2. Core Community Specifications for Purchase and Acceptance of Electron Beam Instruments John J. Donovan Department of Chemistry CAMCOR (Center for Advanced Materials Characterization in Oregon) University of Oregon Eugene, OR donovan@uoregon.edu http://epmalab.uoregon.edu (541) 346-4632 MAS 2006, Chicago

3. Paul Carpenter, MFC/NASA John Fournelle, U Wisconsin Dan Kremser, Battelle AMA Ed Vicenzi, Smithsonian Inst. Chi Ma, Cal Tech Greg Meeker, USGS Ryna Marinenko, NIST John Armstrong, American Univ Dale Newbury, NIST plus many others... A personal, opinionated and incomplete perspective... Acknowledgements Special thanks to:

4. Possible Application EPMA (variants: high resolution, geochron, low voltage) SEM (variants: high vacuum, variable pressure, environmental) Specification of Configuration: Electron Guns (Tungsten, LaB6, Thermal, Cold) Vacuum systems (Diffusion vs. Turbo) WDS- Wavelength dispersive spectrometers (Ar vs. Xe) EDS- Energy dispersive spectrometers (Si(Li) vs. SDD) EBSD- Electron Backscatter Diffraction, CL, SE, etc. Specification of Performance: Column/gun stability Stage/Spectrometer reproducibility Optical depth of field/auto-focus Crystal/Detector count rate and resolution

5. Religion? EPMA: Jeol vs. Cameca SEM: Zeiss vs. FEI vs. Hitachi vs. Jeol EDS: Oxford vs. Edax vs. Thermo vs. Bruker (PGT/Rontec) EBSD: TSL vs. HKL etc. Cameca SX100 with Thermo System Six Zeiss Ultra with Oxford Inca and Nabity e-Lithography FEI Quanta 200 with ????? EDS/EBSD Agnostic? This is NOT an endorsement! Full Disclosure:

6. Test Results 8/2005 - present Cameca SX100 running Peak Sight and Probe for Windows

7. Configuration- Gun Type Tungsten (W) Cheap, reliable, easy to change, stable, resolution? LaB6 Small initial cost, stable?, lifetime cost?, fragile, 2x W resolution Thermal Field Emission Expensive, high resolution, stable?, easy to use Cold Field Emission (Jeol SEM only?) Poor stability?, 2x FEG resolution?, cost?

8. Gun/Column Performance Hv stability and accuracy Beam stability and accuracy Column alignment: beam collimation, beam shift Magnification accuracy, scan rotation accuracy Faraday cup beam current linearity

9. Hv Stability @ 50 keV Spec: 50 ppm per hour

10. Beam Stability Beam current stability @ 15keV 10nA during 1 hour ~0.06% Beam current stability @ 15keV 10nA during 12 hours ~0.15% Beam current stability @ 30keV 20nA during 1 hour ~0.025% Beam current stability Without Regulation @ 20keV 20nA during 12 hours ~0.2% Spec: 0.1% or less per hour and 0.6% or less per 12 hours and 1.0% or less in 24 hours

11. Beam current stability during stage movement? ~0.06%

12. Faraday Cup Linearity

13. Column (image) Stability Over Time Spec +/- 0.5 um

14. Column Alignment (beam shift vs high voltage) Spec: < 1 um

15. Beam collimation 100 um W aperture Spec: K = 0.0001 (0.01wt% or 100 ppm)

16. Configuration- Vacuum System Diffusion oil pump Cheap, reliable, easy to maintain, backstreams oil Turbo molecular pump Expensive, reliable?, lifetime cost? Turbo molecular with scroll pump (totally dry) Expensive, reliable?, lifetime cost? Diffusion pump with Freon chilled baffle (Jeol & Cameca) ~$28K, reliable?, minimal backstreaming

17. Poly-Cold “Freon” chilled baffle

18. Configuration- WDS Take-off angle: 40 degrees 52.5 degrees for ARL) Focal Circle: 140mm vs. 160mm (127mm for ARL) Geared vs cable driven (band driven for ARL) Optical encoding vs step counting (fundamental engineering decisions)

19. Proportional Detectors Flow Detectors: Cost of gas (Ar, P10) Clean (low noise) Short term instability Long term stability Poor high energy sensitivity* Sealed Detectors: Xe, Xe-Kr, etc. Short term stability Long term instability (contamination = noise) Good high energy sensitivity *Transmission of Zn Ka (~10 KeV) in 30 PSIA of 2 cm of P-10 gas is over 50%. Add low resolution solid state detector to exit window!

20. EDS Configuration SDD (Silicon Drift Detector) new technology (~1-3 years) peak shift, peak shape artifacts? limited high energy sensitivity? unbelievable throughput! Si(Li): tried and true, mature technology stable, known electronic response excellent high energy sensitivity (also Ge) throughput limited to ~30k cps Micro-calorimeter?

21. EDS Performance* Light element sensitivity Si sum peak intensity Count rate linerarity Si internal fluorescence peak Energy shift vs. count rate Detector resolution, e.g. E.g., resolution shall be 129 eV (or less) at Mn ka at 15 KeV and at a count rate of 2500 cps (or more) and the resolution shall degrade by less than 2 eV when the count rate is varied from 1,000 to 10,000 cps. Under the same conditions, the detector resolution shall also be 65 eV at F ka (or less) at a count rate of 2500 cps (or more). Under the same conditions used to measure these resolutions, confirm that the full width at 1/10th the maximum for Mn ka is less than or equal to 1.9 FWHM. The detector shall maintain this specification after repeated thermal cycling and/or thermal conditioning. In addition the detector resolution shall be 138 eV (or less) at Mn ka (measured on Mn metal) at 15 KeV and at a count rate of 10,000 cps (or more). *defined at specified count rates and dead times

22. Stage Performance Stage reproducibility based on video imaging of a small feature in SE mode Spec: < 1 um

23. Testing All Four Stage Limits as well...

24. Auto Focus Reproducibility Cu coated with 20 nm of carbon (dark blue color)

25. Spectrometer Reproducibility Spec: the intensities shall vary by less than 1.2% without a backlash or re- peak procedure Moving Peak to Peak...

27. Spectrometer Reproducibility (w/ crystal flip) Spec: 2%

28. K-ratio Reproducibility Spec: K-ratio agreement within 0.5% with a counting period sufficient to achieve 0.2% relative standard deviation. ZCOR ~20% 5 PET crystals

29. K-ratio Reproducibility Spec; K-ratio agreement within 0.3% for all TAP crystals with a counting period sufficient to achieve 0.1% relative standard deviation. ZCOR ~45% TAP crystals (Sp 1,2,4)

30. Conclusions 1. The effort to develop and share instrument configuration and performance specifications will produce better educated instrument purchasers who are paying for exactly what they really need. 2. This effort will help to inform instrument manufacturers what the users of these instrument value and care about in terms of instrument configuration and performance. Quantitatively! 3. The process of testing the instrument performance based on specific quantitative criteria will help assure the delivery of instruments that can actually meet these specifications. 4. This process will result in well-characterized instruments with measured limits to precision and accuracy (see P. Carpenter’s talk next on characterization of quantitative instrument parameters).