Introduction to EELS in the AEMIntroduction to EELS in the AEM

Nestor J. ZaluzecNestor J. Zaluzec

zaluzec@microscopyzaluzec@microscopy.com.com

zaluzec@aaemzaluzec@aaem..amcamc..anlanl..govgov

What are the Limits - Today?What are the Limits - Today?

S. Pennycook etal from the TEAM Project Report

5 Å Spectroscopicidentification atsubnanometerresolution.

8% collectionefficiency

820 850 880

Inte

nsity

Energy (eV)

La M4/5

La in CaTiO3grown by MBE

P.E. Batson IBM Yorktown, Jan 2004

AEM - High Spatial Resolution Analysis

Electronic Structure changes at the Si/SiO2 Interface

Brief Review of Energy Loss ProcessesBrief Review of Energy Loss Processes

Instrumentation: Detector SystemsInstrumentation: Detector Systems

Instrumentation: AEM Systems Instrumentation: AEM Systems

Data Analysis and Quantification: Data Analysis and Quantification:

Advanced TopicsAdvanced Topics

Brief Review of Energy Loss ProcessesBrief Review of Energy Loss Processes

Electron Excitation of Inner Shell & Continuum Processes Electron Excitation of Inner Shell & Continuum Processes Spectral Shapes Spectral Shapes Notation of Edges Notation of Edges

Electron Scattering Angular DistributionsElectron Scattering Angular Distributions

The Emission Process:The Emission Process:

1-Excitation1-Excitation2-Relaxation2-Relaxation3-Emission3-Emission

Electron Distribution Relaxation

Ejected Inner ShellElectron

Incident Electron

Inelastically ScatteredPrimary Electron

X-ray PhotonEmission

Internal Conversion and Auger Electron

Emission

Electron Excitation of Inner Shell Processes

Eo

Eo-δE δE

Electron Energy Loss SpectroscopyElectron Energy Loss Spectroscopy

Measure the changes in the energy distribution of an electron beam transmittedMeasure the changes in the energy distribution of an electron beam transmittedthrough a through a thinthin specimen. specimen.

Each type of interaction between the electron beam and the specimen produces aEach type of interaction between the electron beam and the specimen produces acharacteristiccharacteristic change in the energy and angular distribution of scattered change in the energy and angular distribution of scatteredelectrons.electrons.

The energy loss process is the The energy loss process is the primaryprimary interaction event. All other sources of interaction event. All other sources ofanalytical information ( i.e. X-rays, Auger electrons, etc.) are analytical information ( i.e. X-rays, Auger electrons, etc.) are secondarysecondary products productsof the initial inelastic event. Thus, EELS has the highest potential yield ofof the initial inelastic event. Thus, EELS has the highest potential yield ofinformation/inelastic eventinformation/inelastic event

Geometrical Collection Efficiency in XEDS

!" = "4#=

14#

AR2

Geometrical Collection Efficiency in EELS

!$ = ln 1 + $

%E

2

ln 2%E

%E = &E2'oTo

; 'o = 11-$2

; To = 12

mov 2 = 255.530 1- 11+VO(kV)

511.060

2 k e V

Accelerating Voltage Vo

(kV) To =12 mov2 (keV).

Energy Loss

&E (eV)

Accelerating Voltage (kV) Scattering Angle %E (mr)

50 43.5 10 120 0.057100 76.8 100 120 0.569120 87.9 200 120 1.14150 102.8 500 120 2.84200 123.5 1000 120 5.69250 140.3 500 100 3.26300 153.8 500 200 2.02350 165.5 500 300 1.62400 175.1 500 400 1.431000 226.3 500 1000 1.10

!

2 "

e -

Coll imator

Aper tu re

XEDS

Character i s t i cX-ray Emission

Angular Dependance(Isotropic Distr ibution)

Character i s t i cElectron Energy Loss Angular Dependance

(Aniostropic Distr ibution)

d-band

s-band

1 2 3

AES

XPS/UPS

XEDS/XRF

EELS/XAS

DO

S

Incident Electrons Incident Photons

L shell M shell2 4 531

}

Schematic Diagram Illustrating Sources ofSchematic Diagram Illustrating Sources ofInelastic Scattering SignalsInelastic Scattering Signals

Experimental XEDS, XPS, and EELS data from the Copper L shell. Note theExperimental XEDS, XPS, and EELS data from the Copper L shell. Note thedifferences in energy resolution, and spectral featuresdifferences in energy resolution, and spectral features.

