applications of xafs to materials science and...

SILS School on SR, Duino (Trieste), Italy - September 2007

1

Applications of XAFS tomaterials science and nanostructures

Federico BoscheriniDepartment of Physics

University of Bologna, [email protected]

www.df.unibo.it/fismat/rad-sinc


2

Plan

• Why use XAFS as a structural tool in materials science & nanostructure research

• Examples of applications, using both– results which have “stood the test of time”– recent results

X-Ray Absorption Fine Structure

“XANES”: Absorber symmetry and valence/oxidation state Electronic structure of unoccupied statesMedium range structure

“EXAFS”: Coordination numbersInteratomic distancesDisorder of distances

Ener

gy

1s2s2p ωh

ωh


4

EXAFS• Extended X-ray Absorption Fine Structure• Fit fine structure oscillations with

Debye Waller factor-thermal vibration-static disorder

Coordinationnumber

Interatomicdistance

λσδϕχjr

j eekrSinkANSk kjjj

shellsjj

22221

20 ]22[)()( −−

=

++= ∑

Measure:

From ab-initio calculations or from reference compounds

( )h

h BEmk

−=

ω2


5

XANES• X-ray Absorption Near Edge Structure

(also NEXAFS)

• “Molecular orbital” approach: 1 electron approximation, constant matrix element: probe site and symmetryprojected density of states of final states

• “Multiple scattering” approach: structural interpretationthrough simulation

)(ˆ422

fEfri ρεωαπσ rh •=

)(,

0,1

sdpps

m

→→

=±=light) pol. (lin.

ll ∆∆


6

Characteristics of XAFS• Atomic selectivity

– most elements can bestudied

• High resolution for first few coordination shells– signal from higher shells

is damped• Sensitivity to high dilutions• Equally applicable to ordered

or disordered matter

• Today’s topics– Dopants, defects– Alloys, local distortions– Thin films, interfaces– Amorphous solids– Phase transitions– Vibrational dynamics– Nanostructures

• Ge quantum dots• Metallic clusters• Core-shell nanoparticles


7

Role of XAFS in Materials Science

Nakamura et al., Jpn. J. Appl. Phys. 35, L217, 1996

Growth

MBE@TASC

Structure

Physical Properties

XAS


8

XAFS and dopants• Only the structure around the photo-

excited atom is probed• Fluorescence detection greatly

enhances sensitivity• Present sensitivity limit

– dopants in the bulk ∼ 1018 at/cm3, – thin films (single layer) ∼ 1014 at/cm2


9

Arsenic in a-Si:H• As in a-Si expected to be

3-fold coordinated• But: doping is observed• Doping in c-Si due to 4-fold As• Knights et. al. observe increase

in CN at 1% As concentration– evidence of active As

• Knights, Hayes, and MikkelsenJr.Phys. Rev. Lett. 39, 712 (1977)

As

Coor

dina

tion

Num

ber

(CN

)

1 % 10 %

3.0

2.5


10

Si/GaAs δ-doped layers and superlattices

• δ-doping: physical separation of dopants and charge carriers– Si-GaAs a technologically

relevant case• Number of charge carriers

not proportional to numberof dopants above a critical

concentration

Boscherini, Ferretti, Bonanni, Orani, Rubini, Piccin, and Franciosi APL 81, 1639 (2002)


11

Si-GaAs: Open issue• What is the structural origin of the

defects which give rise to charge carriersaturation?

• Growth conditions optimized for Si/GaAssuperlattice formation: – MBE, 540 °C, 4 nm/h, @TASC

• Si thickness: 0.02 ML – 4 ML; repetitionoptimized for XAS

• Fraction of electrically active Si atoms:2×10–4 – 6× 10-2


12

Si-GaAs: data• Si K edge XAS @

LURE• Fluorescence

mode, bulk sensitivity

• Si – Si clusteringapparent in rawdata

2 4 6 8 10 12k (Å-1)

0.5 Å-1

0.2 ML Si

2 ML Si

Bulk Silicon

Si - As theory

k χ(

k) (Å

-1)


13

Si-GaAs: results

• Si – Si CN > 2 always

• Presence of Si clusters

• This is the originof lowconcentration of charge carriers

2

2.5

3

3.5

4

10-4

10-3

10-2

10-1

10-2 10-1 100 101

N (Si) R*

Si -

Si C

oord

inat

ion

Num

ber Fraction of active Si atoms

Thickness (ML)


