UO EPMA Workshop 2008 Electron-Probe Microanalysis: Instrumental Calibration, Standards, Quantitative Analysis, and Problem Systems Paul Carpenter Earth and Planetary Sciences 1 Brookings Drive Washington University St Louis, MO 63130 [email protected]
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
UO EPMA Workshop 2008
Electron-Probe Microanalysis:Instrumental Calibration, Standards,Quantitative Analysis, andProblem SystemsPaul CarpenterEarth and Planetary Sciences1 Brookings DriveWashington UniversitySt Louis, MO [email protected]
UO EPMA Workshop 2008
The Big Picture for EPMA
• Instrumental issues for EPMA:Column-spectrometer alignmentDetector linearity and stability (flow, sealed)WDS deadtime calibrationSpectrometer resolution, reproducibilityNew developments: SDD EDS mapping and quantitative analysis
• EPMA Standards:Proper selection of standards (sample vs. standard)Internal consistency of stds in your lab vs. international environment
• Problem Systems:Peak overlaps, high-order WDS interferencesAnalytical problems, high absorption correctionCorrection algorithms and mass absorption coefficient data sets
• Solutions:Interlaboratory collaboration, educationMultiple KV and multiple spectrometer analysis of core std setPayoff – proof of internal std comps and empirical macs
UO EPMA Workshop 2008
Washington University, Saint LouisEarth and Planetary Sciences JEOL JXA-8200
UO EPMA Workshop 2008
Calibration Issues for Electron-Probe Microanalysis
• Microprobe performance specifications are:Driven by capabilities and address problem solving for customersCapabilities are funded by purchases, user/vendor developmentRealistic specifications for WDS vs. EDS systems
• Instrument calibration during installation and testingSpectrometer alignment – to electron column and mutual agreementDetector linearity with count rate and deadtime issuesPrecision = reproducibility (mechanical, electronic)Accuracy = correct K-ratio measured
• Instrument calibration – short vs. long termConsistent performance with timeAccuracy in international interlaboratory environment
• Geological EPMACMAS silicate standards used for acceptance testing (CIT, WU)
•Low energy pulses must be discriminated from baseline noise. Need propersetting of noise threshold, baseline, and window settings of WDS pulseheight analyzer.•The pulse processing circuitry of WDS does not need to deal with pulseshaping like that of EDS, and is inherently faster.•Pulse energy shift with varying count rate results in instability. At highcount rates pulses are poorly discriminated from baseline noise. Use similarcount rates on standard and sample.•Avoid tight PHA window, use integral mode unless a PHA interference isobserved.•The P-10 detector gas flow rate must be stable or else gas amplificationfactor varies, and so does count rate.•Temperature variation will affect gas amplification factor as well as thermalexpansion of analyzer crystal.•Low energy peaks need to be integrated due to peak centroid and peakshape/area factors. Use area-peak factor or perform integration.
UO EPMA Workshop 2008
Bias Scan Si Kα 15kv 25 nADetector Bias for 4 volt SCA Peak: 1632
Detector Bias Scan Si KαVary Bias at PHA Narrow Window
Detector bias scan using 3.9 volt baseline and 0.2 volt window on PHA.Intended to minimize energy gain shift of PHA.MSFC Spec 1, P-10 flow counter, TAP, 64x gain, Si Kα on SiO2 metal @ 10k cps.
UO EPMA Workshop 2008
SCA Scan Si Kα 15kv 25 nABias 1632 Volts, SCA Peak at 4 Volts
0
500
1000
1500
2000
2500
0 1 2 3 4 5 6 7 8 9 10
Volts
Coun
ts p
er 1
sec
TAP1, 64x, 1632 Bias
Detector PHA Scan Si Kα
For Si Kα there is good separation between baseline and Si pulses.Nominal baseline is 0.5 V with 9.5 V window (integral mode)MSFC Spec 1, P-10 flow counter, TAP, 64x gain, Si Kα on SiO2 @ 10k cps.
