1. 1 Summer 2001: Arrival of new accelerator, courtesy of CFI and other partners Its main job is to provide a 3 MeV proton beam for PIXE activities

2. 2 Thanks to: B. Teesdale, J.Maxwell, Z,Nejedly, T.Hopman, T.Papp, B. Morton, T.Riddolls, C. Gielen, D. Urbshas (accelerator installation) R. Protz (LRS, deceased): H. Jamieson (Queens) G. Grime (Oxford): A.Denker (Berlin) L.Cabri, B.Kjaarsgard (Nat. Resources Canada) N.Halden (U.Manitoba) J.Babaluk, J.Reist, A.Kristofferson (DFO/Arctic) R. Eldred, G.Czamanske (California) R.Hoff, J.Brook, R.Vet, M.Shepherd (Environment Canada)

3. 3 Origins of PIXE and micro-PIXE 1970: Johansson et al at Lund: urban air particulate collected on carbon foil: can “analyze” sub-mg samples at ppm limits using proton-induced X-rays. Ionization cross-sections poorly known: “PIXE analysis” needs standards or better physics 1972: Cookson (UK) – first micron-diameter proton beam permits micro-PIXE 1989: GUPIX program launched from Guelph

4. 4 Our timeline 1970s Experiments on atomic inner-shell processes Dabbling in PIXE as a curiosity with potential 1980s Develop PIXE as a quantitative technique Refine the atomic physics database Start GUPIX Start a proton microprobe 1990s Consolidate micro-PIXE and identify new applications GUPIX widely used – 90 groups – feedback identifies physics needs – extensive development 2000 Create national facility for environmental analysis

5. 5 Electron microprobe Proton microprobe Synchrotron radiation microprobe

6. 6 Niches PIXE can handle small samples (100 μg) – e.g. PM2.5 air particulate (component of smog – health issue) Can focus to micron-size spot – analyze mineral grains in situ Can bring beam into the air – handle large or delicate objects – archaeometry and art, manuscripts (eg Galileo, Vinland map) ppm detection limits in point analysis Trace element imaging in mineral grains, fly ash particles, otoliths, etc

7. 7 Recent contributions to mineralogy of the Noril’sk sulfide deposit in Siberia: looking at PGE-rich ores – distribution of Pd, Rh and Ru among pyrrhotite, pentlandite etc. Objectives of such work: Economics and technology of recovery and processing: Test genetic models involving differentiation of mantle sulfide-silicate melts Small-scale niche

8. 8 Egyptian artefact undergoing 68 MeV PIXE analysis at Institut Hahn-Meitner in Berlin Large- scale niche

9. 9 Guelph proton probe

10. 10 X300 optical microscope Trace element Si(Li) looks thru absorber foil that suppresses bremsstrahlung Thin-window detector for 1-6 keV X-rays of light, major elements Focussing quadrupole doublet

11. 11 GUPIX : The Guelph PIXE Software Package Provides manipulation of the database Fits the spectra in single and batch modes ( a few seconds per spectrum) Converts peak intensities to concentrations by the Guelph H-value method Provides concentrations, two error estimates, detection limits, Y/N/? outcomes, depth probed, etc: detailed individual output files plus four project spreadsheets

12. 12 ION-SOLID PHYSICS: 1 Characteristic X-ray yield is Y(Z) = Y 1 (Z) Q  C Z trans Z eff Z Y 1 (Z) is COMPUTED yield per unit charge, unit solid angle, unit concentration – needs a lot of atomic physics information Q is charge or charge equivalent C Z is concentration  is detector solid angle

13. 13 ION-SOLID PHYSICS: 2 E f Y 1 (Z) =   Z (E) T Z (E) / S M (E)dE E 0 where E T Z (E) = exp [ -(  /  ) G(  )  dE / S M (E) ] E 0 Integration is over proton range of a few tens of microns

