6

4 Detection and artifacts Take care when looking for trace elements (low concentrations). Don t confuse small peaks with escape peaks! 7 Characteristic X-ray peaks EDX spectrum of (K,Na)NbO 3 Electron beam: 10keV Continuum, Bremsstrahlun g 10keV Duane-Hunt limit 8

5 c) Quantification First approach: compare X-ray intensity with a standard (sample with known concentration, same beam current of the electron beam) c i : wt concentration of element i I i : X-ray intensity of char. Line k i : concentration ratio c c i std i Yes, but.. I I i std i k i 9 Quantification When the going gets tough.. Correction matrix c c i i Z A F ki std i I I std i "Z" describe how the electron beam penetrates in the sample (Zdependant and density dependant) and loose energy "A" takes in account the absorption of the X-rays photons along the path to sample surface "F" adds some photons when (secondary) fluorescence occurs 10

6 Flow chart of quantification Measure the intensities and calculate the concentrations without ZAF corrections Calculate the ZAF corrections and the density of the sample Calculate the concentrations with the corrections Is the difference between the new and the old concentrations smaller than the calculation error? no Yes! stop 11 Correction methods: ZAF (purely theoretical) PROZA Phi-Rho-Z PaP (Pouchou and Pichoir) XPP (extended Puchou/Pichoir) with standards (same HT, current, detector settings) Standardless: theoretical calculation of I std Standardless optimized: «hidden» standards, user defined peak profiles 12

7 EDS in TEM Thin samples -> correction factors weak (A and F can be neglected) Very weak beam broadening -> high spatial resolution ~ beam diameter (~nm) High energy: artifacts! 13 PZT ceramics Interaction volume SEM (30KeV), bulk versus TEM (300KeV), thin film bulk Small interaction volume -> high spatial resolution for EDX Analysis! 20nm thick PZT TEM

8

9 Ternary systems

10

11

12 STEM point analysis PbMg 1/3 Nb 2/3 O 3 (bulk) Processing option : Oxygen by stoichiometry (Normalised) Spectrum Mg Si Nb Pb O Total Spectrum Spectrum Spectrum Spectrum Spectrum Spectrum Spectrum Spectrum Spectrum Max Min All results in Atomic Percent Pb(Zr,Ti)O 3 Thick films for MEMS SEM image of a wet etched (in HF/HCl solution) side-wall of 2 µm PZT film. All the 8 interfaces corresponding to the intermediate crystallization steps became visible indicating a compositional gradient (preferential etching) across the PZT layers.

13 CTEM dark field image STEM dark field TEM dark field images. Strong diffraction contrast STEM dark field images. The ramps in the gray level indicate changes of density or chemical composition (~atomic number). EDX Line-scan

14 EDX point analysis Quantitative EDX Analysis of the points indicated in the image Processing options : Oxygen by stoichiometry (normalised) results a Percent Spectrum Zr Ti Pb O Total Zr/Ti / / /39 STEM Element Mapping PMN/PT 90/10 (bulk)

15 Artifacts how to recognize/minimize them Analytical TEM of multifilament Nb 3 Sn superconducting wires Superconducting Nb 3 Sn cables for high magnetic fields 10-20T: increase current density, lower cost Potential Applications: NMR, Tokamak fusion reactors Large Hadron Collider (LHC), CERN Typical cable: 1 x 1.5mm cross-section 121x121 filaments of Nb 3 Sn in a bronze (Cu/Sn) matrix 0.5 mm Prof. R. Flükiger, V. Abächerli, D. Uglietti, B. Seeber Dept. Condensed Matter Physics (DPMC), University of Geneva

16 Processing bronze route Nano -engineering: controlled creation of imperfections of nm scale (coherence length) Cu and Ti are believed to play an important role at the grain boundaries: dirty grain boundaries = pinning Nb Ta Is it possible to detect Cu and Ti at the grain boundaries? What is the difference between the grain boundaries depending on where the additives are added to the unreacted material? Cu,Sn bronze Ti Nb 3 Sn Ti Heat treatment Nb Cu,Sn SEM: reacted filament (1 out of ) Typical problems: thinning of heterogeneous specimens: selective thinning bronze Nb 3 Sn filament Cross-section, polished mechanically to 30 um, ion milled until perforation STEM, Dark field: core of filament too thick, preferential etching of bronze matrix

17 Specimen preparation by focused Ion Beam (FIB): large areas with uniform thickness ideally for EDX Analysis in the TEM (STEM mode) 15um SEM (FIB) thickness:40-50nm STEM-DF Ion milling EDS, element maps STEM, Bright field FIB Sample #21 Spot analysis Line profile Nb 3 Sn Point Ti %at Nb %at Sn %at Ta %at Nb Tc/Jc useful bronze Sample #21 Sn

18 grain boundaries? Ti/Cu EDX line-scan Cu Nb Ti Ta Cu and Ti at the grain boundaries: width ~ coherence lenght (4nm) possible pinning centers!! Sn Sample #21 grain boundary without Ti Cu Nb Ta Ti Quantitative Line-scan Sn Sample #24