UO EPMA Workshop 2008
Calibration of PHAUsing Bias vs. ln(E) plots
• For JEOL microprobe want SCA pulse at 4 volts, Cameca at 2 volts• Spectrometer at peak position• Bias scan with 3.8v base, 0.2v window gives bias for 4 volt SCA• Plot of bias vs ln of x-ray energy is linear• Calibration performed for minimum element set which spans energy
range of spectrometer for all analyzing crystals• Detector should give same bias for Ti Kα on PET vs. LIF, others• Calibration confirms systematic behavior of x-ray counter• As P-10 tank empties and Ar/CH4 changes, requires recalibration• Use y = mx + b fit to bias data to provide quick calibration• Similar plot for escape peak as function of x-ray energy
UO EPMA Workshop 2008
PHA Bias Plot for LIF/PET DataSi, Ti, Ni Bias data at 8, 16, 32, 64, and 128x Gain
WU 8200 Detector Bias Spectrometer 3Settings for 4 Volt Pulse
LIF 8xLIF 16xLIF 32xLIF 64xLIF 128xPET 32xPET 64xPET 128xSi, PET
Ti, PET and LIFNi, LIF
y = mx + by = bias, x=ln(E)
UO EPMA Workshop 2008
Gain Shift Due to Count Rate
Gain shift due to count rate, detector bias arbitrarily set to 1700 volts.Observed shift is ~ 0.008 volts per 1 K cps (1.95 volt shift over 245 K cps range).At ~125k cps baseline noise discrimination deteriorates.Older PCS electronics exhibit complete shift into baseline noise.MSFC Spec 1, P-10 flow counter, TAP, 32x gain, Si Kα on Si metal.
5K cps3.75 V
60K cps3 V
125K cps2.6 V
250K cps1.8 V
UO EPMA Workshop 2008
PHA Scan Ti Lα 20K CPS
Light element / low energy x-rays are poorly resolved from baseline noise.Gain shifts with count rate – PHA peak shifts toward baseline with increasingcount rate. Use integral mode unless PHA energy discrimination required –counts extend to upper limit of PHA scan.MSFC Spec 1 with P-10 flow counter, LDE2, 128x gain, Ti Lα on Ti metal @20k cps.
UO EPMA Workshop 2008
Carbon Kα PHA Scans Graphite, Fe3C
Scaled PHA Scans C Kα LDE2 10 KeV
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 1 2 3 4 5 6 7 8 9 10Baseline, Volts
Rel
ativ
e In
tens
ity
Graphite 100 nA > 65 kcps
Graphite 25 nA ~ 23 kcps
Fe3C 100 nA ~ 2 kcps
Scaled PHA scans demonstrateconsistent PHA behaviorNo baseline – noise is presentLDE2, 128x gain, 1750V
Baseline 0.5V, 9.5V Window
UO EPMA Workshop 2008
Deadtime Measurement on theWavelength-Dispersive Spectrometer
UO EPMA Workshop 2008
WDS Deadtime Issues in EPMA
• Deadtime – time interval during which counting electronics are unable toprocess subsequent incoming pulses• Deadtime error is non-negligible, systematic, affects all measurements• General problem:• Counting behavior of WDS systems is undocumented and poorly known• End-users make measurements with assumed WDS deadtime behavior• User knowledge of deadtime issues needs improvement• Specific problem areas:
No software to conveniently evaluate deadtime on turnkey systemsNo agreed method for setting bias, gain, and sca on systemsSCA pulse shift behavior with count rate undocumentedDeadtime dependence on X-ray energy undocumented and unknownLow vs. high count rate behavior and deadtimes inconsistent
UO EPMA Workshop 2008
Deadtime Behavior:Extending vs. Nonextending
Nonextending vs Extending Deadtime
0.01
0.10
1.00
0.01 0.10 1.00 10.00Normalized Input Count Rate
Nor
mal
ized
Out
put C
ount
Rat
e
Nonextending
Extending
Nonextending: N = Nm (1 – N τ), Nmax = 1/τExtending: N = Nm e –Nm τ , Nmax = 1/eτ
Nm τ
N τ
UO EPMA Workshop 2008
Deadtime LossesInput – Output Curves for μsec Deadtime Constants
WDS Deadtime Losses
0
50000
100000
150000
200000
0 50000 100000 150000 200000
Input Counts per second
Out
put C
ount
s per
sec.
