Spectrum-based Phase Mapping of Apatite and Zoned Monazite Grains Using PCA. Stephen Seddio, PhD Application Scientist

Transcription

1 Spectrum-based Phase Mapping of Apatite and Zoned Monazite Grains Using PCA Stephen Seddio, PhD Application Scientist

2 Phosphate Mineralogy 101 Apatite (Ca 5 [PO 4 ] 3 [F,Cl,OH]) Mineral identified in terrestrial, lunar, martian, and asteroidal rocks (also in bones) Can contain trace concentrations of REE, Th, and U 2 Proprietary & Confidential

3 Phosphate Mineralogy 101 Apatite (Ca 5 [PO 4 ] 3 [F,Cl,OH]) Mineral identified in terrestrial, lunar, martian, and asteroidal rocks (also bones) Can contain trace concentrations of REE, Th, and U Monazite ([La,Ce,Pr,Nd,Sm,Th]PO 4 ) Rare mineral in terrestrial and lunar rocks Monoclinic counterpart of the tetragonal xenotime ([HREE,Th,U)(P,Si)O 4 LREE ore Magnets Hybrid car batteries Th, U, and radiogenic Pb concentrations can be high enough for geochronology by EPMA 3 Proprietary & Confidential

4 Introduction Considerations associated with EDS and WDS X-ray mapping Minimizing the interaction volume of the electron probe Collecting a statistically meaningful number of counts above background Selecting the appropriate elements for mapping Avoiding sample damage 4 Proprietary & Confidential

5 Introduction Considerations associated with EDS and WDS X-ray mapping Minimizing the interaction volume of the electron probe Collecting a statistically meaningful number of counts above background Selecting the appropriate elements for mapping Avoiding sample damage Confusing results may be unavoidable Multiple phases with similar compositions Elements that produce interfering X-ray lines In same phase In different phases in the mapped area 5 Proprietary & Confidential

6 Introduction Considerations associated with EDS and WDS X-ray mapping Minimizing the interaction volume of the electron probe Collecting a statistically meaningful number of counts above background Selecting the appropriate elements for mapping Avoiding sample damage Confusing results may be unavoidable Multiple phases with similar compositions Elements that produce interfering X-ray lines In same phase In different phases in the mapped area Contrast enhancements and image filtering can help but do not eliminate the confusion 6 Proprietary & Confidential

7 7 Proprietary & Confidential

8 8 Proprietary & Confidential

9 What about spectrum-based phase mapping? 2.0 Unknown Liquid 1.5 Int Gamma-Butyrolactone, 99+% Library Match: 96.34% 0.6 Int Raman shift (cm -1 ) 9 Proprietary & Confidential

10 What about spectrum-based phase mapping? Software compares a spectrum at each pixel with a spectral library Uses a χ 2 test to compare spectra Requires a thorough and correct spectral library 10 Proprietary & Confidential

11 What about spectrum-based phase mapping? Does 217 counts = 935,000 counts??? 11 Proprietary & Confidential

12 What about spectrum-based phase mapping? Apatite (Ap) Spectra are compared to a 15 kv monazite spectrum High Th Monazite (HTM) Monazite (Mon) 12 Proprietary & Confidential

13 What about spectrum-based phase mapping? Spectroscopy is done here Apatite (Ap) Mapping is done here Spectra are compared to a 15 kv monazite spectrum High Th Monazite (HTM) Monazite (Mon) 13 Proprietary & Confidential

14 What about low kv spectrum-based phase mapping? There is less spectrum for comparison 14 Proprietary & Confidential

15 What about low kv spectrum-based phase mapping? Ap Spectra are compared to a 5 kv monazite spectrum HTM Mon 15 Proprietary & Confidential

16 What about low kv spectrum-based phase mapping? Spectroscopy is done here Ap Mapping is done here Spectra are compared to a 5 kv monazite spectrum Mon HTM 16 Proprietary & Confidential

17 What about spectrum-based phase mapping? Thought Experiment: How many counts to distinguish Apatite and HTM from Monazite? At 15 kv? Compared to itself, χ 2 Mon = ~2.5 for all counts To be distinguishable from monazite, χ 2 Phase > χ 2 Mon χ 2 Ap Counts Ap χ 2 HTM Counts HTM For χ 2 = 2.5, Counts Ap 700; Counts HTM Proprietary & Confidential

18 What about SBPM? Thought Experiment: How many counts to distinguish Apatite and HTM from Monazite? At 5 kv? Compared to itself, χ 2 Mon = ~1 for all counts To be distinguishable from monazite, χ 2 Phase χ 2 Mon χ 2 Ap Counts Ap χ 2 HTM Counts HTM For χ 2 = 1, Counts Ap 1800; Counts HTM Proprietary & Confidential

19 What about spectrum-based phase mapping? Thought Experiment: Generate a px map 100,000 counts per second How long to be able to distinguish At 15 kv? Apatite is distinguished from monazite after 1.5 hours HTM is distinguished from monazite after 2.0 hours At 5 kv? Apatite is distinguished from monazite after 4.0 hours HTM is distinguished from monazite after 5.5 hours 19 Proprietary & Confidential

20 A Better Way Elemental mapping can yield messy results Spectral matching is SLOW. Is there a better way????? What does it take to distinguish phases at 25 counts per pixel? Principal component analysis to the rescue! Let s look at an example 20 Proprietary & Confidential

21 Ti-Al-V 15 kvalloy SE Image 21 Proprietary & Confidential BSE Transmission of BSEs through overlying phases. BSE Image Warrants low-kv mapping 10 µm

22 22 Proprietary & Confidential From Seddio et al., 2011

23 15 kv BSE 5 kv 10 µm 23 Proprietary & Confidential

24 15 kv Both mapping runs were acquired for 2.5 hr 1575 counts per px 232 counts per px 5 kv 10 µm 24 Proprietary & Confidential

25 Principal Component Analysis 10 µm 15 kv 1575 counts per px 25 Proprietary & Confidential

26 Principal Component Analysis 10 µm 5 kv 232 counts per px 26 Proprietary & Confidential

27 Principal Component Analysis 10 µm 12 COUNTS PER PIXEL!!! 5 kv 232 counts per px 12 COUNTS PER PIXEL!!! 27 Proprietary & Confidential

28 Principal Component Analysis 10 µm 15 kv 5 kv 28 Proprietary & Confidential

29 Principal Component Analysis 15 kv 5 kv 10 µm 15 kv 5 kv 29 Proprietary & Confidential

30 Conclusions Principal component analysis is required for understanding and interpreting otherwise confusing X-ray maps It provides identification of phases in a fraction of the time Principal component analysis is an accepted technique in ever other industry Delivering packages Scheduling airlines Credit card campaigns Telemarketing Remote sensing 10 µm 15 kv 30 Proprietary & Confidential