14. 14 ION-SOLID PHYSICS: 3 In addition there is secondary fluorescence; And in a multiple-layer target, X-rays from 1 layer may fluoresce elements in a different layer: need tabulation of photo-electric cross- sections

15. 15 SPECTRUM FITTING: 1 Non-linear least squares approach Model spectrum takes library of relative X-ray line intensities and modifies them for : - relative absorption in target -relative detector efficiency -transmission through absorbers inserted to reduce bremsstrahlung and low energy X- rays (electron probes use no absorbers – work entirely at low energy)

16. 16 SPECTRUM FITTING 2 Fitting can not be divorced from major element concentrations – which may not be known! Fitting depends upon detector and absorber description – demands a characterization exercise that has no counterpart in RBS or ERD: PIXE is more demanding in this regard The following are important issues but are ignored today -continuum background -Detector resolution function ie peak shape

17. 17 Example of trace element fit:

18. 18 CONCENTRATIONS? STANDARDIZATION BY THE GUELPH H-VALUE METHOD: Y(Z) = Y 1 (Z) Q H C Z trans Z eff Z H is effectively the solid angle H should be a constant PROVIDED  (a) the database is accurate;  (b) the detector is well-characterized  (c ) beam charge is properly integrated Look at single-element H values for 3 MeV protons

19. 19 H-values measured for K X-rays (squares) and L X-rays (triangles)

20. 20 STANDARDIZATION BY H-VALUE: 2 Now we need a quick method to determine H daily Y(Z) = Y 1 (Z) Q H C Z trans Z eff Z This is a compromise between: -a fundamental parameters approach (all physics, no standards – analytical chemists do not like this); and -total reliance on standards (necessitates far too much chemistry) - H must be determined with standard reference materials or SRMs, but these need not be of identical matrix to the specimens

21. 21 Our H-determination Method with NIST alloys Use a homogeneous reference material with a high energy X-ray and a low-energy X-ray Example Mo and Fe in a NIST Mo-steel standard Mo Kα provides the H value, only slightly affected by absorber thickness With that H value fixed, Fe Kα determines the absorber thickness (typically 100-500 μm) Iterate back to the H value using Mo

22. 22 PIXE spectra of NIST alloy standards

23. 23 H-values from alloys

24. 24 Quality assurance/Quality control Quality assurance (QA): Run SRMs and participate in intercomparisons * BCR air filters: 2 nd most accurate of 14 labs in IAEA inter-comparison * NIST glasses *Geochemical standards eg BHVO1 and 30 others Quality Control: Run micromatter films or NIST steels every day prior to analysis

25. 25 Ni in geochemical SRMs Relevant to Ni-in-garnet geothermometer used to identify potentially diamondiferous kimberlites

26. 26 PIXE data for air quality across US-Canada border: Guelph versus Univ. California at Davis

27. 27 What about the ionization cross-sections? GUPIX uses ECPSSR of Brandt with DHS wave functions (Chen and Crasemann) We also have “reference cross-sections” created from the entire literature data by Paul (K) and by Orlic (L), using statistical selection processes Measure H using each of these sets Is the basic atomic physics well enough known?

28. 28 H-values for K X-rays (squares) and L X-rays (triangles) ECPSSR theory OK for K shell but has problems in L shells AND What is wrong with the reference cross- sections?

29. 29 Thorium L X-ray spectra: Voigtians versus Gaussians (with tailing) Papp and Campbell 1996 Nucl. Instr. Meth. B114, 225

30. 30 L2 and L3 natural level widths Campbell and Papp: 2001. ADNDT 77,1

31. 31 Natural width of M3 level Campbell and Papp: 2001. ADNDT 77,1

32. 32 The independent-particle model (IPM) heavily over-estimates atomic level widths in situations (L1, M1, M2, M3, N1 – N6) where 2-vacancy final states are involved: therefore X-ray widths are poorly known – causes fitting errors – serious for L1 As a result, Coster-Kronig probabilities for vacancy rearrangement are badly predicted – induces error in working back from measured L spectra to L sub-shell cross-sections May explain part of L cross-section problem Also introduces error into fitting PIXE spectra