1.00E-062.00E-064.00E-065.00E-066.00E-06
UO EPMA Workshop 2008
Percentage Deadtime LossesPercent Level Corrections Apply to All Measurements
Deadtime Correction, Percentage
0123456789
10
0 10000 20000 30000 40000 50000Input Counts per second
Dea
dtim
e C
orre
ctio
n, %
6.00E-065.00E-064.00E-062.00E-061.00E-06
~1.25 μsec
At 1 μsec, 1% correction at 10Kcps, 5% at 50KcpsAt 2 μsec, 2% correction at 10Kcps, 10% at 50Kcps
UO EPMA Workshop 2008
Deadtime RelationsCalculation of Deadtime Constant
• N = true count rate, Nm = measured count rate withdeadtime losses (Nm < N), and τ is the deadtime constant,which ranges from 1 to several μsec for WDS countingsystems. It is necessary to know Nm and N to calculate τ.We assume the proportionality of N to the probe current iis constant. This may not be true at low count rates.
• Nm / i = measured count rate in counts per second per nA,and c is the constant N / iForm: y = mx +b (Nm / i) is y, x is Nm, y-intercept b is constant c (= N / i).
• Equivalent to τ = (1 – y / b) / xMeasure x-ray intensity at increasing probe currentUse count rate Nm and Nm / i to evaluate the deadtimeconstant τ over a range of intensity values
=
) N - (1N N
m
m
τ
)= τ N - c(1
iN m
m
Ν] / ) / Ν( −[1
=m
m ci τ
UO EPMA Workshop 2008
Deadtime Evaluation PlotNm vs. Nm / i to determine LS Fit to τ
UO EPMA Workshop 2008
Verification of Probe Current vs. Absorbed Current Linearityand/or Detection of Sample Charging
Identification of Sample Charging:Ratio of Absorbed Current to Probe Current
Conductive Si WaferConductive Si Wafer Run 2Charging Si Wafer
UO EPMA Workshop 2008
Deadtime Calculation from Excel Spreadsheet
nA Abs Cur Abs/Probe Time Cps (x) Cps/nA (y) Fit All Fit Last DT us All DT Last2.00 1.63 0.82 100 4607.9 2302.57 2299.81 0.615.00 4.05 0.81 80 11436.9 2287.20 2286.17 0.83
Regression Output: Mean deadtime 0.86 0.84All Y intercept 2309.01 Slope -0.0020 Sigma 0.02 0.01High CR Y intercept 2304.04 Slope -0.0019 Regression DT 0.87 0.84
Excel Sheet: X is Nm and Y is Nm/i. Deadtime evaluated from each intensity (DT) andfrom least squares fit to data (Fit) using Excel linest function. All data and high intensityonly data are compared with average values (Mean deadtime) and standard deviation.Ratio of absorbed/probe current checks conductivity. If linear all data agree.
