XRF at the WSU GeoAnalytical Lab - University of Western Ontario

XRF at the WSU GeoAnalytical Lab:
decades of service with graphite
crucibles and a smile
Rick and Laureen
Richard M Conrey
Laureen C Wagoner*
Diane J Cornelius*
John A Wolff
GeoAnalytical Lab
http://www.sees.wsu.edu/Geolab/index.html
School of Earth and
Environmental Sciences
P.O. Box 642812
Pullman, WA 99164 USA
* graduates of Univ. of Western Ontario XRF course

Mission of the XRF portion of our lab
Since our founding circa 1973 by Peter Hooper we have served thousands of
researchers with X-Ray Fluorescence data
Our client base is chiefly other universities around the USA and the world, our own
students and faculty, and government agencies both local and federal
We strive to provide research quality data at an affordable price and to educate both
our students and visitors in the methods of sample preparation and X-Ray
Fluorescence analysis
A graduate level X-Ray Analysis course has been taught at WSU by Nick Foit for
three decades, with a focus on XRD, XRF and electron microprobe analysis
Diane
Our lab also houses a JEOL 8500F field emission electron microprobe, Siemens D-500 powder X-ray
diffractometer, Agilent 7700 ICP-MS, Thermo-Finnegan Neptune multi-collector ICP-MS, Thermo-Finnegan
Element2 high precision ICP-MS, Finnegan Delta S gas source mass spectrometer, and a Perkin-Elmer
Spectrum GX FTIR
Nick
John at the JEOL console

XRF at WSU - a brief history
� ~1973: XRF lab started by Peter
Hooper and his graduate students
working on the Columbia River basalts
� Peter adopted the low dilution fusion
in graphite method to measure both
major and trace elements based upon
his prior experience in Wales
� ~1976: first XRF: a “biologically
automated” Philips PW1410
� 1980s: radioactive waste project at
Hanford provides steady work
� 1986: automated Rigaku 3370 XRF
� 1984 - 2009: the XRF lab blossoms
under Diane Johnson Cornelius
� 1997: Peter retires, John Wolff
assumes directorship of the GAL
� 2004: Thermo-ARL Advant’XP+
XRF spectrometer installed
� 2010: XRF lab now run by Rick
Conrey and Laureen Wagoner, both
former students of Peter Hooper
7000
6000
5000
4000
3000
2000
1000
0
Steve
Reidel
XRF, ICP-MS samples/year through 2009
XRF
ICP-MS
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Peter
Hooper
Vic
Camp
Marty
Ross

Graphite has many advantages
� Ease of maintenance, fabrication,
and cleaning, inert, inexpensive,
efficient, rare to make unusable
pellet, no releasing agent required
� Allows single recipe with Litetraborate
for nearly all samples -
we analyze nearly all rock types,
minerals and soils with the
exception of ores - little need* to
know beforehand what you are
analyzing
� Allows low dilution strategy - traces
measured on same pellet as majors;
full matrix correction for all
elements
� * Mg-, Fe-, Ca-, P, or Mn-rich
samples can be diluted with pure
fused silica to ensure good glass
formation upon cooling
Life of a graphite crucible
XRF ICP
custom boring tool
~ 60 fusions/crucible
• Graphite crucible cost - $US 8
• machined in-house from ultra-pure rod
• nearly all WSU ICP-MS analyses begin
with a graphite fusion
24 fusions possible
every 45 minutes

Graphite doubly fused beads - glass with some carbon on top
graphite crucible loading
Kimwipe +
compressed air
cleaning
� homogeneity requires re-grind (30 sec/bead) and
re-fusion of graphite static fused beads
� loss of powder is immaterial at re-grind stage
� buffing reduces carbon load before re-grind
powders are weighed into
plastic jars, mixed with a
Vortex mixer, and passed
over an anti-static bar
prior to loading
Graphite fusions
re-grinding

� diamond grinding in water is fast (total 2-3 min/bead)
� final cleaning in sonicator with ethanol
� lapping removes diffusion profile at contact of glass and crucible - final surface finish is 15 microns
29 mm 15 mm 8 mm
Mask Sample wt Minimum wt
29 mm 3.5 gm 1.5 gm
15 mm 1.1 gm 0.5(?) gm
8 mm 0.4 gm 0.2(?) gm
Sizing/lapping of graphite fused beads
machined template assures accurate sizing
We calibrate with smaller masks for beads down to 8 mm diameter
Not simply a reduction in counts across the spectrum - varies with 2theta
because of crystal geometry so many parameters change
Precision varies as square root of counts so degraded by factor ~2 each
step down in size - we count twice as long at 15 mm and 4x at 8 mm
Small sample analyses
1600
1400
1200
1000
800
600
400
200
0
Concentration (ppm)
vs. matrix corrected
intensity (kcps)
Zr calibration at 8 mm
SEE = 4.2 ppm; n =14
r 2 = 0.9999
0 0.2 0.4 0.6 0.8 1

50 ton anvil press
“mini”-chipper
(sub-2 mm)
Breaking and chipping, splitting and grinding
the Cadillac of splitters
It’s important not to
contaminate during
chipping - need a
very hard surface
Our custom-made
(WSU Tech Services)
chipmunks use
hardened tool steel
plates - we’re still
using the original
35+ yr old plates
Rocklabs ring mill
Random sub-sampling of rocks is critical: we mini-chip coarse samples and reduce
splitting error to the level of analytical error with a Rocklabs rotary splitter
Coarse
chipmunk
Retsch planetary ball mill
We normally grind in Ta-free Rocklabs WC (< 120 g) for both XRF and ICP-MS or
in agate (< 20 g) on request or for soils; small ball mills are used for < 3 g samples WC and agate ball mills

