Wolff, John A. and Conrey, Richard M. GeoAnalytical Laboratory School of Earth and Environmental Sciences Washington State University Pullman, WA 99164.

Wolff, John A. and Conrey, Richard M.

GeoAnalytical LaboratorySchool of Earth and Environmental Sciences

Washington State UniversityPullman, WA 99164 USA

Application of portable X-Ray Fluorescence to problems in volcanology

Portable (handheld) and mobile EDXRF bulk analysis

Bruker handheld Tracer IV

Innov-X 5000 mobile

Recent advances in portability meet need for field measurement (e.g. customs, soil contamination, mine reclamation, scrap yards, mineral exploration etc)

Employ miniature XRF tubes, microamp electronics, SDD detectors

Moxtek 50 kV Au tube

Bruker handheld benchtop setup

-- many people are disappointed with their pXRF on rock outcrops

-- these are NOT tricorders, they are instruments that need consistent consistent sample prepsample prep and and analytical methodsanalytical methods

Portable XRF

Difficulties with portable XRF analysis

• Vacuum only inside instrument, not surrounding the sample

• Sensitivity for lightest elements poor - down to Mg (Z = 12) only

• Matrix (absorption and secondary enhancement) corrections must be approximated if all elements are not analyzed

• Resolution (dispersion) is not as good as WDXRF, so overlaps and interferences may be problems

• No pulse height discrimination so many spurious peaks (sum, escape, diffraction, tube, collimator) can be present in spectra

• Manufactured software good for scrap yards, but not optimized for earth sciences

• Analytical surface often not flat

12%

77%

5%

0.33%

0.09%0.05%

1.2%

128 ppm 202 ppm

High silica rhyolite spectrum with a handheld EDXRF

Sensitivity improves dramatically with Z (due to increasing fluorescent yields of higher energy X-rays combined with their lower absorption coefficients)

ZrRb

Fe

Mn

Ti

CaK

Si

AlRh

RhComp

What do you need for good pXRF analysis?

• Matrix correction for other elements present - even if only approximate

• Calibration and validation of your method

• Development of routines for single applications - one size does not fit all

• A few examples, chiefly volcanologic….

• Consistent sample preparation, especially sample surface but grain size too if you can

Handheld XRF analyses of pumices

Fused bead WDXRF data

-- high sensitivity trace elements critical to discrimination of pumice chemistries

-- samples from the Bandelier Tuff, all high-silica rhyolite

pXRF samples

-- sample prep: mortar and pestle grind to sub-250 micron powder sample prep: mortar and pestle grind to sub-250 micron powder (loose powder in cups)(loose powder in cups)

-- analysis at 45 kV; Compton scatter and approximate matrix corrections employed (no matrix variation)

Average of 10repeat pXRFanalyses 2 std dev wdXRF

TiO2, wt% 0.048 0.004 0.053FeO, wt% 1.31 0.02 1.34MnO, wt% 0.083 0.004 0.079CaO, wt% 0.24 0.01 0.25K2O, wt% 4.44 0.05 4.51Zn, ppm 133 4 134Rb, ppm 367 4 356Y, ppm 110 2 112Zr, ppm 255 9 262Nb, ppm 191 4 186

Uncertainties in pXRF analysis of high-silica rhyolite pumice

Repeatability is very good, as is comparison with WDXRF analyses

Average of 10 repeat pXRF analyses

y = 0.9935x + 0.7284

R2 = 0.99350

50

100

150

200

0 50 100 150 200

Nb

y = 0.996x + 0.9253

R2 = 0.9961

0

100

200

300

400

0 100 200 300 400

Rb

y = 0.9963x + 0.1091

R2 = 0.9961

0

20

40

60

80

100

0 20 40 60 80 100

Sr

pXRF vs WDXRF (all values in ppm)

CRMs

CRMs

CRMs

• Loose powder certified reference material analyses (of silica-rich CRMs) agree well with certified values

• pXRF data can be used to discriminate Bandelier pumices in the field with minimal sample prep and uncertainties for these elements approaching that of fused bead WDXRF