Comparision Comparision Light Element Spectroscopy ResolutionLight Element Spectroscopy ResolutionXEDS XEDS vs vs EELSEELS

Comparision Comparision of WL XEDS Detector and EELS spectraof WL XEDS Detector and EELS spectrataken from the same taken from the same NiO NiO specimenspecimen

Note the enhanced spectral information in the EELS data. Vertical scale isNote the enhanced spectral information in the EELS data. Vertical scale isarbitrary andarbitrary and chozen chozen for clarity of presentation. for clarity of presentation.

HeliumHeliumIn PdIn Pd

D. D. Taverna etalTaverna etalPRL 100, 035301 (2008)PRL 100, 035301 (2008)

Electron Scattering

Scattering from an isolated atom in free space

0 10 20

Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

PhononInner-Shell

Unscattered

R = Average Interatomic Spacing

Scattering from an collection of an amorphous collection of atoms - neighboring atoms give rise to interference

AmorphousSilica (SiO2)

0 10 20

Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

PhononInner-Shell

Unscattered

0 10 20

Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

PhononInner-Shell

Unscattered

Unfiltered Elastic

Plasmon

Elastic Plasmon

0

1 105

2 105

3 105

4 105

5 105

6 105

7 105

-12 -8 -4 0 4 8 12

Amorphous Carbon

Expt - 50eV Theory - Normalized

Rel

ativ

e In

tens

ity

Angle (mR)

0

5000

1 104

1.5 104

2 104

2.5 104

3 104

3.5 104

4 104

270 280 290 300 310 320 330

0.110.35.50

Rel

ativ

e In

tens

ity

Energy Loss (eV)

θE ~1.15 mR

Intensity of Edges is Directly Related to Mean Scattering AngleExample: Amorphous Carbon

Intensity of Edges is Directly Related to Mean Scattering AngleExample: Boron - Nitride

θE ~1.62 mRθE ~0.777 mR

0 10 20Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

Phonon

Unscattered

Inner-Shell

0 10 20Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

Phonon

Unscattered

Inner-Shell

Unfiltered Elastic

1st Plasmon2nd Plasmon

0 eV

15 eV30 eV

Instrumentation: Detector Systems

Energy Loss Spectrometers Basic Principles Electrostatic/Electromagnetic Serial/Parallel Detector Systems Spectral Artifacts

Multichannel Analyzers

CommericalCommericalSpectrometersSpectrometers

VsVsImaging FiltersImaging Filters

Imaging Filters Imaging Filters vs vs SpectrometersSpectrometers

Spectral ImagingSpectral ImagingSlice-by-Slice Slice-by-Slice vs vs Point-by-PointPoint-by-Point

Unfiltered Elastic

PlasmonUnfiltered + Obj. Aperture

Unfiltered Elastic

1st Plasmon2nd Plasmon

Silicon 111 ZAP- Gamma Corrected

Spectroscopy Spectroscopy vs vs Filtered ImagingFiltered Imaging

W. W. Grogger etal Grogger etal MM 2003MM 2003 9, s.3 , 729, s.3 , 72

Data Analysis and Quantification:Data Analysis and Quantification:

Spectral Processing Spectral Processing Thin Film Quantification Methods Thin Film Quantification Methods Specimen Thickness Effects Specimen Thickness Effects

MM2323CC6 6 in Steel :in Steel :Spectral OverlapSpectral Overlap

SpectralSpectralOverLapOverLap

Problems alsoProblems alsoexist in EELSexist in EELS

Two Related Methods are sometimes used:Two Related Methods are sometimes used:

Second Difference Filtering ( Second Difference Filtering (Shuman etal MAS,1983Shuman etal MAS,1983))

Record 3 Energy Loss Spectra which are displaced in energy by dERecord 3 Energy Loss Spectra which are displaced in energy by dEMathematically combine in computer to form the Second DifferenceMathematically combine in computer to form the Second Difference(SD) Spectrum(SD) Spectrum

SD(E) = ISD(E) = I11(E-dE) - 2 I(E-dE) - 2 I22 (E) + I (E) + I33 (E+dE) (E+dE)

Spectrum has the appearance of a derivitive, removes channel toSpectrum has the appearance of a derivitive, removes channel tochannel gain variation in parallel EELS and slowly varyingchannel gain variation in parallel EELS and slowly varyingbackgrounds. Sharp features are enhanced in visibility.backgrounds. Sharp features are enhanced in visibility.

