microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the...

34
Phase microscopy: Fresnel fringes in the TEM Interferences between waves not being in phase overfocussed Interferences between waves not being in phase 2 waves propagating through different media (sample, vacuum) => difference of mean inner potentials => difference of optical paths => phase differences Transformed into intensity contrast by the objective lens underfocussed in focus Phase TEM relies on coherency of incident beam (energy spread, point source) => FEG underfocussed in focus

Transcript of microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the...

Page 1: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Phase microscopy: Fresnel fringes in the TEM

Interferences between waves not being in phase

overfocussed

Interferences between waves not being in phase

2 waves propagating through different media (sample, vacuum)

=> difference of mean inner potentialsp

=> difference of optical paths

=> phase differences

Transformed into intensity contrast by the objective lens j

underfocussed in focus

Phase TEM relies on coherency of incident beam (energy spread, point source) => FEG

underfocussed in focus

Page 2: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Phase microscopy: High Resolution Electron Microscopy

Multi‐beam mode image: transmitted + several diffracted beams interfereseveral diffracted beams interfere => contrast = f (phase relationships of these beams)

i “ ” h lliFT

‐ image “represents” the crystalline structure of the specimen => structural characterization at the atomic scale

FT‐1

‐ phase differences result from:(i) interactions between the e‐beam and the electrostatic potential in the object ( i di l f t l)FT 1 (periodical for a crystal)(ii) phase shift induced by the objective lens (transfer function)

specimen thin enough => beam amplitude variations can be neglected (“phase object”)=> phase shift proportional to the electrostatic potential at every point of the exit surface

Page 3: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

High Resolution Electron Microscopy: WPO approx

• Transmission function of the sample:if small phase shift, Weak Phase Object (WPO) approximationq(x,y)= 1-iσΦ(x,y) Projected potentialq(x,y) 1 iσΦ(x,y)

• Image directly related to the projected potential of the object (atoms) after modification by the objective lens (filter)

Projected potential

y j ( )I(x,y)= |ψi(x,y)| 2 = |q (x, y)*t(x,y) |2

= 1+2σΦ(x,y) *s(x,y)

s(x,y)=TF{O (u, v)sin(χ(u, v)} is the imaginary part of t(x,y)

Projected potential (object)

Im. of t(x,y)(objective lens)

( ,y) F{ ( , ) (χ( , )} g y p ( ,y)

• best “reading” of the projected potential when‐largest possible frequency bandwith‐objective lens is underfocussed by Δf=-1.2(Csλ)1/2 (Scherzer condition)

Page 4: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

High Resolution Electron Microscopy

• In all practical cases, WPO approximation is not valid=> comparison HREM images/numerical simulation of the images for ≠ Df is necessary to access to the real structure of the sample (ex: where are the atomic columns?)

• However, although non linear, information regarding the atomic structure can be directly obtained in some cases: stacking sequences, extra plane, orientation relationship presence of dislocationsrelationship, presence of dislocations…

• Good experimental conditions: thin specimen (10 nm range)‐thin specimen (10 nm range)‐ incident beam // to a simple (low index) direction of the crystal => diffraction= known plane of the RS‐ column perfectly alignedcolumn perfectly aligned‐ appropriate defocus of the objective lens (Schertzer condition)‐microscope resolution smaller than the distance between the atomic columns that you want to image!g

Never trust ONE image !Never trust ONE image !

Page 5: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Rod‐like defect in silicon Si « dumbells »B=[011]

High Resolution Electron Microscopy

Rod like defect in silicon Si « dumbells »

[-1-11] [-11-1][-200][-11-

1][-1-11]

B [011]

[ ]

[1-11] [11-1][200]

[ ]

0.14 nm

[1-13]

[ 3 3 2][-111][1-11]

0.543 nm

with Cs corrector

I f i i d f h id ifi i d f

[-3 3 2] with Cs corrector

Cherkashin 2005Houdellier 2005

Interfaces, precipitates, defects, phase identification, defects, strain, etc…

Page 6: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

What else ?

Incident beam

specimen

2ΘDiffracted l t

Transmitted electrons

EELSelectrons Energy loss

electrons

Page 7: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

BCEV intensity jumps

Electron Energy Loss Spectroscopy

BC

BVEF

M

intensity jumpsionization energy levels

=> edges =core level distributions

ML

K

Im(-1/ε) = ε2 / (ε1 2 +ε2

2 )

Low-losses (the first few ten eV)Core-losses (up to a few thousand eV)

El m nt l ntit ti n l i

Valence electrons (VEELS)

Collective excitation : PlasmonIndividual excitation :Interband Transition

Core losses (up to a few thousand eV)

Inelastic interactions with inner or core level electrons

Elemental quantitative analysis

Electronic structureI~Im(-1/ε) = ε2 ∝ unoccupied DOS

Thickness: It/Izl=t/λpDielectric function

ε = ε1 + i ε2 I~Im(-1/ε) = ε2 ∝ unoccupied DOSEnergy Loss Near Edge fine Structure