14

Low Z dopants and XAS• C, N & O often used as dopants• Experimentally difficult: low fluorescence

yield, soft X-rays, UHV

ALOISA beamline @ ELETTRA10-4

10-3

10-2

10-1

100

5 10 15 20 25 30

Fluo

resc

ence

Yie

ld

Z

K edges LIII

edges

K edge yield

LIII

edge yield


15

Dilute nitrides: GaAs1-yNy, InxGa1-xAs1-yNy

0.0 0.5 1.0

3.25 eV

Eg (eV)

GaAs

GaN

y

1.42 eV

Anomalous non-linear optical and electronic proprieties of III-V nitrides

• Red shift of the band gap by adding few % of nitrogen (≈ 0.05-0.1 eVper N atomic percent in InGaAsN )

• Huge and composition dependent optical bowing


16

Hydrogen – nitrogencomplexes in dilute nitrides

• Hydrogenation/deuteration leads toreversible changes– compressive strain– opening of band gap

• [1018 ions/cm2, 100 eV]

Ciatto et al, Phys. Rev. B 71, 201301 (2005) Bisognin et al, APL 89, 061904 (2006)

X-ray diffraction photoluminescence

GaAs

GaAsN


17

Hydrogen – nitrogen complexesin dilute nitrides

• Which is the hydrogen –nitrogencomplex responsiblefor these changes?

Some candidate low energy structures


18

N K edge experiment

• Full multiple scattering XANES simulations

• Good agreement forGaAsN

400 410 420 430 440Energy (eV)

experiment

simulation

GaAsN

Nor

mal

ized

XA

NES

(a)


19

Hydrogen – nitrogen complexesin dilute nitrides

• DFT calculations to determinelowest energy geometries

• Full multiple scattering XANES simulations

• Answer: C2v – like complexesare mostly present

• 3-D sensitivity of XANES!!390 400 410 420 430

Abs

orpt

ion

cros

s-se

ctio

nEnergy (eV)

N-HBC

N-H3

C2v

As grown

Hydrogenated


20

How many H atoms are there?

• Measurement of D concentration: in as-deuteratedsamples there are 3D atoms per N!– Where is the third

D atom? – What is its effect

on lattice and gap? Nuclear reaction analysis


21

Annealing studies

• 250 °C annealing:– 2 H/N– Compressive strain

released– No effect on gap

• 328 °C annealing– 0 H/N– Strain back to tensile– Gap back to that of

GaAsN


22

XANES: the optically activecomplex really is the C2v one

GaN

H


23

XAFS and alloys• High resolution in probing the local

coordination in first few coordinationshells

• Study, as a function of composition– Deviation of local structure from

average structure– Atomic ordering


24

Bimodal distribution of bondlengths in InGaAs

• InxGa1-xAs: tuning of lattice parameter and band gap

• Does the local structure followthe average structuresuggested by Vegard’s law?

• NO! Bond legths stay close tosum of covalent radii

• Mikkelsen Jr., and Boyce, Phys. Rev. Lett. 49, 1412 (1982)


25

XAS and amorphous solids

• Sensitivity only to very local coordination isuseful to avoid overlap of RDF peaks

• In alloys, atomic selectivity is further help


26

XAFS vs. X-ray scattering

• Comparison between X-ray scattering and Si K-edge XAFS in a-Si1-xCx:H alloys– F. Boscherini,

Mat. Res. Soc. Symp. Proc. 258, 217 (1992)

XRS XAFS


27

Amorphous semiconductors: ordering• Different possibilities for relative arrangement

for A1-xBx alloy (A,B equal CNtot = 4 ):– Random: CNA-B = 4x, CNB-A=4(1-x)– Clustering: CNA-B = 0, CNA-A = CNB-B =4– Chemical order: for x < 0.5, CNB-A = 4, CNB-B = 0

• Compare composition (from other method) with CNi-j (from XAFS)


28

Amorphous semic.: ordering• Clustering never observed• Degree of chemical

ordering depends on enthalpy of heteroatomicbond, e.g.:– a-Si1-xCx:H, ordered– a-Ge1-xSnx , random

Pascarelli, et al Phys. Rev. B 45, 1650 (1992);Phys. Rev. B 46, 6718 (1992).

