© SAS, ICXOM XVIII

Alexandre SimionoviciLab. de Sciences de la Terre, ENS - Lyon

© SAS, ICXOM XVIII

Lab. de Sciences de la Terre, École Normale Supérieure - Lyon

L. LemellePh. Gillet

P. Soudant

ESRF, GrenobleP. Bleuet

Univ. of SassariB. Golosio

IMT,Russian Acad. of Sciences

M. ChukalinaLaszlo Vincze, Univ. of GhentCh. Rau, Univ. of ChicagoAntonio Brunetti, Univ. of Sassari

© SAS, ICXOM XVIII

Uncertainties for Kα et Lα

5 − 25 %

Fluorescence cross-sections σ

M.O. Krause et al., ORNL-5399

© SAS, ICXOM XVIII

Brunetti, Sanchez del Rio, Golosio, Simionovici, SomogyiSpect. Acta B 59, 1725-1731,(2004)

X-ray database: XRAYLIB

multi-OS/multi-languagefluorescence/absorptionCompton/Rayleigh (pol/non-pol)form-factor, scatt. functionTransition/Edge energies

Z=13

Z=80

http://www.esrf.fr

© SAS, ICXOM XVIII

Accurate quantification by Monte-CarloL. Vincze, Univ. of Ghent

250 µm thick Si-waferLive time: 1000 sInstrument: ID18FE0 = 27 keVEphotoelectron = 25.16 keV

K photoelectron bremsstrahlungL photoelectron bremsstrahlung

fluorescence

photoelectron impact ionization

Si

Ar

full simulation

experiment

Single + multiple Compton

Simulation without photoelectron tracing

Collaboration: J. Osan, Sz. Török,, KFKI, Budapest, Hungary

© SAS, ICXOM XVIII

I0 Abs. Det.

Totalfluo.

Thick target mapping - IFast

counterE dispersive det.Slow (>0.5 s/pt)

Fluodet.

© SAS, ICXOM XVIII

Fluodet.

Δx

Δy

Thick target mapping - II

loss of resolution ~ Δyaverage of concentrationsNO self-absorption corrections

Limited imaging

Qualitative

But FAST

© SAS, ICXOM XVIII

GEOMETRY problem

SELF-ABSORPTION problem

LOW Z (≤ 14) fluorescence

Problems in X-ray mapping

DETFocused

beam

penetration

multi-element

Z > 14

+ -

© SAS, ICXOM XVIII

e--beam mapping

“skin” mapping

Energy distribution

Range distributions

Quartz Steel

resolution ~ μmlocal concentrations > 1‰

E = 20 keV

© SAS, ICXOM XVIII

Spectroscopy- fluorescence- absorptionImaging- abs. tomographyDiffraction- WAXS/SAXS

Beamline for X-ray microanalysis & microscopyCOUPLED ANALYSES- Fluorescence + Diffraction- Tomography + Fluorescence

K-B

microscope

CCDCCDDiff

Si(Li)

detectorYZ Rθ

2 µm/step

Sample holder0.1 µm/step or

5 nm/stepXYZ Rθ Rφ θ

YZ1 µm/step

< ppm

≤ µm

Talk by Isabelle Letard, ESRF: 13 element Si(Li)

© SAS, ICXOM XVIII

Solution :fluorescence tomography

polar scan z et θlong acquisitionreconstruction : inverse pb.

Simionovici et al., IEEE Trans. Nucl. Sci., 47, 2736-2740 (2000)SPIE 3772, 304 (1999)SPIE 5535, 232 (2004)

© SAS, ICXOM XVIII

Problem: unmeasured fluorescences (Z<14)- self-absorption- matrix density

Solution:

I T T: Integrated Tomographic Techniques

SART (FT, Compton, transmission)