UO EPMA Workshop 2008
Deadtime Si Kα TAP Spec 2 Caltech 733
1.001.101.201.301.401.501.601.701.801.902.002.10
0 25000 50000 75000 100000 125000 150000 175000Counts per second
Dea
dtim
e, μ
sec
4/25/91 Original Noran
Deadtime Variation with Count Rate
Caltech 733 Spec 2 with P-10 flow counterDeadtime at low count rates include darkcurrent and are not representative of dynamicrange of spectrometer
UO EPMA Workshop 2008
Deadtime Variation With Time and P-10 Gas ChemistryComparison of Original and New Tracor PCS Electronics
UO EPMA Workshop 2008
Alignment and Quantitative Analysis:Wavelength-Dispersive and Energy-Dispersive Spectrometers
UO EPMA Workshop 2008
Establishing Calibration of an Electron Microprobe
• Wavelength spectrometer aligned vertically (baseplate) to coincide withoptical microscope focal point in z-space• Diffracting crystal aligned to be on Roland circle• All WDS should focus on same z-axis and coincident xy area ~ 50 um indiameter• Characteristics of correct alignment
All WDS & EDS have identical X-ray takeoff angleMaximum X-ray intensity at z focus position, but also require:Measure identical k-ratio within counting statistics
• Simultaneous k-ratio measurement is ultimate test of alignment• Initial CMAS standard set used on Caltech MAC and JEOL JXA-733• Expanded CMASTF standard set used for Wash U JXA-8200
UO EPMA Workshop 2008
Electron Microprobe ColumnSpectrometer Alignment: Baseplate and Crystal
Baseplate: Place Rowland circle at Z focusCrystal: Align all crystals on Rowland circleSpectrometer design keeps detector on RCNote: Different K-ratio = misalignmentMultiple spectrometer comparison requiredto demonstrate all WDS and EDS aremutually aligned
UO EPMA Workshop 2008
CMASTF Silicate StandardsGeological materials are multicomponent
• End-member stoichiometric silicate and oxide mineral standards• Primary standards:
• Analyzed suite of stoichiometric standards, natural and synthetic materials:Second set of primary standards on different mountsSpinel MgAl2O4, Enstatite MgSiO3, Forsterite Mg2SiO4Kyanite Al2SiO5Fayalite Fe2SiO4
• Well characterized natural mineral standards and glasses:Olivines (Mg,Fe)2SiO4
Diopside CaMgSi2O6, Anorthite CaAl2Si2O8, Sphene CaTiSiO5Ilmenite FeTiO3Synthetic glasses in CMAS and CMASF system:
Weill CMAS glasses, NBS K411, K412
UO EPMA Workshop 2008
CMASTF Standard Inventory: Natural & SyntheticComposition in Wt% Oxide
36.9942.9819.021.01Weill J26.0152.952.0119.03Weill I21.9730.9141.905.22Weill H2.8961.123.3132.69Weill G6.9452.0630.9310.07Weill F
59.850.0040.15Weill Enstatite Glass5.0479.978.996.00Weill E*16.0045.0720.9617.97Weill D20.9748.9916.0513.99Weill B23.1549.7216.0711.05Weill A
Washington University JXA-8200 SDDCorning 95IRV: K, Ti, Cr, Fe, Ce, Hf
Washington University JXA-8200 SDDCorning 95-Series Trace Element Glasses @ 15 KeV, 50 nA
1
10
100
1000
10000
100000
1000000
0 1 2 3 4 5 6 7 8 9 10
Energy, KeV
Cou
nts
95IRV, T3, 250s95IRV, T3, 5s
K
O
CrTi,Ce
Ca
Mg, Al, Si
Fe Hf
UO EPMA Workshop 2008
Washington University JXA-8200 SDDCorning 95IRW: V, Mn, Co, Cu, Cs, Ba, La, Th
Washington University JXA-8200 SDDCorning 95-Series Trace Element Glasses @ 15 KeV, 50 nA
1
10
100
1000
10000
100000
1000000
0 1 2 3 4 5 6 7 8 9 10
Energy, KeV
Cou
nts
95IRW, T3, 250s95IRW, T3, 5s
O
Ca
Mg, Al, Si
BaMn Co Cu
VLa
Th
UO EPMA Workshop 2008
Washington University JEOL JXA-8200SDD Quantitative Analysis Data
• SDD great for mapping, what about quantitative analysis?• SDD EDS data acquired at 120s, 60s, and 3s acquisitions at T3• Standards used: MgO, Al2O3, SiO2, CaSiO3 (CaO 48.27, SiO2 51.73), TiO2,and Fe2O3