NSi

-C/N

Si


29

Using SR linear polarization to study thin films

θθσ 2)( Cos∝θ

e-Electron orbit

ε̂


30

Hexagonal and cubic GaN


31

Cubic and hexagonal GaN• N K-edge XAS to study

relative amounts of cubic and hexagonal GaN

• Exploit– linear polarization of SR– polarization dependence of

cross-section• XAS signal must exhibit (at

least) point group symmetryof the crystal– Td: isotropic signal– C6V: σ tot(E,θ)=σ iso(E)+(3Cos2θ −1)σ1(E)


32

GaN

• Katsikini et. al., APL 69, 4206 (1996); JAP 83, 1440 (1998)

CUBIC HEXAGONAL


33

The Fe/NiO(001) interface

P. Luches, V. Bellini, S. Colonna, L. Di Giustino, F. Manghi, S. Valeri, and F. Boscherini ,

Phys. Rev. Lett. 96, 106106 (2006)


34

FM/AFM systemsFM/AFM systems

AFM

FM

exchange bias

heating up to TN+

cooling in H


35

The Fe/The Fe/NiONiO(100) interface(100) interface

NiOAF TN=520 K

aNiO= 4.176 Å

rock-salt

aFe= 2.866 ÅdFe = 4.053 Å

bcc

FeFM TC=1040 K

m=-2.8%


36

Fe K-edge XAFS experimentFe local atomic environment

Polarization dependence of XAFS cross section

in-plane local environment

kE

out-of-plane local environment

kE

GILDA beamline @ ESRF


37

ex-situ growth @ S3 laboratories

Ag(001)

10 ML NiO

Fe

10 nm Au

samplessamples


38

XANESXANES

• deviation frommetallic character

• shift of the white line towards FeO


39

XANES: XANES: polarizationpolarization dependencedependence

kE

θ

• planar FeO phase


40

EXAFSEXAFSkE

θ


41

EXAFSEXAFS--quantitativequantitative analysisanalysis

5.095.014.9688th shell

4.844.87167th shell

4.744.644.75

86th shell

4.074.1745th shell

3.924.044.05

84th shell

2.872.9543rd shell

2.812.772.87

22nd shell

2.502.502.4881st shell

fit results, R (Å)

BCT Fe, R (Å)

BCC Fe, R (Å)

N

the epitaxial strain ispartially released at 10 ML


42

2 ML 2 ML samplesample


43

The modelThe model

buckledpseudomorphicFeO layer

BCT Fe

• expanded FeO-NiO and FeO-Fe distance


44

XAS and highly correlated oxides• Manganites, HTC’s, perovskites• High resolution in the determination

of local structure• Local distortions couple to physical

properties• Local deviations from average

structure


45

Ferroelectric Phase transitions in PbTiO3

• At Tc = 763 K PbTiO3 undergoes tetragonal to cubicphase transition

• T < Tc it is ferroelectric (permanent dipole moment)• Phase transition believed to be purely displacive (no

local distortion for T > Tc)

low temp

Sicron, Ravel, Yacoby, Stern, Dogan and Stern, Phys. Rev. B 50, 13168 (1994)


46

Ferroelectric Phase transitions in PbTiO3

• Ti and Pb XAFS data • Local lattice parameters

and local distortionsdo not change at Tc

• The ferroelectrictransition has a largeorder – disordercharacter: the orientationof local distortionschanges at Tc

c

a


47

La1-xCaxMnO3 Manganites

• For 0.2 < x < 0.5– Metallic ferromagnet T < Tc

– Insulating paramagnet T > Tc

• Colossal Magneto Resistance• Double exchange model:

hopping conductivity betweenadjacent Mn ions is enhancedif Mn ions are spin aligned Mn3+

Mn4+


48

La1-xCaxMnO3 Manganites• Mn(3+)O6 octahedron in

LaMnO3 is Jahn-Tellerdistorted– Mn4+ is not

• Conductivity in the presence of polarons: anelectron + a lattice distortion

• Range of polaronicdistortion?