Digital FilteringDigital Filtering

This is related to a simple numerical differentiation of a singleThis is related to a simple numerical differentiation of a singlespectrumspectrum

IIkk = P = Pkk* I* Ioo

IIkk = Number of electron having excited a kth inner shell= Number of electron having excited a kth inner shell

PPkk = Probability of excitation of the kth shell = Probability of excitation of the kth shell

IIoo = Incident electron current= Incident electron current

PPkk = N = N σσkk

N N = Number of atoms of the element analyzed= Number of atoms of the element analyzedσσkk = Ionization cross-section for the kth shell= Ionization cross-section for the kth shell

Alternatively;Alternatively;

Consider the ratio of Intensities of any two Edges in the same spectrum IConsider the ratio of Intensities of any two Edges in the same spectrum IA A and I and IBB

Invoke the Ratio Method and obtain the exact equation:Invoke the Ratio Method and obtain the exact equation:

Note the similiarity of this equation with that of Thin Film XEDSNote the similiarity of this equation with that of Thin Film XEDS

But in the real world the assumptions used in theabove simple arguments are never realized:

• Measure all scattered electrons (!="=180 0)• Integration over all energy Losses

Because we must measure over a finite energywindow (#E) we modify the expression to:

NA= IA(#E)

!$A(#E)*I 0

we also measure over a finite angular window (!)and therefore, we modify the expression to:

NA= !IA(#E,!)$A(#E,!)*I0

and the ratio equation becomes:

NANB = kAB

IA(#E,!)I B(#E,!)

with

kAB = $B(#E,!)$A(#E,!)

!

"

Specimen

IncidentBeam

Convergence

InelasticScattering

Angle

e -

Hexagonal Boron- Nitride

θE ~1.62 mRθE ~0.777 mR

Problems in EELS QuantificationProblems in EELS QuantificationCross-section CalculationsCross-section Calculations

Problems In EELS QuantificationProblems In EELS Quantification Collection Angle Errors Collection Angle Errors

0 10 20Scattering Angle [mR]

I( )

/Io!

Elastic

Plasmon

Phonon

Unscattered

Inner-Shell

Effects of Specimen ThicknessEffects of Specimen Thickness

• Multiple ScatteringMultiple Scattering

•• Low Loss Low Loss

•• Core Loss Core Loss -Visiblity-Visiblity

•• Quantification EffectsQuantification Effects

•• To measure the thickness ofTo measure the thickness ofcompare the intensity of thecompare the intensity of thezero loss (Izero loss (I00) to the total) to the totalintegrated intensity in theintegrated intensity in thespectrum (Ispectrum (ITT).).

•• This ratio is directlyThis ratio is directlyproportional to the localproportional to the localthickness of the specimen.thickness of the specimen.

t = t = λ λ * * ln ln (I(Io/o/// IITT))λ =λ = mean free path mean free path

0

1 105

2 105

3 105

4 105

5 105

-100 0 100 200 300 400

Inten

sit

y

Energy Loss (eV)

I0

IT

450 500 550 600 650 700 750

Experimental EELS Edge/Background Ratioas a Function of Specimen Thickness

at Constant Accelerating Voltage

EELS

Inte

nsity

Nor

mal

ized

to P

re-E

dge

Bgn

d

Energy Loss (eV)

t/ λ = 0.23

t/ λ = 0.62

t/ λ = 1.21

O K Edge in NiO

VariationVariationin EELSin EELSEdge P/BEdge P/B

withwithThicknessThickness

VariationVariationin EELSin EELSEdge P/BEdge P/B

withwithkVkV

450 500 550 600 650 700 750

EELS

Inte

nsity

Nor

mal

ized

to P

re-E

dge

Bac

kgro

und

Energy Loss (eV)