Optimal estimate functions ρ, μ

First fully quantitative reconstruction method

© SAS, ICXOM XVIII

Transmission tomographyART (Algebraic Reconstruction Technique)

f1 f2

fn+1

fn

f2n

fn×n

gi

j

gij+1

jth rayof width τ

Area of the intersectionijka

τ=

Image : square grid of n × n = K pixels

fk = absorption coefficient at the pixel k

Ray integrals : gij = g(θi , sj ) θi (i=1,…, Z) , sj (j=1,…,J)

∑=

≈K

kk

ijk

ij fag

1

)()(

τ

© SAS, ICXOM XVIII

ART (Algebraic Reconstruction Technique)

• System of I× J linear equations with the unknowns fk• Initial guess of the image fk (normally set to zero)• Estimate for the ray integral:

( )(

1

)K

ijk k

k

ij a fq

=

≈ ∑

• This must be compared to the measured ray integral gij

• The content of all pixels intersected by the ray is updated in order to obtain the correct value of the integral :

( ))

( )

((

1

) iji

k jk Ki

jk

ij

k

qa

a

gf

′′=

−Δ =

∑

• The difference between gij and qi

j is distributed among all pixels intersected by the ray in proportion to their weights

© SAS, ICXOM XVIII

Fluorescence Tomography Fluorescence signal on the detector at 90°• Only the points intersected by the beam contribute• Contribution of a small path du :

• εD is the efficiency of the detector• f(s,u) is the probability that a photon from the

source reaches the point U

• pα(s,u) is the probability of fluorescent emission by an atom of type α along the path du

• gα (s,u) is the probability that a photon emitted from U reaches the detector

( ) ( ) ( )duusguspusfI D ,,,0 ααε

( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛′′−= ∫

∞−

udusEusfu

,,exp),( 0μ

( ) duNduusp ααα σ=,

( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛−Ω= ∫∫

Ω

dllEdusgDet

usD ),(

,exp41),( αα μπ

( ) ( ) ( ) ( )∫+∞

∞−

= duusguspusfIsS D θθθεθ ααα ,,,,,,, 0Total signal measured by the detector :

Translation-rotation systemEnergy sensitive detector at 90°Counter/detector in the forward direction

Y

X

s translation

u

Fluorescencedetector

Sample

Beam

Transmissiondetector

rotationθΩDl

© SAS, ICXOM XVIII

Compton Tomography

• Analogous to fluorescence tomography, • Based on the signal due to Compton scattered photons

• Non-isotropic differential cross section

Momentum transfer :

• For q ≥ 20 keV, the binding of electrons to the atoms is neglected (free-electrons approximation) ⇒ σC ∝ Z

200

)cos1(1 cmEEEe

=−+

=′ αθα

⇒ provides a map of the electronic density spatial distribution

• For q < 20 keV, the atomic wavefunctions of the electrons must be considereσC ∝ S(q,Z) incoherent scattering function

• In the same conditions, Rayleigh scattering is also important: σR ∝ Z2

0 sin2

q E θ=

© SAS, ICXOM XVIII

Golosio, Simionovici et al,J. App. Phys. 94, 145-156, 2003

Compton/(Rayleigh/Transmission)μ′ and ρ optimal functions

0 0( ), ( , ) ( '),, 0; 2

, ; /min max

, ( )

ScatE p E Emeasurements

bounded

optimal rms

μ μ ρρ μ

ρ μ

ρ μ

Θ> ≥

∃

→

↓

↓

+

E0=25 keVΘ = 90°, φ =90°

6 ≥ Z ≥ 15

Global valuesfor Z elem.unavailable

ε ≈ 12 % ε ≈ 10 %

ε ≈ 5 %

μ′ μ′

ρ

© SAS, ICXOM XVIII

Projections: 180° for FBT, 360° for ART

FBT: no corrections for reflection geometry, no (self) absorptionfaster, adapted to homogenous samples but « artifact-prone »

ART: robust, easy to add corrections, better resolution

Reconstruction: ART versus FBTART

FBT

Fe Ni Cd Sr

resolutio

n = 5 µm

Simionovici et al, SPIE 3772, 304 , 1999

© SAS, ICXOM XVIII

Mycorrhizal root of tomato plantMycorrhizal root of tomato plantroot root -- Ø < 0.5 mm; resolutionØ < 0.5 mm; resolution ≈≈ 1 µm1 µm