49

La1-xCaxMnO3 Manganites• Presence of

localizedstructuraldistortions at MI transition

Booth, Bridges, Kwei, Lawrence, Cornelius and NeumeierPhys. Rev. Lett. 80, 853 (1998)


50

La1-xCaxMnO3 Manganites

• Magnetic field reduces local distortions

Meneghini, Castellano, Mobilio, Kumar, Ray and Sarma, J Phys.: Condens. Matter 14, 1967 (2002)


51

Thermal vibrations in crystals• XAFS detects relative atomic motion

– Useful to study of thermal vibrations• In the harmonic approximation the effect of

thermal motion on χ(k) is via( )[ ]20

20

ˆjojj uuR rr

−•=σ

jR0

r

our

jur


52

Correlated atomic motion in AgI

• Dalba, Fornasini, Rocca and Mobilio, Phys. Rev. B 41, 9668 (1990)

1st shell2nd shell


53

Anharmonic vibrations in AgI• High maximum wave-

number leads to great sensitivity to details of radial distribution function

• Detect asymmetries in RDF connected to deviations from harmonic approximation to interatomic potential

• Dalba, Fornasini, Gotter and Rocca, Phys. Rev. B 52, 149 (1995)


54

Vibrations parallel & perpendicular to interatomic vector in Ge

• Distance measured in EXAFS is not equal to distance between average position of atoms:

• Perpendicular motion does not contribute to σ2, so that

R

uRC 2

*1

2⊥+=

∆

<u2⊥>

<u2||>

0 600 K2||

2 u∆σ =• Dalba, Grisenti, Fornasini, and Purans,

Phys. Rev. Lett. B 82, 4280 (1999) Å

2


55

Nanostructures


56

XAS to study nanostructures• XAS is a local, short range, effect

– Origin: core hole lifetime (τhole = 10-16 – 10-15 s) and electron mean free path (5 – 10 Å).

– EXAFS signal is damped by e-(r/λ) where

• Same formalism applies to molecule, cluster or crystallinesolid– insenstive to variations of morphology– sensitive to low thicknesses, high dilutions

• Excellent probe of variations in local environment uponreduction of dimensions and/or dimensionality

( )hole

e

mfp km

τλλ111

h+=


57

Local nature of XAS


58

Semiconductor quantum dots

• The prototypical nanoscience system• Self organized hetero-epitaxial

growth can lead to quantum dots withnarrow size distribution

• Due to small size the quantum dotsexhibit new physical properties


59

Ge Quantum Dots

• Need for understanding of local bonding

• Preparation:– Ge/Si(001) by CVD @ 600 °C,

Univ. Roma Tre• Ex-situ AFM used to characterize

degree of relaxation– Ge/Si(111) by MBE @450 - 550

°C, Univ. Roma II• In-situ STM/AFM

Boscherini, Capellini, DiGaspare, Rosei, Motta, and Mobilio, Appl. Phys. Lett. 76, 682 (2000)


60

Energetics of island formation• Competing energies:

– strain– surface– dislocations

• Contributions from:- wetting layer- islands

Wetting layer WL+Strained island WL+Relaxed island

"Coverage"

WL+2D platelet


61

Q. Dots: AFM• Analysis of aspect ratio

provides measurementof relative amount of relaxed islands

• Ge/Si(001): Full rangeof relaxation examined

0

20

40

60

80

100

0 5 10 15 20 25 30 35 40

Rel

axed

vol

ume

(%)

equivalent thickness (nm)

Ge/Si(001)

(1 ML = 0.135 nm, WL = 3 ML)


62

Q. Dots: Ge edge data

0 2 4 6

Ge in Si

Ge bulk

dots

interatomic distance (Å)

Ge

Fou

Tra

nsf

0

1

2

3

4

2 4 6 8 10 12 14 16k (Å-1)

k χ(

k) (

Å-1

)

Ge:Si

(111)1 nm

(001)7.8 nm

(001)38 nm

Ge

•Assuming random alloyaverage compositionis Ge0.70Si0.30


63

Conventional SK growth


64

SK growth with interdiffusion


65

Physical origin of intermixing• Keating potential

– intermixing reduces the number of “stretched”bonds and hence the strain energy

– thermodynamical and kinetic effects– strain enhanced diffusion– intermixing is reproduced by MC simulations which

include “atomic identity flips” (P. Kelires, 2001)• Intermixing must be considered in realistic models of

Stranski-Krastanov growth

V({Rij}, {θ ijk}) = α2 (Rij – Rij

0)2Σij

+β8 Re

2 (Cos θijk + 13)

2Σijk


66

Metallic nanostructures


67

Clusters: bond length contraction

• A bond length contraction has beenfound for weakly supported metallicclusters (Ni, Cu, Au……) for d < 100 Å

Apai et. al.Phys. Rev. Lett. 43, 165 (1979) Montano et. al.