• Linear least-squares peak deconvolution (JEOL software)• Extracted raw K-ratios processed using Armstrong Φ(ρz) and FFAST macsfor comparison with WDS data
UO EPMA Workshop 2008
CMASTF Standard AnalysesWU8200 SDD LLSQ 120 sec. Acquisition T3
Measured K Relative to Calculated KW U8200 SDD 120 sec Acquisition
0.80
0.90
1.00
1.10
1.20
0 10 20 30 40 50 60 70 80 90 100
Weight Percent Oxide
Km
/ K
c
Mg Km/KcAl Km/KcSi Km/KcCa Km/KcTi Km/KcFe Km/Kc
UO EPMA Workshop 2008
CMASTF Standard AnalysesWU8200 SDD LLSQ 60 sec. Acquisition T3
Measured K Relative to Calculated KWU8200 SDD 60 sec Acquisition
Advances in EPMA:Geological Materials -- Standards
•EPMA standards requirements: Homogeneous on micron scale, grain tograin, well characterized on both scales, and available in large enoughquantity to be used by microanalysis communities.•Most materials fail one or more of these requirements.•Natural and synthetic minerals, oxides, and glasses. Minerals impose stoichiometry but may be inhomogeneous Glasses lack stoichiometric control but can be homogeneous•Glasses: targeted compositions that can be made in bulk and utilized by themicroanalysis community.(Corning 95-series trace element glasses)•Internal consistency of EPMA standards used by the community is poorlyknown. Few comparison reports, generally anecdotal.•Solution: calculate expected x-ray intensity for element of interest in suite ofstandards, compare measured intensities relative to end-member standard(oxide), i.e., k = ZAF / C. This highlights errors in composition as well assystematic errors in algorithm.
UO EPMA Workshop 2008
Basalt Glass Indian Ocean USNM 113716:EPMA vs. Wet Chemistry Data
Of
Of the 3-5 mounts of UNSM 113716, this is the first observation of mineralinclusions or crystallites in the glass. This is otherwise a homogeneous standard,consistent with EPMA of other glasses, but based on wet chemistry comparison.How representative is this of the wet chemical analysis?
DevitrificationPlag
Glass
UO EPMA Workshop 2008
Olivine Standards: Mg-rich (Mg,Fe)2SiO4
Minor/Trace Els.Nat./Syn.Standard
<none>SyntheticLLNL “Fo67” (Fo70)
Ca, Cr, MnNaturalSpringwater olivine Fo82
Al, Ca, Cr, Mn, Co?, Ni?SyntheticLLNL “Fo80” (Fo85)
Al, Ca, Mn, ZnSyntheticFujisawa sintered Fo90
Na?, Mg, Al, Ca, Ti?, Cr, Mn, Co, NiNaturalSan Carlos olivine Fo90
<none>SyntheticLLNL “Fo85” (Fo93)
Mn, Co, Ni, Zn?NaturalBoyd olivine Fo93
Fe?SyntheticShankland forsterite Fo100
Shankland from ORNLLLNL olivines from George Rossman, Boyd and Fujisawa from CaltechSan Carlos and Springwater olivine from Smithsonian
UO EPMA Workshop 2008
Olivine Standards: Mn, Fe, Ni
Minor/Trace Els.Nat./Syn.Standard
Cr?, Fe, CoSyntheticNi-olivine P-877
Al?, Ca?, CrSyntheticFayalite ORNL
Mg, Ca, Cr, Mn, ZnNaturalRockport Fayalite
Mg, Cr, MnSyntheticFayalite RDS P-1086
MnSyntheticFayalite GRR-391
Mg, Ca, FeSyntheticMn-olivine RDS P-1087
FeSyntheticMn-olivine GRR-392
GRR and RDS from George Rossman, P numbers Caltech probe standardsRockport Fayalite from Smithsonian
UO EPMA Workshop 2008
Rockport Fayalite
• RF is widely used as primary Fe standardBut Mg and Zn present, not in wet chemistry analysis[Low level oxides suspected to be variable not reported in wc analysis]
• Is ferric iron present? – apparently not: Wet Chemistry: Fe2O3 1.32, FeO 66.36 %, Tot: 99.18 Dyar XANES: RF iron is completely reduced.• Grunerite in separate: Fe7
2+Si8O22(OH)2
• Magnetite at locality, in separate (Fe23+Fe2+O4) ??