Experimental EELS Edge/Background Ratioas a Function of Accelerating Voltage

at Constant Specimen Thickness

O K Edge in NiO

300 kV

200 kV100 kV

Specimen Thickness Effects on EELSSpecimen Thickness Effects on EELS

Deconvolution Deconvolution of Multiple Scatteringof Multiple Scatteringusing using Leapman/Swyt Leapman/Swyt MethodMethod

Orientation & Thickness Effects on EELS SignalsOrientation & Thickness Effects on EELS Signals

Specimen ContaminationSpecimen ContaminationThe The Microscopists Microscopists BaneBane

15 sec15 sec

30 sec 30 sec

60 sec 60 sec

120 sec 120 sec

300 sec 300 sec

Spectral ImagingSpectral ImagingSlice-by-Slice Slice-by-Slice vs vs Point-by-PointPoint-by-Point

Boron Nitride on Holey Carbon

Elastic Plasmon

Energy Loss0 200 400

170 eV 195 eV

265 e V 290 eV

BN Data

170 eV 195 eV

Filtered Elemental Imaging

Net Boron K Image

265 e V 290 eV

BN Data

Filtered Elemental Imaging

Net Carbon K Image

EFTEM elementalmaps of mousepancreaticIslet cells

Nitrogen PhosphorusSulfur

Mut

Mut

Cnt

Cnt

Leapman et al. (2003)Leapman et al. (2003)

1µm

Shows sulfur-richinsulin granules in β cells

Model of axon andoligodendrocyte showinganatomy of degeneration

0

2

4

6

8

10

12

600 650 700 750 800

CCD

Coun

ts/1

000

Energy Loss (eV)

EELS of ferritin molecule extracted from spectrum-image

Fe L2,3 edge

100 nm

FerritinFe

STEM-ADF

Ferritinlocalizedin invaginationsofoligodendrocytes

Mapping Ferritin in Brain:Misregulation of Iron Metabolism inIRP Knockout Mice

Pre Fe L2,3 Post Fe L2,3

STEM-EELS

(P. Zhang et al., J. Struct. Biol. 2005)

Hofer & Hofer & Warbichler Warbichler - Graz- Graz TUTU

Hofer & Hofer & Warbichler Warbichler - Graz- Graz TUTU

Steps in Quantitative (Elemental) Analysis

Select the operating mode: Is it appropriate? CTEM, STEM..... Image Coupled, Diffraction Coupled

Obtain a typical spectrum Optimize the experimental conditions

Accelerating Voltage (maximum consistent with your specimen) Chooze the best Edges to analyze (K, L, M,....) Optimize β for the weakest edges (θE= δE/2Eo) Optimize α to minimize problems (α < β/2 ) Optimize the Acquisition Mode

Normal - High Concentrations Difference - Low Concentrations

Optimize the Data Acquisition Select Energy Resolution and Range DQE & Statistics

Process the Data to Extract Intensities Normal Difference Reference Spectra

Check the relative specimen thickness (t/λ< 1) Calculate the compositions

Absolute # of atoms Relative # of atoms Standards/Standardless?

ReCheck for Artifacts/Problems Diffraction Orientation/Channeling Unidentified Edges Spectral Overlaps Radiation Damage

Core Loss Electron Core Loss Electron SpectrosopySpectrosopy

Linear Linear dichroismdichroism::

The linear polarization of theincident electrons parallel to thedirection of momentum transfer.This polarization can be usedsense the anisotropy of thevalence states involved in thecore excitation process. It candetect the number or changes inthe number of the valence holesin different directions of theatomic volume.

In many cases, the anisotropy ofthe charge in the atomic volumeis caused by crystal-fieldinteraction and is due to ananisotropy in the bonding.

Linear Linear Dichroism Dichroism andand

Momentum Resolved EELS (MREELS) in Core Loss SpectroscopyMomentum Resolved EELS (MREELS) in Core Loss Spectroscopyderives its information from transitions from initial to final statesderives its information from transitions from initial to final states

K shell p-> s L shell p->s and p->dK shell p-> s L shell p->s and p->d

**

A Conventional EELS experiment in the TEM

Angular Resolution typically Angular Resolution typically 2-102-10 mRmR

Beam ConvergenceBeam Convergence2-10+2-10+ mR mR

0

0.2

0.4

0.6

0.8

1

1.2

270 280 290 300 310 320 330

.11

.35

.50

Rel

ativ

e In

tens

ity

Energy Loss (eV)

In some systems/conditions the difference isIn some systems/conditions the difference is subtlesubtleBut But ……. this also depends upon what your looking. this also depends upon what your looking for.for.