K Ca Fe

Cu Zn Pb TransmissionSimionovici et al, Nucl. Instr.& Meth. A 467-468, 889-892, 2001

Schroer et al, SPIE 4503, 230-239, 2001

Environmental study of metal extraction from polluted soils

© SAS, ICXOM XVIII

Search for traces of life on micro-meteorites

Non destructive imaging of carbonatesites of formation of bacteriomorphsComplementary to IR, SEM/TEM studiesPreparation for MARS RETURN- mini-container (BSL 4)

Simionovici et al,SPIE 4503, 222-229, 2001

Golosio et al., Jrnl. App. Phys. 94, 145-157, 2003

Lemelle et al, Am.Mineral. 87,547-553, 2004

Simionovici, Schroer, LengelerAdvances in X-Ray Spectrometry,John Wiley & Sons, 2004

-30

-20

-10

0

10

-30 -20 -10 0 10

Fe

CaSi Cr

Fe

Si2 μ resolution

2 sec/pt

-30

-20

-10

0

10

-30 -20 -10 0 10

Tatahouine

Distance (μm) Distance (μm)

© SAS, ICXOM XVIII

Reference sample :6 thin metal wires in a quartz capillary

Quartz capillary• 100 μm diameter at the bottom• 10 μm wall thickness

Quartz capillaryGlue2 Aluminum wires d = 20 μm1 Copper wire d = 20 μm2 Nickel wires d = 25 μm1 Tungsten wire d = 10 μm

Sample geometry and composition

© SAS, ICXOM XVIII

Metal wires inside a capillary:Transmission and Compton tomography

• Translation : 160 steps of 3 μm each• Rotation : 180 angular steps of 2° each

Transmission tomography Compton tomography

Transmission tomography• Only the higher Z metal wires are clearly distinguishable in the image

( Absorption cross section ~ Z4 )

Compton tomography• The lower Z objects (Al wires, glue, capillary) are clearly distinguishable

( Compton cross section ~ Z )

© SAS, ICXOM XVIII

Combined experiments:FT - AT

tomo (1 μm) fluo-tomo (2 μm)

Ca/Fe ratio

A

B

A B

Radiography

Martian meteoriteNWA 817

© SAS, ICXOM XVIII

Fluorescence tomography reconstructions

Iron Manganese

• Fe density 1.45 g/cm3

• Mn density 0.058 g/cm3

• Total density 3.89 g/cm3

• Fe weight fraction 37.2%• Mn weight fraction 1.50%

Compatible with EPMA analysis

Golosio, Simionovici et al,J. App. Phys. 94, 145-156, 2003

Transmission ~ Z4

Compton~ Z

Rayleigh~ Z2

Results

© SAS, ICXOM XVIII

diatom: voxel: 2x2x2 µm3 , Vol: 100x 100 x 80 µm3

FI (full scan) interpolation, 12 µm pitch, 120 rot.dose/ time, same or better resolutionvolume segmentation, visualization, renderingvolume processing: extraction, measurement

density Br

Golosio, Somogyi, Simionovici, Bleuet, Lemelle,App. Phys. Lett, 84, 2199-2201, 2004

3D elemental imaging byspiral fluorescence tomography

200 μm

© SAS, ICXOM XVIII

Compton CdCa Fe

Camerani Pinzani et al., Anal. Chem. 76, 1586-1595, 2004

Flyash chemistry:C. Camerani et al. , Chalmers Univ., Gøteborg

Density

Mn ––––

Rb ––––

Fe ––––

© SAS, ICXOM XVIII

Origin of primitive meteorites:chondrites (Semarkona/Krymka)

oxydized chondrules(FeO, FeS)

reduced chondrules(metals)

Study of processes during chondrule formation:evaporation + chemical fractionation

(trace anal. Cu, Zn)