33. 33 Critically selected best values of vacancy transfer probability f13 c/f with IPM theory: Campbell, to appear in ADNDT 2003

34. 34 Earth, air, fire and water: Some applications of micro-PIXE at Guelph Mineral grains: precious metals; impact of mine tailings on environment Mineral grain boundaries: percolation (fire) Imaging in contaminated soil (earth) Metals dispersal in and settling from smelter and power plant plumes (TSRI) (air) Fish otoliths: anadromy; stock identification; fisheries management (water!)

35. 35 Mineral grains in a sulfide rock: courtesy of L.J. Cabri

36. 36 Halden et al: 1993. Can. Mineralogist 33, 961. Linescans to study a lozenge-shaped titanite grain in a radiation-damaged rock

37. 37 LS1 shows Mn, Zn, Y, Pb, U on edges of the grain

38. 38 LS2 shows Mn, Fe, Zn, Sr, Y, Nd, Pb, Th, U in cracks in surrounding material

39. 39 Tomlin et al: 1993. Geoderma 57, 89. Artistic portrayal of a soil block

40. 40 View down vertical worm tunnel in methacryalate- fixed soil: Note fecal material on burrow edge

41. 41 PIXE spectra far from burrow (above) and near the burrow (below): note heavy deposit of metals at sides of burrow: they are carried in fecal material deposited by the worm

42. 42 2-D images of soil section

43. 43 DFO collaborators collect fish in Canadian Arctic (courtesy of DFO)

44. 44 Lake Hazen on Ellesmere Island

45. 45 Arctic char

46. 46 Char otolith (residual earbone)

47. 47 1-inch probe mount with 7 otoliths aligned

48. 48 Note the intense strontium peaks in PIXE spectra of otoliths

49. 49

50. 50 Linescans of known non-anadromous arctic char from Lake 104 and Capron Lake This slide and following three: Babaluk et al: 1997. Arctic 50, 224

51. 51 Known anadromous charr from Halovik River, Jayco River, Paliryuak River, Ekalluk River

52. 52 Conclude: Small morphotype from Lake Hazen is anadromous

53. 53 Surprise: large morphotype from Lake Hazen is NOT anadromous

54. 54 STOCK IDENTIFICATION ISSUE In a mixed summer fishery at sea, can we identify stock or sub-species being exploited?

55. 55

56. 56 ``

57. 57 Otolith of a “young-of- year” Dolly Varden char under oil- immersion objective: PIXE- imaged at OXFORD and at GUELPH

58. 58 Guelph Linescan results Central Sr peak is maternally-derived: outer Sr signal reflects environment

59. 59 FishSr ppm LeftRight Vittrewka 43989445 ± 15518 ± 53 Vittrewka 43984372 ± 39425 ± 26 Cache BF5681 ± 162650 ± 64 Cache BF61004 ± 981001 ± 66 Mean Sr from point analyses in outer lobes

60. 60 Sr from 50 micron rasters of outer lobes FishSr ppm L DorsalL ventralR DorsalR VentralMean V9423 ±7406 ±7444 ±7428 ±7425 ±13 V4404 ±7390 ±7423 ±7405 ±7406 ±12 C BF5652 ±9607 ±8610 ±8579 ±8612 ±26 C BF6965 ±11997 ±11998 ±111041 ±171003 ±27 Technique now shows promise for stock identification

61. 61 Last remarks PIXE and micro-PIXE are well-established methods for trace analysis and imaging, but exist at only a few labs in any country They occupy special niches where they are accurate and versatile Environmental science is well served We get to meet interesting collaborators! Experimental and theoretical work is needed to refine the atomic physics database because neither the IPM nor the experimental data are quite good enough