• Analysts should use EPMA analysis when using RF as primary standard.
Averages of total and cation stoichiometry for all olivines from test data set.For olivines, Mg/(Mg+Fe) = 0.860 ± 0.080 (ox) vs. 0.861 ± 0.079 (oliv).Identical k-ratios corrected using PAP full Φ(ρz) and Heinrich 1986 macs,relative to oxide vs. synthetic olivine standards.Olivine Formula: M2+
• EPMA using synthetic olivine standards better than oxide standards:Superior analysis total, Si cation ~1.0, and ΣM2+ ~ 2.0
• Improvement in EPMA accuracy for olivine usingArmstrong Φ(ρz) coupled with FFAST mac data set.
• Using oxide standards we observe: Overcorrection of Mg and Fe in olivine across Fo-Fa binary
Undercorrection of Si in low-Mg olivine (Fayalite, Mn-ol, and Ni-ol)Marginal underestimation of Mg/(Mg+Fe).
• These relationships extend to all MgFe silicates relative to composition.• Alpha-factor analysis of systematic errors in Fo-Fa system: EPMA and wet chemistry of natural olivines are not internally consistent.
Worst: Boyd Forsterite Mg and Fe not consistent (Caltech standard)Best: Springwater Mg,Fe, and Mg in San Carlos (Fe in SC less so)
UO EPMA Workshop 2008
Accuracy of EPMA:Quantitative Analysis of Olivine
UO EPMA Workshop 2008
Olivine EPMA Accuracy Study:Alpha factor method extended to olivine
• Alpha factor (α-factor) method used to evaluate:Systematic errors for Φ(ρz) correction algorithmsInternal consistency of EPMA-only data and EPMA vs. wet chemistry.
• Synthetic olivines, pure Fo and Fa, used as primary standards for α-factoranalysis.• Natural olivines require projection onto Fo-Fa binary for comparison.• Anticipate decreased reliance on correction algorithm and fundamentalparameters as one moves from pure element end members to olivine endmembers. Measurement errors ultimately control accuracy.
UO EPMA Workshop 2008
Olivine Alpha Factor Study:Comparison of Calculated and Measured Concentration•Calculated α-factors using Φ(ρz) algorithms and mass absorptioncoefficients and a polynomial fit to the individually calculated values. C known, K calculated.•Experimental α-factors extracted from olivine analyses. K measured, C calculated. Compared with calculated α-factors. If measurements, algorithms, and data sets are correct, all experimentally determined analyses would lie on theoretical lines.•Wet chemistry data evaluated. K measured, C obtained from wet chemistry.
If wet chemistry data, algorithms and data sets are correct, all wet chemistry analyses would also lie on the theoretical lines.