19 New possibilities due to SDD (silicon drift detector) technology X-rays SDD SDD Si(Li) Thickness µm 3mm Area 30,50,80mm 2 30mm 2 Det. Interval 4-10µsec µsec. Speed: cps cps Cooling Peltier Liq. N TECNAI OSIRIS Analytical TEM CIME 5x electrons 10x Det. area

20 Sn Nb Cu 400x400 pixels (5umx5um) spectra 4msec., (10min.) 2.5nA STEM DF Cu Sn 400x400 pixels (500nmx500nm) spectra 4msec., (10min.), 2.5nA

21 Synthetic sapphire: Al 2 O 3 Al2O3 with La (250ppm) in the grain boundaries Paul Bowen, M. Stuer EPFL-LPT HAADF STEM La at grain boundaries Cl (from synthesis)

22 QUALITATIVE EDX Mapping: GREAT! Chemical analysis on atom columns

23 STEM EDX analysis of Cr diffusion into the Cu stabilizer of Nb 3 Sn strand annealed for 200h B. Bartova, G. Arnau Izquierdo, B. Bordini, P. Alknes, M. Cantoni TEM lamella was prepared by FIB. 30 microns wide area was lifted out next to the chromium plating. Three windows of 7 microns were thinned down to the electron transparency. 2-3 microns thick areas were kept in order to ensure the stability of the sample. In following slides STEM EDX analysis from each of the thinned windows is presented. Marker made by FIB to recognize the Cr side of the lamella. 1 st window in TEM lamella 3 microns from Cr plating BF and DF HAADF images of window 1 Cr S O STEM EDS mapping revealed the presence of Cr-S rich precipitates that might contain also oxygen. However, oxygen was not taken into account since its proper quantification at very low quantities is difficult.

24 1 st window in TEM lamella 3 microns from Cr plating 3 spectra were analyzed in window 1 at higher magnification. 2 precipitates and the matrix st window in TEM lamella 3 microns from Cr plating Precipitate 1- window 1 Precipitate 2 window 1 Element series [wt.%] [norm. wt.%] [norm. at.%] Error in wt.% Chromium K-series Copper K-series Sulfur K-series Sum: Element series [wt.%] [norm. wt.%] [norm. at.%] Error in wt.% Chromium K-series Copper K-series Sulfur K-series Sum: The precipitates contain chromium and sulfur and their ratio is close to 1:1. The peaks at K (5.411) and K lines for chromium and K for sulfur are evident.

25 1 st window in TEM lamella 3 microns from Cr plating Element series [wt.%] [norm. wt.%] [norm. at.%] Error in wt.% Chromium K-series Copper K-series Sulfur K-series Sum: The quantification of the spectra from matrix is showing 0.01 at.% of chromium. In order to verify chromium presence, original spectra needs to be magnified. For certain spectra the automatic background deconvolution was slightly ajusted.

26 Outlook: STEM EDX tomography on nanometric devices Kevin Lepinay 1 *, F. Lorut 1, R. Pantel 1, T. Epicier 2 1 STMicroelectronics, Crolles, France 2 MATEIS Lab, INSA Lyon, France kevin.lepinay@st.com Microscopy needs in semiconductor industry Needs in chemical analysis materials composition, diffusion and dopants EDX technique (photon detection from electron ionization) EDX mappings in only a few minutes, detecting most of elements even dopants such as Arsenic. TEM EDX STEM O Si Ti Hf As Ni EDX mapping of a 32nm NMOS transistor [R.Pantel] performed on the Tecnai OSIRIS The idea is the coupling of these two techniques to obtain a nanometric 3D chemical analysis.

27 STEM EDX tomography principle EDX mapping acquisition at each tilt angle Tomography specific FIB sample preparation needle shaped Sample preparation strategy (SEM images) Prepared sample needle diameter between 100nm and 200nm(TEM image) STEM EDX tomography analysis flow ACQUISITION Esprit Manual Step: Tilt range defined Dose (probe current, EHT) EDX Dwell time EDX mapping 800x800px 5min/map, step 2 DATA EXTRACTION Esprit / ImageJ Nickel Tungsten Copper Tantalum Titanium Manual Step: Extraction of all elemental mapping from each EDX mapping ImageJ treatment (8 bit conversion, operations, filtering) Semi Automated Step: MRC volume creation for every element Alignment shifts calculation with 1 element Apply alignment shifts to every volume SIRT Reconstruction Alignment parameters RECONSTRUCTION Inspect3D VISUALISATION Amira Manual Step: Volume rendering creation Information extraction (ex: defect morphology)

28 Experimental method and case study Acquisition parameter: EHT 120keV, probe current : 1nA Tilt range 180, tilt step 2 Dwell time 500µs, mapping acquisition 5min/map Voxel representation can be obtained Element per element Combination of elements Ni WCu Ti 3D chemical rendering of FDSOI transistor 3D chemical representation per element Ta Ti Si Cu W Ni Hf Experimental method and case study Extracted slices in all directions Parallel to the gate stack Ni / Si / Ti/ Hf STI step height No masking effect Perpendicular to the gate Ni / Si / Ti/ Hf Cu Metal line TaN Barrier Ta Ti Si Cu W Ni Hf In plugs W plugs TiN Barrier NiSi siliciuration