Variation of Amorphous Carbon NES with Collection Angle (ββ)

For Anisotropic Materials there can be large variations in the relative intensities of spectral featuresas a function of Angle (Momentum)

Variation of Graphite NES with Collection Angle (ββ)

0

1 104

2 104

3 104

4 104

5 104

6 104

7 104

260 280 300 320 340 360

A

0

2000

4000

6000

8000

1 104

1.2 104

260 280 300 320 340 360

A

B

0

1 104

2 104

3 104

4 104

5 104

260 280 300 320 340 360

A

0

5000

1 104

1.5 104

2 104

2.5 104

3 104

3.5 104

260 280 300 320 340 360

A

0

5000

1 104

1.5 104

2 104

2.5 104

260 280 300 320 340 360 0

200

400

600

800

1000

260 280 300 320 340 360A

0 0 mRmR 0.35 mr0.35 mr 0.7mr0.7mr

1.05 mR1.05 mR 1.73 mR1.73 mR 7 mR7 mR

θE ~1.15 mRα1/2 ~ 0.025 mR β1/2 ~ 0.1 mR

Variation of CK (π* σ*) in Graphite with Scattering Angle

Advanced TopicsAdvanced Topics

Low Loss SpectroscopyLow Loss SpectroscopyPlasmon Plasmon Losses StudiesLosses StudiesDielectric PropertiesDielectric Properties

Core Loss Spectroscopy Core Loss SpectroscopyNear Edge StructureNear Edge StructureExtended Fine StructureExtended Fine Structure

Radiation Damage Radiation Damage

StructureStructure in in EEL Spectra EEL Spectra is a manifestation of: is a manifestation of:Bonding Bonding (Chemistry/Physics(Chemistry/Physics) ) and/and/or Crystallographyor Crystallography ((Structure)Structure)

Low Loss RegimeLow Loss Regime

Optical, Dielectric,Optical, Dielectric,Electronic, MagneticElectronic, Magnetic

PropertiesProperties

Core Loss RegimeCore Loss Regime

Composition, Bonding,Composition, Bonding,Electronic, Magnetic ,Electronic, Magnetic ,Structural PropertiesStructural Properties

EELS Measurements of Valence Electron DensitiesEELS Measurements of Valence Electron Densities

Information in Core-Loss Profiles

ELNES

Copper L-shell Core Loss Spectroscopyin

Metallic, Oxide and High TcSuperconductor Phases

Magnetism in EELS is detectableusing the L shell transitions

p—>d

p —> s

• d electrons are dominant in determining magnetic properties• the L3/L2 "white lines" are the principle signals used to measure magnetism

x=0.2

x=0.3

x=0.4

x=0.5

x=0.6

x=1.0 Fe x Ge 1-x

Energy Loss (eV)640 680 720 760

AmorphousAmorphousFeFexxGeGe1-x1-x Magnetic Magnetic

MaterialsMaterials

700 705 710 715 720 725 730

FeL-RT-Norm

FeL-LN-Norm

Fe L

She

ll N

orm

aliz

ed In

tens

ity

Energy Loss (eV)

Summed Spectra 4 x 300 sec ~ 200KCnts at L3 Summed Spectra 4 x 300 sec ~ 200KCnts at L3 PkPkNormalized to LShell IntegralNormalized to LShell Integral

FeFe22OO3 3 (hematite)(hematite)

700 705 710 715 720 725 730

FeL-RT-Norm

FeL-RT0ffaxis-Norm

Fe L

She

ll No

rmal

ized

Inte

nsity

Energy Loss (eV)

Normalized to Normalized to LShell LShell IntegralIntegral

Momentum Resolved On-Axis vs Off-Axis

RutileRutile

Dc = 0Native State

Dc = 5Fully Amorphized State

Typical Specimenused in the EELS work

Anatase

Dc = 0

Anatase

Dc = 5

OKOKTiLTiL

Rutile Comparison Dc = 0 / 5