UO EPMA Workshop 2008
Extraction of α-factors from Experimental Measurements
AAB
AABA
AB
AAB
AAB C
CKC
−
⎥⎦
⎤⎢⎣
⎡−
=1
α
0 CA, weight fraction 1
C / Kα = 1
α2 > 1
α1 < 1α1
α2slope = 1 - α
Abs
Enh
( )
stdsmp
AAB
AAB
AABA
AB
AAB
BPBPKmxby
CKC
)/()(
1
−−=
+=
−+= αα
For olivine:Forsterite: C = Wt Fraction MgO / [MgO in Mg2SiO4]Fayalite C = Wt Fraction FeO / [FeO in Fe2SiO4]
UO EPMA Workshop 2008
Analysis of Mg Kα in Shankland Forsteriteusing MgO Std: Experimental Determination of α-factors
C / K Data Mg Kα in SiO2 - MgOShankland Forsterite
Mg Kα in Olivine Using Forsterite StandardCalculated and Experimental Data, Effect of MAC’s
1.450
1.500
1.550
1.600
1.650
1.700
1.750
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Calc -- Heinrich 1966
Exp -- Heinrich 1966
Calc -- Heinrich 1986
Exp -- Heinrich 1986
Calc -- Henke
Exp -- Henke
Boyd Forsterite WC
San Carlos WC
Springwater WC
Calc -- FFAST
Armstrong φ(ρz) 15 KV 40 TOA
[MgO] / [MgO in Forsterite], wt fraction
RF Fo25 Fo35 Fo40 Fo60
Fo70 Fo85 Fo90 Fo93
SC
BF
SP
Mg
Kα
α-f
acto
r [(C
/ K
- C
)/(1
- C)]
1. Historical overcorrection, esp. H66 at Fo-rich, FFAST macs reduce overcorrection.2. Good agreement EPMA of syn. and natural olivines, esp. H86, Henke using Armstrong Φ(ρz).3. Boyd Forsterite Mg value of wet chemistry inconsistent with wc of San Carlos and Springwater.
UO EPMA Workshop 2008
Fe Kα A-factors in Fo-Fa BinaryBetter accuracy compared to Mg
Armstrong φ(ρz) 15KV 40 degrees
1.060
1.070
1.080
1.090
1.100
1.110
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
[FeO]/[FeO in Fayalite], wt fraction
Fe K
α α
-fac
tor [
C/K
-C] /
(1-C
)
Calc a-factors Henke
Calc a-factors Heinrich1986 and 1966Exp -- Heinrich 1986
Exp -- Henke
Exp -- Heinrich 1986sinteredExp -- GMR/Heinrich1986Boyd Forsterite
San Carlos Olivine
Springwater OlivineFo60 Fo40 Fo35 Fo25
Fo93 Fo90 Fo85 Fo80 Fo70
BFSP
SC
FFAST values intermediatebetween Henke and Heinrich
1. Minimal dependence on mac data set. Could calculate Mg by 2-Fe for binary olivines only.2. Good agreement EPMA of syn. and natural olivines with Armstrong Φ(ρz).3. Continuum fluorescence important for Fo-rich olivines.4. Boyd Forsterite and San Carlos Fe value of wet chemistry least consistent with others.
UO EPMA Workshop 2008
PDR: Philibert-Duncumb-Reed ZAF, oxide stds, Heinrich 1966 macsPAPF-1 and Arm-1: Φ(ρz) algorithms, oxide stds, Heinrich 1986 macsPAPF-2 and Arm-2: synthetic olivine stds, Heinrich 1986 macsPAPF-3 and Arm-3: synthetic olivine stds, FFAST macsSame k-ratios, n=4, CaO 0.09, Cr2O3 0.06, MnO 0.14, NiO 0.37 (wt %)Olivine Formula: M2+
2SiO4 [PFW 7/2004, PDR and PAPF algorithm errors corrected]
EPMA of San Carlos OlivineCorrection Method and macs @ 20 KV, 40 TOA
GMR Thin Film program: calculate emitted intensity at each takeoff angle.(We don’t want k-ratio relative to standard at each takeoff angle)X-ray intensity relative to 40 degree value used to scale weight %Al range 0.88 wt% / 20 degrees = 0.044 wt% per degree = 1.5% relative/deg.Cr range 0.88 wt%, = 0.12% relative/degreeNi range 1.25 wt%, = 0.063 wt% per degree = 0.11% relative/degreeConclusion: 10 degree tilt error results in percent level analytical errors
Nominal
UO EPMA Workshop 2008
Problem Systems
UO EPMA Workshop 2008