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
The nitty gritty of a WSU GeoAnalytical Lab XRF analysis
� collect intensities at 36 peak and 25 background positions (~ 66 min/sample) for all samples
� calculate net peak minus background intensity for each element (often require linear and even
parabolic* matrix-dependent background slope correction)
� correct the net intensities for 135 spectral interferences (pk on pk, pk on bgd, Compton tails)
� calculate raw concentrations from corrected intensities using the calibration curves (set with 75 to 105
highly diverse CRMs)
� calculate matrix corrected concentrations from the raw concentrations by iteration using the
fundamental parameters method (with the absorbance values of the NIST-GSC using Al, Ba, Ca, Cl, Cr,
Fe, K, Mg, Mn, Na, P, S, Si, Sr, Ti, Zr and estimated LOF as matrix elements)
� recalculate the corrected concentrations by iteration using an estimated loss on fusion (LOF; simply 100
minus the analytical sum) to correct for raw powder** analysis and volatile loss of weighed rock powder
upon fusion (correction for not truly 2:1 flux:rock mixture)
** raw powder is much preferred due to potential sample damage and partial analyte losses (chiefly of Pb, K, and Rb) during
ignition loss measurements, especially for halide-rich samples
* parabolic background
correction for Zr
Zr peak minus
background vs.
RhKb Compton
intensity
y = 0.00003424x 2 + 0.00485338x + 0.07247217
R 2 = 0.99536438
mixtures of silica and
puratronic compounds
20 40 60 80 100
estimated loss on
fusion is a robust
approximation of loss
on ignition - we have
employed this
correction for several
years now on diverse
lithologies. Very Ferich
samples (gray
squares) have high
estimated LOF due to
reduction of Fe +3
Estimated LOF (wt%)
55
45
35
25
15
5
-5
(A)
y = 0.9972x + 0.9522
R 2 = 0.9891; n = 2552
-5 5 15 25 35 45 55
Measured LOI (wt%)

Matrix corrected intensity (kcps)
Matrix corrected intensity (kcps)
Major and trace element calibration with graphite double fusions
700
600
500
400
300
200
100
0
350
300
250
200
150
100
50
0
K 2O
SEE = 0.044 wt%
R 2 = 0.9998
n = 102
0 4 8 12 16
Al 2O 3
Concentration (wt %)
SEE = 0.155 wt%
R 2 = 0.9998
n = 104
0 10 20 30 40 50 60 70 80
Concentration (wt %)
Matrix corrected intensity (kcps)
Matrix corrected intensity (kcps)
0.25
0.2
0.15
0.1
0.05
0
2.25
2
1.75
1.5
1.25
1
0.75
0.5
0.25
0
Sc
Calibration requires 5 days at ~66 min/CRM but calibrations hold for 3-10 months
SEE = 0.8 ppm
R 2 = 0.9995
n = 85
0 10 20 30 40 50 60 70 80
V
Concentration (ppm)
SEE = 4.9 ppm
R 2 = 0.9987
n = 84
0 100 200 300 400 500 600
Concentration (ppm)

Reproducibility of major and trace elements using graphite double fusions (n = 850)
Difference from average (wt%)
Difference from average (wt%)
0.015
0.010
0.005
0.000
-0.005
-0.010
-0.015
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
P 2O 5
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Concentration (wt %)
CaO
0 4 8 12 16 20
Concentration (wt %)
Difference from average (ppm)
Difference from average (ppm)
8
6
4
2
0
-2
-4
-6
-8
10
8
6
4
2
0
-2
-4
-6
-8
-10
0 20 40 60 80 100
Rb
Nb
Concentration (ppm)
0 100 200 300 400 500
Concentration (ppm)
Duplicate analyses of 850 diverse same sample powders performed over a 5 year period, 2004-2009

WSU XRF (ppm)
WSU XRF (ppm)
Graphite fused XRF versus graphite fused- acid digested ICP-MS trace elements (n = 1224)
1000
800
600
400
200
200
160
120
80
40
0
0
Ba
y = 0.9827x + 5.763
R 2 = 0.9995
0 200 400 600 800 1000
Ce
WSU ICP-MS (ppm)
y = 0.9908x - 0.6172
R 2 = 0.9985
0 40 80 120 160 200
WSU ICP-MS (ppm)
WSU XRF (ppm)
WSU XRF (ppm)
200
160
120
80
60
40
20
80
40
0
0
Y
y = 0.9715x + 0.6721
R 2 = 0.9988
0 40 80 120 160 200
Th
WSU ICP-MS (ppm)
y = 0.9885x + 0.5574
R 2 = 0.9906
0 20 40 60 80
WSU ICP-MS (ppm)
All 1224 analyses performed over an eight month period in 2009 on a single XRF calibration

Summary
� Our lab has served a broad Earth science research clientele over the
past 35 years using low dilution graphite fusions for XRF analysis
� Graphite provides more than sufficient reproducibility for the vast
majority of bulk rock analyses (our lab is often chosen for certification
studies due to our consistent performance on GeoPT samples)
� Graphite allows employment of a robust low dilution fusion strategy
for nearly all geologic materials save ores
� We thank Charles Wu and the instructors of the University of Western
Ontario XRF course for their excellence in instruction and in preparation
of course materials. The UWO courses have deepened the knowledge
base of our technical staff and helped us further our goals of providing
education and good data at a reasonable cost.