ELNES

1 2Optical parametersn, k, μ, R = f (ε1, ε2)

Electronic structure JDOSε2 ∝ JDOS

Page 8: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Electron Energy Loss Spectroscopy

Page 9: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Electron Energy Loss Spectroscopy

. Electron Energy Loss Spectroscopy

Energy filtered TEMEnergy filtered TEM

Page 10: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

TEM/EELS

EELS adds one dimension to TEM imaging

B bi th it ti f l t b d d t th lid ith th l t fBy probing the excitations of electrons bounded to the solid with the electrons from the incident beam

Gives information on: ‐specimen thickness‐elemental chemical composition and spatial distribution of elements‐spatial distribution of first neighbors relative to an atomic site‐chemical bonding‐electronic band structure‐dielectric function

with an energy resolution ranging between 1 eV and 0.3 eV

with a spatial resolution ranging between 10 nm and 1 nmwith a spatial resolution ranging between 10 nm and 1 nm 

Page 11: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Concept of holography

ϕiAe=ΨComplex wave : amplitude and phase Take a picture :

²² AI =Ψ=You loose the phase         !!! ϕ

AeΨWe want to retrieve the phase !

Coherent photon wave

Object

Object wave

Reference wave

)2cos(2²²² 0 refobjectobjectrefobjectrefobjectref xRAAAAI ϕϕπ −+++=Ψ+Ψ=

Page 12: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Information in the phase

∫ dE )()(φ ∫= dzzxVE ),(C)x( Eφ

∫∫= dxdz)zx,(- )x( nBeM

EMT φφφ +=

Page 13: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Electron Holography in a TEM

Sample

)exp(o K.ri=ψ ))((exp)(s r.rr ss KiA ϕ+=ψ

Objective lens

Focal plane

V+

Image plane

V+

Projector lensesProjector lenses

IHolo=⏐Ψo Ψs*⏐

2)yx(2A= 1 +Holo ⏐ o s ⏐ )yx,(sA= 1 +

)]yx,(s.xR2cos[)yx,(sA2 0 ϕ+π+

Page 14: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

GPA analysis of the hologram

FT MaskFT

FT‐1FT

−π < ϕ < π

d

)yx,(sAImage d’amplitude

)yx,(sϕImage de phase

2π jumps unwrapped

Page 15: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Experimental aspects

• Coherent Electron Beam

Field Emission Gun

Fresnel fringesd t th hi hdue to the high coherence of the beam 

• Elliptic illumination to increase the spatial coherence biprism

meilleure cohérence

Page 16: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Reference hologram

(without object)

Hologram with the object

FT FT

20 nm

FT-1FT-1

siesAsϕ=Ι RRR

ieA ϕ=Ι

)s(ieA

sAs RRR

ϕ−ϕ=ΙΙ

20 nmDérivée de la phase

ϕΔ=ϕ−ϕ RsNR

AA

sA=Normalised Amplitude Phase shift due to the object only

Page 17: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Non magnetic sample (B = 0)

∫∫∫ −= dxdzzyxBedzzyxV )()(C)yx(ϕ

∫ dV )(C)(

∫∫∫ dxdzzyxBdzzyxV ),,(),,(C)yx,( nEh

ϕ

∫= dzzyxVelect ),,(C )yx,( Eϕ⎟⎟⎠

⎞⎜⎜⎝

⎛++

⎟⎠⎞

⎜⎝⎛=

)2EE(EEE

λπ2C :with

0

0E

V( ) V ( ) V ( )V(x,y,z) = Vi (x,y,z) + V (x,y,z)

• Vi (x,y,z) = Mean Inner Potential (MIP)

V • V (x,y,z) = Local electric potential (junctions, CMOS, diode ..)

• if Vi et V are constants along the e-beam direction « z » :

ϕelect = CE (Vi + V ) t t = sample thickness

Need to perfectly know the sample thickness (sphere, cleaved sample, CBED hi k ) if V d Vthickness measurement...) to quantify Vi and V

Page 18: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

ϕelect = CE Vi tMean Inner Potential (MIP) measurement

F NiFe0.5Ni0.5

10 nm

10 nm

Page 19: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

10Fe Ni 910nmFe0.3Ni0.7

6

7

8

s FeNi theory

3

4

5

6

Pha

se/ra

dian

s yFeNi exp

tVEc 0=φ1

2

3P tVEc 0φ

00 5 10 15 20 25 30 35 40 45 50 55 60

Position/nm

• MIP measurement

• Composition analysis

• Morphological studies

Page 20: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

∫∫∫ −=ϕ dxdz)z,x(Bedz)z,x(VC )x( nEh

Magnetic samples

Pb. 1: The objective must be witched off. => use of Lorentz lens to get sufficient resolution

• Object in a magnetic field free area

• Field of view up to 1µm 

f = 30 mm

Cs = 7430 mm (!!)Cs = 7430 mm (!!)

