Basic XRF Workshop

Tim ElamUniversity of Washington - Applied Physics Lab

and Ametek, Inc. Materials Analysis Division(EDAX Research Group)

andGeorge Havrilla

Los Alamos National Laboratory

Denver X-ray Conference August 2, 2010

Workshop Outline

1. What is an XRF instrument?2. How does an XRF spectrum look?

And why does it look like that3. How do we do quantitative analysis?

Specimen effects Instrument effects

4. Applications (George)

Material -> sample -> specimen

Material to measure - solid, powder,liquid, suspended particulates

Sample - chip off rock, scoop,precipitate, pre-concentrate, etc.

Specimen - an individual preparationfor measurement, unit placed ininstrument

Continuum Ag Kα

Ag KβAg L

30 kilovolts tube voltage

What happens when X-rayshit your specimen?

Coherent or Rayleigh scatterWave phenomenon, no energy lossSame process that causes rainbowsAtoms are the right size to scatter X-rays

Incoherent or Compton scatterX-ray hits a single electronLoses energy and changes directionDepends on angle (fixed for most instruments)

Photoelectric absorption (dominant process)Leaves atoms excited (electron shell vacancy)Independent of angle

XRF Mechanism

Atomic vacancy created by:X-rays - XRFProtons - Proton Induced X-ray Emission (PIXE)Electrons - Electron microprobe Picture from AmpTek, Inc.

Used with permission.

A vacancy in theelement must becreated to get XRF

XRF Spectrum

Quantitative overview

Source: James WillisUsed with permission

XRF Explanation - Detailed

Specimen

DetectorSource

(usually an X-ray tube)

Incident Beam Optics(Minimum - aperture) Emergent Beam Optics

(Mono for WDXRF)

Inst

rum

ent

Spe

cim

enIn

tera

ctio

ns

Start with Specimen

X-ray processes in specimen -matrix and scatter

Incident beam is absorbed by all atoms, including analyteEmitted beam is absorbed by all atoms, including analyteX-rays absorbed by Ni atoms emit Ni Kα x-rays,

which can excite additional Fe atomsThis produces both absorption and enhancement matrix effects

Cr Fe Ni

The incident X-rays can also scatter from the atoms. This produces background fromthe continuum primary radiation and peaks from the primary characteristic lines. Raleigh(coherent) scatter is at the same energy while Compton (incoherent ) is shifted.

X-ray processes in specimen -thickness, homogeneity, etc.

Highenergy

Lowenergy

Inhomogeneousspecimen

XRF Explanation - Detailed

Specimen

Detector

Source (usually an X-ray tube)

Incident Beam Optics(Minimum - aperture)

Emergent Beam Optics(Mono for WDXRF)

Instrument - Sources• Isotope and monochromatic SR - less background• Optics - filter, secondary fluorescer, polarizer, or

focusing optic• All optics modify spectrum - this has implications for

analysis, especially sensitivity and background

Data from R. Gorgl, P. Wobrauschek, Ch. Streli, H. Aiginger andM. Benedikt, "Enery-Dispersive Measurement and Comparison of DifferentSpectra from X-ray Tubes", X-ray Spectrometry, 24, 157-162 (1995).

Tungsten X-ray Tube 45 kVKey consideration:primary spectrum energyvselement absorption edges

Instrument - geometry andatmosphere

Detector

2!

Geometry factor (independentof elements or lines; includesarea of specimen, acceptanceangle of detector, etc)

4π steradians

Source solid angleDetector acceptance solid angle

1!

Attenuation of bothbeams in chamberatmosphere

Incident angle Emergence angle

Both angles affect path, especially for thin specimens.

Instrument - detectors

Monochromator for WDXRF, slitsFocusing optics for confocalFilters (used occasionally)

Specimen

Amplifier andPulse shaper

Spectrum

Active layer - Si diode or prop ctrElectrode and/or dead layerBe or plastic window

time

volta

ge

time

volta

ge

Pulse height analyzer

energy

resp

onse

QuantificationSpectrum processing (ED)

Energy

Inte

nsity

1) Subtract background to getnet intensity

2) Sum all counts under peak

Energy

Inte

nsity

Resolve overlaps and interferences•Fixed regions•Fixed energy

Linear least squaresBayesian

•Full least squares

Identify peaks and associate intensitywith analyte elements

Scan setup (WD)

Energy

Inte

nsity

Energy

Inte

nsity

3-point scan setup for peak and background

Fewer interferencesbut harder to handle

Conversion of net intensity toconcentration

Mostly, concentration is proportional to intensity

!

Ci" I

i

!

Ci=Ii

I(i)

Where I(i) is the net intensity from a pure elementspecimen.But there are absorption and enhancement matrix effects,and we may not know the pure element intensity,so

!

Ci= K

i" I

i"M

i" S

i

•Ci concentration of analyte i•Ki proportionality factor•Ii net intensity of analyte i•Mi interelement effect of specimen matrix •Si specimen preparation and heterogeneity

Matrix correction

Ignore S, since we don't know and often can't control homogeneity.

Source: James WillisUsed with permission

This ratio is MFe

In general, M is a function of allconcentrations (or intensities)

Ci = Ki x Ii x Mi x Si

Empirical Matrix Correction

• Empirical methods require many standards• They only cover the composition space of the

standard set• They can be in terms of intensity or

concentration• M is closer to linear function of concentration• Completely dependent on quality of standards

Empirical methods can still provide the best possibleaccuracy (provided the standards are adequate).

Ci = r0 + Ii [ ri + ∑nrin In ]

Theoretical Matrix Correction

• Fundamental parameters method allows calculation of I from C• Can get M from intensity, this allows matrix correction• In principle no standards required - "standardless"• Actually involves some standards, typically measured at factory• If all elements are observed in spectrum, can use sum=100%,

"true standardless"

Ii = G • ∫

S-1S • Ci • pi • ωi • τi(E)

µ(E)Sin ψ1

µ(i)Sin ψ2

+• I(E) • dE

E

Quantification Results

Source: James WillisUsed with permission

Uncertainty and accuracy• Unlike optical spectra, always count limited• X-rays arrive at random, unpredictable times• Sources of uncertainty

– Statistics always lower limit– Calibration and matrix effects– Specimen inhomoqeneity effects

• Accuracy– “Standardless”: limited by atomic data– Calibration– Specimen preparation

• Minimum Detection Limits– Usually calculated from calibration, assuming statistical

fluctuations of background– XRF- background relatively small (vs electron microprobe,

for example)

Summary of XRF capabilities

• Qualitative and quantitative analysis• Solids and liquids• Atomic number ≥ 10 (Na) [special cases ≥ 5]• Energy 0.5 to 25 keV (Wavelength 25 to 0.5 Å)• Sample preparation relatively simple• Concentration range ppm to 100%• Typically about 1% relative accuracy• Time required - a few seconds to several

minutesFrom John GilfrichDon’t tell him.