Problem Systems in EPMA:Analytical Elements in Problem Matrices
• Analytical Problems:• X-ray peak overlaps• High x-ray absorption• Spatial issues, inhomogeneity• Particle, thin film, etc.• ----• Perform proper measurement to obtain correct k-ratio• Use standard as close to sample for high correction analysis• Necessary to evaluate all correction algorithms and mass absorptioncoefficients – do not blindly accept one pairing as the best
UO EPMA Workshop 2008
WDS Background Selection (Natural Peak Width)And High Order WDS Interferences
• Example AlCrNi alloy used in Lehigh Microscopy School lab• WDS scan on pure Al necessary to establish full natural peak width• People choose backgrounds too close to peak
If background is not true on pure element, also not true on any sample• Cr Kb IV reflection observed at Al Ka peak position on TAP
UO EPMA Workshop 2008
WDS Scan LIF: Al, Cr, Ni and NiCrAl SampleNote background intensity as function of Z
1
10
100
1000
1000014
9
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
Spectrometer position, mm LIF
Cou
nts
Cr stdNiCrAlNi stdAl std
Cr Kα 1,2
UO EPMA Workshop 2008
WDS Scan LIF: Al, Cr, Ni, and NiCrAl SampleNote background intensity as function of Z
1
10
100
1000
1000010
5
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
Spectrometer position, mm LIFH
Cou
nts
Ni stdNiCrAlCr stdAl std
Ni Kα 1,2
Ni Kβ 1,3
UO EPMA Workshop 2008
WDS Scan TAP: Al, Cr, Ni, and NiCrAl SampleNote Cr Kβ IV-order interference on Al Kα, Full Al peak width
Cs Ba La Ce Pr Nd PmSm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg
UO EPMA Workshop 2008
Error Analysis Ta Lα in TaSi2All algorithms and MAC sets (PAPF-FFAST = 1.0)
Error Histogram Ta Lα in TaSi2Relative to PAP-FFAST Nominal K-ratios
0
2
4
6
8
10
12
14
0.95
0
0.95
5
0.96
0
0.96
5
0.97
0
0.97
5
0.98
0
0.98
5
0.99
0
0.99
5
1.00
0
1.00
5
1.01
0
1.01
5
1.02
0
1.02
5
Kcorr / Kexp
Freq
uenc
y Philibert-Duncumb-ReedZAF
Φ(ρz) Models
PAPF
-FFA
ST =
1.0
UO EPMA Workshop 2008
Error Histogram Si Kα in TaSi2Relative to PAP--FFAST Nominal K-ratio
0
1
2
3
4
5
6
7
8
0.79
0
0.86
0
0.93
0
1.00
0
1.07
0
1.14
0
1.21
0
1.28
0
1.35
0
1.42
0
1.49
0
1.56
0
1.63
0
1.70
0
1.77
0
1.84
0
Kexp / Kcorr
Freq
uenc
yError Analysis Si Kα in TaSi2All algorithms and MAC sets (PAPF-FFAST=1.0)
Phili
bert-
Dun
cum
b-R
eed
ZAF
Φ(ρz) ModelsM
cMas
ter
Line
mu
Mac
30
citz
mu
Φ(ρz) Models
Phili
bert-
Dun
cum
b-R
eed
ZAF
PAPF
-FFA
ST =
1.0
Dramatic demonstration of choice of MAC data set:Variation of Φ(ρz) models and 4 mac data sets
UO EPMA Workshop 2008
Calculated Compositions of TaSi2Relative to PAP—FFAST Nominal K-ratios
102.8176.8525.96MM
101.3976.5224.87LM
10076.3123.69(FFAST)
89.6674.0215.64M30
88.4873.7414.74CM
TotalWt% TaWt% SiPAPFwith MAC
UO EPMA Workshop 2008
Conclusions
• Geological applications require multicomponent accuracy evaluation• Use of Kmeas/Kcalc plot used for data analysis, WDS and EDS• CMASTF standards provide instrument calibration data set• Experimental K-ratio data set available for development and testing• Identification of inconsistent compositions• Accuracy of analysis in CMASTF system better than 2%, precision limited• SDD quantitative analysis data highly competitive with WDS• Excellent prospects for high speed SDD quantitative analysis in particle,mapping applications