Cc = 39 mm

Point resolution # 2 nmPoint resolution # 2 nm

Page 21: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Pb. 2: How to separate the magnetic and electrostatic contributions  B and CEVit ?

Magnetic samples

Phase shift induced bt the local electrostatic potential:potential:

∫= dzzxVelect ),(C )x( Eϕ

Phase shift induced by the local magnetic 

y gfield :

∫∫ d d)()( Be∫∫= dxdz)zx,(-)x( nBmag

Total phase shift

)x()x()x( ϕϕϕ += )x()x()x( magelectTot ϕϕϕ +=

Page 22: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Pb. 2: How to separate the magnetic and electrostatic contributions  B and CEVit ?

1/ Taking two holograms (4 with the reference holograms) switching upside down the sample => The sign of the magnetic contribution is reversed and the electrostatic one remains. 

VBt Vi BVi

∫∫−= dsBetVi nE .C 1 h

ϕ ∫∫+= dsBetVi nE2 .C h

ϕh

∫∫= dsBen2-

12 hϕϕ tVC iE2

12=+ϕϕ

Problem : finding back the same area of interest

2/ Taking two holograms with two different high voltages⎟⎟⎞

⎜⎜⎛ +

⎟⎞

⎜⎛= 0EEπ2

EC2/ Taking two holograms with two different high voltages

∫∫−= dsBetVi nE .C 11 h

ϕ)(

- 12

CCtVi =

ϕϕ

⎟⎟⎠

⎜⎜⎝ +

⎟⎠

⎜⎝

=)02EE(EλEC

)(12

EE CC −

P bl k i th i iti diti ( ifi ti

∫∫−= dsBetVi nE .C 22 h

ϕ

Problem : keeping the same acquisition conditions (magnification, alignment…) when changing the high voltage. 

Page 23: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

3/ Using the high magnetic field of the objective (~2 Tesla) to saturate the magnetic sample in two opposite directionstwo opposite directions

HObjectiveHObjectiveHObjective

+Θ -Θ

Back to Θ = 0 – objective switch off

∫−=ϕ dx).x(B.t.et.VC niE1h ∫+=ϕ dx).x(B.t.et.VC niE2

h

t

-d)(Bte 12 ϕϕ∫ tVC 12 ϕ+ϕ

∫h ∫h

2dx).x(B.t.e 12n

ϕϕ=∫h 2tVC iEϕϕ=

Magnetic contribution to the  Electrostatic contribution to the phase shift phase shift

Page 24: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

x1

∫∫=

−=−=Δ dzdzyBeyxyx nx

ξξϕϕϕξ

),,(),(),(1

2

12h

txBexx

n ).()(h

=∂

∂ϕ • The gradient of the phase is proportional to the in‐plane component of the magnetic induction Bn

• The equiphase contour gives the direction of the magnetic induction B

Page 25: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Magnetic Configuration of an isolated Fe Nanocube42

982)

, pp

4293

–4t.,

200

8, 8

(12

al.,

Nan

o Le

ttS

noec

ket

a

(a) TEM image, (b) Magnetic contribution to the phase shift, (c) Magnetic induction mapping, (d) micromagnetic simulation (OOMMF)

Page 26: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Magnetic Configuration of two neighbouring Fe Nanocubes

(a) TEM image, (b) Magnetic contribution to the phase shift, (c) Magnetic induction ( ) g , ( ) g p , ( ) gmapping, (d) micromagnetic simulation (OOMMF)

Page 27: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Magnetic Configuration of four neighbouring Fe Nanocubes

(a) TEM image, (b) Magnetic contribution to the phase shift, (c) Magnetic induction mapping, (d) micromagnetic simulation (OOMMF)

Page 28: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

φ(x) = e/ħ ∫ B┴(x) t(x) dx

Magnetic multilayers

φ ∫

dφ(x)/dx = (e / ħ) . t . B┴(x) 

E. Snoeck, P. Baules, G. BenAssayag, C. Tiusan, F. Greullet, M. Hehn, A. Schuhl

J. Phys.: Condens. Matter 20 (2008) 055219.

Page 29: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Sample preparation

Cross Section

specimen 1 t t

glue

specimen 1st cut

Ar+ ionsspecimen

electrons

2nd cutReady for TEMMechanical grinding

(abrasive papers + dimpler)(abrasive  papers   dimpler)

‐> 10‐20 µm

+ ion milling ‐> hole

Page 30: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Sample preparation

Plan view

electronsAr+ ions

20 µm

specimen mechanical grinding+ dimpling

ion milling ‐> hole Ready for TEM

Page 31: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Sample preparation

Sample preparation

Sample preparation

Page 32: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Sample preparation: FIB

Page 33: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Sample preparation: FIB

Page 34: microscopy: Fresnel fringes in the TEMregpot/winterschool2009/... · •best “reading”of the projected potentialwhen ... Fresnel fringes due to the hi hhigh coherence of the beam

Summary

d l fGood TEM results from

50% specimen preparation, 

20% TEM’s price

30% ’ i30% guy’s experience…