Phys. Rev. Lett. 56, 2076 (1986)


68

Bond length contraction• A macroscopic surface tension interpretation

(“liquid drop”) can explain the bond lengthcontraction– f surface tension,

κ bulk modulusr radius of spherical particle

– Montano et. al.Phys. Rev. B 30, 672 (1984)Balerna et. al.,Phys. Rev. B 31 5058 (1985)

rf

RR κ

32

−=∆


69

Dynamic properties of Au clusters• The Mean-Square-Relative-

Displacement, σ2, damps the EXAFS signal with a term

• σ2 :– depends on thermal motion– can have structural

contribution

222 σke−

( ) 2

0020

ˆjjj Ruu •−=

rrσ

Bulk Au


70

Dynamic properties of Au clusters• As the cluster dimensions

decrease an enhacement of σ2 is evident

• Surface atoms have lessmotion constraints

• High surface-to-volumeratio for nanoclusters

• Values reproduced bynumerical model for freesphere phonon DOS whichincludes surface modes

• Balerna and Mobilio, Phys. Rev. B 34, 2293 (1986)


71

XX--ray absorption ray absorption spectroscopy studies of spectroscopy studies of Fe/FeFe/Fe--oxide granular oxide granular

nanostructuresnanostructures


72

Granular Granular FeFe//FeFe--oxideoxide nanostructuresnanostructures

• Changes in structure and physical properties induced by reduced dimensions

• Effect of disorder on magnetic properties

• Comparison with 2D systems (surfaces)

• Applications


73

SampleSample depositiondeposition: IGC: IGC

Inert Gas Condensation

In situ compacting

• Evaporation in He (133 Pa)• O2 passivation• Cold compacted at 1.5 GPa


74

SampleSample morphologymorphology

Core: α-Fe nanoparticles<d> = 5 – 21 nmOxide matrix: ultra-fine grains

(2-3 nm) with spinel structure


75

Inverse spinel structure

- Magnetite (Fe3O4)Td: Fe3+; Oh: Fe3+ & Fe2+

- Maghemite (γ-Fe2O3)Fe3+ in both Td and Oh sites

BCC

α-Fe

SampleSample morphologymorphology & & structurestructure

a

b

c


76

MatrixMatrix: magnetite or : magnetite or maghemitemaghemite? ? • Two oxides have significantly different

properties (e.g. conductivity)• XRD not sensitive (broad peaks)• Use XAFS

– Element and local sensitivity– Unaffected by morphology changes– Phys. Rev. B 68, 195423 (2003)

Fe K-edgeGILDA @ ESRF

O K-edgeBEAR @ ELETTRA


77

O O KK--edgeedge XANESXANES

• O edge: informationonly on the matrix

• Clear variation of edgeposition with core size–7 nm ∼ maghemite

(94%)–21 nm ∼ magnetite

(83%)520 525 530 535

Energy (eV)

21 nm

10 nm

7 nm

γ-Fe2O

3

Fe3O

4


78

FeFe and O and O edgeedge EXAFSEXAFS

• Fe edge: core & matrix– Core

• bcc Fe, σ2 decreases with <d>

– matrix• 7nm similar to maghemite

• O edge: matrix– High values of σ2 (~35 × 10-3

Å2)– magnetite → maghemite

transition with decreasing <d>


79

SizeSize--dependentdependent magnetite magnetite totomaghemitemaghemite transformationtransformation

• α-Fe core exposed to O2→ formation of Fe3O4

• Further exposure to O2→ formation of γ-Fe2O3 on surface of grains

• Enhanced transformation due to higherdisorder, grain boundaries in smaller particles


80

Thanks• Funding: MIUR, INFM, Uni Bo, ESRF, EU


81

High resolution XAS @ N edge

• M. Petravic et al., Chem. Phys. Lett.425, 262 (2006)

• Vibrationally split 1s →π* transitionfor N implanted in compoundsemiconductors


82

M. Malvestuto, R. Carboni, and F. Boscherini F. D’Acapito S. Spiga, M. Fanciulli, A. Dimoulas, G.

Vellianitis, and G. Mavrou, Phys. Rev. B 71, 075318 (2005).