© SAS, ICXOM XVIII

10-3 g/cm3g/cm3

Fe Ti Cr

Transmission

cm-1

μ

10-3 g/cm3

Mn Cu ZnNi

10-3 g/cm3 10-3 g/cm3 10-3 g/cm3 10-3 g/cm3

Combined FT&AT- Localization of chondrules/CAI by AT ≈ 1 µm- Direct elemental imaging

250 μm : 1 x 2 μm2

FTslice

250 µm

© SAS, ICXOM XVIII

Environmental markers: foraminifera

Transmission

Ca Ni

100μm

100μm

Ca Cu

100μm100μm

Ca Zn

Ca

100μm

© SAS, ICXOM XVIII

Time savings in FT

Regular scan Sine-fit scan- 30 %

Smart scan- 50 %

Smart scan: - robust, alignment-less- threshold: Compton⊗Fluo- pad to fix matrix- great time gains >50%

Zap scan: - continuous scan for transl.- NO overhead- limited dwell range 0.1-1 s- split line into MCA spectra- abs. time gains >0.5 s/point

1.5th generation : 2 sym. detectors- tricky alignment 1st time- 1/2 time

© SAS, ICXOM XVIII

Confocal μ-XRF

crossed-beam voxel (≈ 15-20 μm)semi-quantitative reconstructiondepth profiling/ layering

possibility of XAS measurements

only way for single-side access

possibility of local tomography

straighforward XYZ scanning

lower scattering volume

© SAS, ICXOM XVIII

B. Kanngiesser, W. Malzer, I. ReicheNucl. Inst. & Meth. B 211, (2003) 259–264B. Kanngiesser, W. Malzer, A. Fuentes Rodriguez, I. ReicheSpectrochimica Acta Part B 60 (2005) 41– 47

Confocal μ-XRF I

μ-layering of paint: Hg, Pb (10 μm)

Dating of a Mughal miniature – 18th century3 layers of paint: Pb, Pb+Zn+Sn, Pb

© SAS, ICXOM XVIII

Z. Smit, K. Janssens, K. Proost, I. Langus, NIM B 552, 219–220 (2004) 35– 40.

K. Janssens, K. Proost, G. Falkenberg,Spect. Acta B 59, 533 (2004) 1637– 1645.

Confocal μ-XRF II

Fe and Sr maps at various depthsGranite 100 μm thin slideResolution : 15 - 30 μmMDL 0.1 ppm, < 1 fg for Fe-Zr

© SAS, ICXOM XVIII

L. Vincze, B. Vekemans, F.E. Brenker, G. Falkenberg, K. Rickers, A. Somogyi, M. Kersten, F. Adams, Anal. Chem. 76 (2004) 6786– 6791.

B. Vekemans, L. Vincze, F. Brenker, F. Adams, JAAS 19 (2004) 1302– 1308.

Confocal μ-XRF III

Inclusions (Zr, Sr, Th) in mm-sized diamondsMDL ≈ 50 ppm (Zr)Resolution 15-30 μm (Zr, Ca)

© SAS, ICXOM XVIII

XCFT C-XRF

DET

Dose deposition

det

© SAS, ICXOM XVIII

Quantification requirementsaccuracy of tabulated values (σ, μ)

normalization, stability: position, angle, flux = f (t)

sample heterogeneity/geometry corrections

enhancement (2nd, 3rd)

photo-ionization from scattering (C, MS)

photo-ionization from photo-e- bremsstrahlung

electron impact ionization by Auger & photo-e-

linear polarization correction for incident flux

detector response (tails, internal Compton, efficiency)

energy dependence of the excitation (ΔE≠0)

angular distribution of signal (detection angle)

spectral purity (higher orders contribution)

© SAS, ICXOM XVIII

EXRS 2006European Conference on X-Ray Spectrometry

June 19-23, 2006 - PARIS, France

Chairs: Marie-Christine LEPY, CEA SaclayAlexandre SIMIONOVICI, ENS Lyon

http://www.nucleide.org/exrs2006