• Loose powder pXRF values are very comparable to same sample fused bead WDXRF values

pXRF analysis of pumices

Handheld XRF analyses of thin section billets

-- analysis at 15, 30, and 45 kV; Compton scatter and approximate matrix corrections employed (wide range of matrix)

-- samples from diverse fresh, fine grained volcanic rocks

-- sample prep: surface lapped flat on coarse diamond lapsample prep: surface lapped flat on coarse diamond lap

-- altered rocks and drill core do not work so well, coarse grained rocks require multiple analyses

y = 0.9316x + 0.2529

R2 = 0.93250

2

4

6

8

10

12

14

16

0 2 4 6 8 10 12 14 16

MgO (15 kV)

y = 0.9695x + 1.7195

R2 = 0.969540

45

50

55

60

65

70

75

40 45 50 55 60 65 70 75

SiO2 (15 kV)

units Wt%

units Wt%

HHXRF vs WDXRF

• Mg, Al, Si, and P all have highest signal to background at 15 kV or lower

• Uncertainties are from 1-2 wt% absolute for Mg, Al, and Si, approximately 0.05 wt% for P

• Calibration is to billets of samples analyzed via fused bead WDXRF (there are no available CRMs for this use)

Billet analyses via pXRF

• Light elements (Z <16) are best excited at low tube voltages

Z < 16

y = 0.9688x + 5.2255

R2 = 0.96880

100

200

300

400

500

0 100 200 300 400 500

V (30 kV)

y = 0.9737x + 0.2117

R2 = 0.9737

0

5

10

15

0 5 10 15

Fe2O3 (30 kV)

HHXRF vs WDXRF

Units ppm

units Wt%

• K, Ca, Ti, and Fe can be usefully analyzed at any voltage, but Cr and Ni appear best at 30 kV

• Uncertainties are improved from light elements, but still no match for WDXRF

Billet analyses via pXRF

• Z = 19-28 elements are best excited at 30 kV (no filters)

Z = 19-28

y = 0.9816x + 4.1672

R2 = 0.9793

0

100

200

300

400

500

600

0 100 200 300 400 500 600

Zr (45 kV)

y = 0.9916x + 4.3004

R2 = 0.99160

300

600

900

1200

0 300 600 900 1200

Sr (45 kV)

units ppm

units ppm

HHXRF vs WDXRF

• 45 kV allows excitation of Ba K lines (Ba L lines have severe interference)

• Uncertainties are improved again but still no match for WDXRF

• Powdered rock in a cup may provide better data, but we have not performed the experiments

Billet analyses via pXRF

• Z = 29-56 elements are best excited at 45 kV (no filters)

Z = 29-56

y = 0.9222x + 0.0565

R2 = 0.92220

0.4

0.8

1.2

1.6

0 0.4 0.8 1.2 1.6

P2O5

y = 0.9842x + 0.0411

R2 = 0.9842

0

1

2

3

4

5

6

0 1 2 3 4 5 6

CaO

y = 1.0021x - 0.3915

R2 = 0.927180

100

120

140

160

180

80 100 120 140 160 180

Rb

Handheld XRF analyses of mudbricks

-- - sample prep: grinding to fine powder in ring mill sample prep: grinding to fine powder in ring mill (loose powder in cups)(loose powder in cups)

-- analysis at 30 kV; Compton scatter and approximate matrix corrections employed (range of matrix)

-- P2O5 data critical to assess presence of cow dung

HHXRF vs WDXRFacknowledgements: Melissa Goodman Elgar and Nichole Bettencourt

units ppm

units Wt %

-- two sample groups from Bolivia provided by WSU Anthropology Dept

Summary and conclusions

• But for now sample prep is critical to good analysis

• Development of good pXRF analytical routines for problems in the Earth Sciences requires some fundamental knowledge of XRF analysis, can’t just rely on manufactured software

• Practical methods for analysis of pumice are easy to develop, routines for analysis of a wide range of lithologies are more challenging

• Thanks for your attention!

• The “tricorder” will be a practical X-ray laser, if it’s ever developed