Y2O3 thin films on Si(001)


83

High-κ oxides

From: G. Lucovsky


84

XAFS and high-κ’s

• Local atomic and CB electronicstructure of high-k films:– Similarity to bulk oxides?– Disorder, interdiffusion?

• Interfaces with Si:– nature of interface phases


85

The bixbyite structure

• Two inequivalentcation sites, bothoctahedral– site I: all O at

same distance– site II: 3

differentdistances


86

Y2O3/Si(001): growth

• Dimoulas group, Athens• MBE from ceramic target at 450 °C• 5 samples: 2, 4 and 20 nm• Some annealed 30 min. 500 °C

Malvestuto et al., Phys. Rev. B 71, 075318 (2005)


87

Y2O3/Si(001): Y K edge XANES• Probes p-type CB

DOS projected on Y• Overall similarity to

bulk spectrum– 2 nm samples:

broader spectrum

thic

knes

s

Energy (keV)

Yttria

# 5

# 3# 4

# 2# 1

Abs

orpt

ion

coef

ficie

nt Y K-edge

17 17.1


88

Y2O3/Si(001): Y XANES simulation• Simulation within

“full multiple scattering “theory (FEFF) for20 nm sample– 150 atom cluster– self consistency up to 4 Å

• Good correspondence with5p PDOS of Y in bixbyitecluster

• Similar PDOS for the twoinequivalent sites

17040 17070

Experiment

PDOS 5p

PDOS 5p

site I

site II

Simulation

Energy (eV)


89

Y2O3/Si(001): Y EXAFS• 4 and 20 nm samples

– Sharper 2nd and 3rd shellpeaks with increasingthickness and withannealing

• 2 nm samples– 1st shell peak shifts to

lower energy– a combination of Y – O

and Y – Si bonds

0 1 2 3 4 5 6M

ag. F

ou. T

rans

f. [k

2 χ(k)

]

R (Å)

# 5# 4# 3

# 2# 1

Yttria

Y-O

Y-Y

Y-Y

Y K-edge

2 4 6 8 10 12k (Å-1)

thic

knes

s


90

Y2O3/Si(001): O XANES• 4 and 20 nm

samples: similarto bulk yttria

• 2 nm samples: contributionfrom SiOx

• Good agreement with simulations 525 535 545 555

Energy (eV)A

bsor

ptio

n co

effic

ient O K-edge

SiOx# 1# 2

# 4 # 3

# 5Yttria

sim.

PDOS2p


91

Y2O3/Si(001): O EXAFS• 20 nm sample• Good agreement with

bulk structure– 4 Y atoms, 2.2 – 2.3 Å– 10 O atoms, 2.9 – 3.7 Å

0 1 2 3 4 5Mag

. Fou

. Tra

nsf.

[k2 χ

(k)]

R(Å)

O K-edge# 5

2 3 4 5 6k (Å-1)


92

Y2O3/Si(001) conclusions: films

• For thickness ≥ 4 nm atomic and electronic structure correspond tobixbyite Y2O3

• Decreasing static disorder withincreasing thickness and/or annealing(less defects seen in electron microscopy)


93

Y2O3/Si(001) conclusions: interface• Presence of SiOx

• Y – Si bonds with R = 2.09 Å (2.07 Åin YSi2)

• At 610 °C deposition temperature YSi2 seen from XRD

• Y – Si bonds are “precursor bonds”


94

“Fingerprint approach” to XANES: Fe oxides

• Change of XANES lineshapewith oxidationstate, localsimmetry


95

Evolution of XAFS• XAFS has greatly benefited from

– Improvements in SR sources• high brilliance, polarization, extended energy range,

stability, reliability– full theoretical understanding and reliable

analysis programs

• XAFS a reliable tool,but still interesting as a physical effect

σ = σ 0

3 Sin2δl=10 Im{T 1 −TG( )−1}1m ,1m

0,0

m∑

applications of xafs to materials science and...

Documents