High Sensitivity EPMA: Past, Present and Future John Donovan CAMCOR University of Oregon (541) camcor.uoregon.edu
The Past: EPMA/SEM (from Goldstein, et. al. 1988) : Comparison of EDS to WDS, Equal Beam Current, pure Si and Fe, A (0.01 nA), 25 keV Comparison of EDS to WDS, Optimized Conditions, 15 keV, 180 seconds counting time: EDS : 2 x A (2 nA) to give 2K cps spectrum to avoid sum peaks WDS : 3 x A (30 nA) to give 13K cps on Si spectrometer (< 1 % dt)
More recent data for EPMA/SEM:
“ The detection limit cannot be reduced indefinitely by accumulating more counts, however, because systematic errors in the background correction eventually become significant. - Stephen Reed 4 Accuracy (not precision) in characterizing the continuum becomes the limiting factor...
Other Artifacts: “Holes” in the Continuum PPM wt.
Reality Check: Accuracy at the 400 PPM Level? Note: Blank level (C level ) can be non-zero
PET LPET 5 Spectrometers Still… We Need to Improve Sensitivity as well..
(normal PET crystal) 1 ppm 170 ppm NIST SDD EDS is Dead! 5 Spectrometers
WDS Analysis of Hg (polymer door frames from suspected Mexican facility) TakeOff = 40.0 KiloVolt = 20.0 Beam Current = 50 Un 6 std-flex Results in Elemental Weight Percents ELEM: Hg Pb Cr TIME: AVER: SDEV: Detection limit at 99 % Confidence ELEM: Hg Pb Cr AVER: PPM of Hg easily detected, with 48 PPM sensitivity Checked with EDS -count 200 sec - several nA -no Hg found Checked with WDS - count 20 sec - 50 nA - Hg found Why?
Polymer door frames (2000s) from suspected Mexican facility, Check with EDS for 100 sec, 20 keV, 50 nA... nothing sec counting time
Still nothing sec counting time
Hg peaks barely visible… 1000 sec counting time, with Be window inserted (to remove C and O)
15 keV, 200nA, 600 sec on-peak, 600 second off peak, Ti Ka, LPET + PET (aggregate intensities) So what exactly can WDS do on a “typical” quantitative analysis?
15 keV, 200nA, 300 sec on-peak, 300 second off peak, Zr La, LPET + PET (aggregate intensities)
Every sample is beam sensitive -at a sufficiently high beam current... Usually thermally insulating samples (e.g., non conductors…) Classical beam sensitive samples (e.g., alkali, hydrous glasses) Orientation dependent intensity changes over time (e.g., apatites) Trace element measurements (high beam currents, long counting) Use alternating on and off-peak measurements (constant delta) Extrapolate to zero time intensities Use a “blank” correction to apply a systematic error offset SiO2 GlassSiO2 Quartz
F K in VG2 Glass (1800 secs total count time)
Correcting for Intensity Loss (and Gain) Results in Oxide Weight Percents ELEM: Na2O SiO2 Al2O3 MgO TiO2 MnO P2O5 Cl FeO K2O CaO O H2O SUM AVER: SDEV: SERR: %RSD: VOL%: DEV%: VOLF: LINEAR LINEAR LINEAR LINEAR LINEAR
But Not Always What You Expect!
Hyper-exponential Loss Results in Oxide Weight Percents ELEM: Na2O SiO2 Al2O3 MgO TiO2 MnO P2O5 Cl FeO K2O CaO O H2O SUM AVER: SDEV: SERR: %RSD: VOL%: DEV%: VOLF: QUADRA LINEAR LINEAR LINEAR LINEAR Two exponential processes with different decay constants overlapping in time (?)
The Present “state-of-the-art”: MultiPoint Backgrounds: Combined Qualitative and Quantitative acquisition ThSiO4 (Pb free, i.e., “blank”) Th Mz1 and Mz2
Fluorescence from Si K Characteristic Fluorescence from Al Continuum ~100 PPM Si
Back To The Future: A proposal for a TEPNA instrument The Transmission Electron Probe Nano Analyzer integrates several new detection technologies to optimize compositional characterization with a target spatial resolution of ~10 nm for “as deposited” films and particles in the range of tens to hundreds of nanometers in thickness, while still attached to electron opaque substrates.
50 nm Bi2Te3, 30 keV, 20 nm beam
Thin films are a trace element problem... ~50 nm Fe, Nb, Se film on Si wafer (20 keV, 30 nA) Is that my signal? Nope… Si sum peak!
18 hour integration at 30 nA can provide significant sensitivity... Newbury ID “blunders” are still here… Si sum peak identified as Sn… Nb peak not identified...
Even though the Nb Ka peak is visible...
Like thin films, nano-particles also present a sensitivity problem... WO 3 nano-particles on Si Si sum peak is not 3 sigma, but neither are the W peaks!
The increase in signal as a 10 m diameter beam is scanned over a region containing a monolayer of Hf atoms deposited on a Si substrate using a Cameca SX50 EPMA WDS Monolayer Detection Demonstrated nm thick, 10 um dia. 100 nm thick, 0.01 um dia. Requires a 1000 fold improvement in sensitivity! (1,000,000 x fewer atoms but 1000 x thicker film)
How do we improve sensitivity 1000 fold? Utilizing high energy emission lines with higher fluorescent yields e.g., Nb L = 3.5%, Nb K = 74% (20-30 fold improvement) Energy filtering of Be exit windows for high energy emission lines (?) Why not do it now? Goldstein et. al. 1992
Highest effective fluorescent yields are found for element emission lines whose absorption edges are higher than 8 keV Zn K is keV Nb K is keV In 2 cm of Ar 37% of Zn K trans. 86% of Nb K trans. In 2 cm of Xe.05% of Zn K trans. 59% of Nb K trans. While still retaining soft x-ray sensitivity!
Other sensitivity improvements are possible... Small FC and/or large area crystals (3 to 4 fold improvement) Multiple WDS spectrometers in “aggregate” mode - 2 to 5 fold improvement using only software Increased counting time/beam current in electron “transmission mode” - 30 keV beam through 100 nm of FeS2 loses ~30 eV of energy - assume 2 to 5 fold improvement by increasing time/current Reduced continuum signal using “faraday cup” TEM grid holders - preliminary measurements show a 30% reduction in continuum
An electron beam instrument that integrates several innovations to optimize compositional characterization with a target spatial resolution of ~20 nm for samples in the range of tens to hundreds of nanometers in thickness on various electron opaque substrates. Transmission Electron Probe Nano Analyzer (TEPNA) The TEPNA complements existing analytical techniques by providing an unmet need for quantitative compositional analysis conveniently intermediate between that currently achieved by wavelength dispersive x-ray (WDX) electron probe micro analysis (EPMA) and energy dispersive x-ray (EDX) analytical electron microscopy (AEM).
EPMA vs. LA-ICP-MS Larger error bars for EPMA reflect actual small scale compositional variation.
Current/Future Capabilities of High Sensitivity EPMA/TEPNA WDS: Bulk Analysis: presently single digit PPM sensitivity (and accuracy) -AFTER correction of various continuum artifacts, e.g., “blank” Thin Film/Particle Analysis: feasible now for major/minor elements -with typical ~1 um beam diameters on samples >50 nm thick -requires more sensitivity for <50 nm beam TEPNA: Transmission Electron Probe Nano Analyzer - utilize high energy nm electron beam (transmission mode) - high fluorescent yield lines (> 8 keV) - tandem gas flow/SDD photon counters (full energy sensitivity) - large area/small FC crystals/spectrometers - aggregate intensities in software