Overview of Scanning Electron Microscope,Transmission Electron Microscope,

Scanning Transmission Electron Microscope,Low Energy Electron Microscope

P.E. Batson

with help from K.A. Mkhoyan, U. Minnesota

• Why electrons rather than light?• What physical processes do we use for obtaining image contrast?• How are the instruments above related?• What are the results like?

KA Mkhoyan

rest mass kinetic energy

Total Energy

Wavelength of light -> 3800 – 7500 Å -> not very interesting for materials characterizationwhere regions of materials are less than 10 nanometer (100 Å) or smaller,

and atomic distances, where much new physics occurs, are smaller than 1 nanometer.

KA Mkhoyan

1

0.1V -> 38 Å1.0V -> 12 Å100V -> 1.2 Å1kV -> 0.38 Å10kV -> 0.12 Å

Now, what happens when electrons interact with the specimen and howdo we collect them?

Large angle, relatively chaotic processes

Small angle, relatively well behaved processes

Image plane, or detector

Source

Optics

Apertures optimized for signal

http://www.purdue.edu/rem/rs/sem.htm#2

SEM System islargely a probe formingoptical system that imagesby scanning and collectingserial, time varyingsignals that are fed to a TV-like display

http://www.tasc.infm.it/research/tem/images/semimage.jpg

Electron Column

Gun

Optics

Specimen

Pumps

Knobs and displays

Detectors

Computers!

The ORION™ Helium ion Microscope from Carl Zeiss SMTCurrently one of these is being installed at Rutgers:

Torgny Gustaffson - Len Feldman

50μm

Anthophyllite asbestos

Size of fibers – about 50μm

Alveoli Size 140 μmCell Size 10-100 μm

Result: Cancer

http://usgsprobe.cr.usgs.gov/picts2.html

Topographical Contrast

ProbeForming

The first 50 Years of Electorn Microscopy

The first Next 50 Years of Electorn Microscopy

7’6”

Yao Ming

Many signals, one microscope: Nion UltraSTEM™

Aberration corrector 1

Aberration corrector 2

Described in: Krivanek et al. Ultramicroscopy 108 (2008) 179-195 and Dellby et al. EPJAP 2011. More info at www.nion.com.

UltraSTEM200* UltraSTEM200*

*instrument shown:CNRS Orsay, France*instrument shown:CNRS Orsay, France

UltraSTEM200*

*instrument shown:CNRS Orsay, France

Fully modular (all lenses, the corrector, etc., are independent modules, with identical mechanical interfaces) and thus very flexible.

Ultra-stable, friction-free sample stage.

Operating voltage range 20-200 kV.

Efficiently coupled EELS.

UHV at the sample (<10-9 torr; <10-7 Pa).

http://www.specs.de/cms/upload/PDFs/SPECS_Prospekte/LEEM.pdf and IBM Corp (Ruud Tromp)

Probe forming

Image forming

Low Energy Electron Microscopy

http://www.research.ibm.com/leem/

2nm resolution today

Deceleration

http://www.specs.de/cms/front_content.php?idcat=209

Electron Column

Gun

Optics

Specimen

Pumps

Knobs and displays

DetectorsComputers!

Specimen

Source

SEM TEM/STEM LEEMEnergy 2-10 KeV 50-200 keV 0-10 eVProbe Size 1-10 nm 0.1-1 nm 2-5 nm

Probe FormingOptics

Imaging Optics

Variations on a Theme!

Energy width -> 0.8 – 3 eV 0.25-0.40 eV

++--

Tungsten thermionic total current -> 0.1mA/srField Emission total current -> 1.0 μA/sr

so where small probe is not needed, tungsten wins.

Cylindrical Electron Lens

Lorentz ForceF =(-e/c) v x B

so Fθ = (-e/c) voBr

producing a spiral: vθ

finally

Fr = (-e/c)vθBz

so that electron beam is deflectedtowards axis

Br

Bz

V

Cylindrical Lens Focusing

Specimen

Source

SEM TEM/STEM LEEMEnergy 2-10 KeV 50-200 keV 0-10 eVProbe Size 1-10 nm 0.1-1 nm 2-5 nm

Probe FormingOptics

Imaging Optics

Reiterating variations on a theme.

Now, what happens when electrons interact with the specimen and howdo we collect them?

First: THICK Specimens

Large angle, relatively chaotic processes

Small angle, relatively well behaved processes

Image plane

Source

Optics

Apertures optimized for signal

With thick specimens, most of the signal comes out the front side.

0 Eo – Incident beam Energy

Primary Imaging Mechanism for the SEM and HIM: Secondary and Backscattered Electrons

Fred Cosandey

http://upload.wikimedia.org/wikipedia/commons/8/82/SEM_blood_cells.jpg

Why does Secondary Electron Imaging look like such a “natural” rendering of the specimen?

Secondary electron image of red blood cells shows surface topography in a very easilyunderstood image.

Unlit areas in the common situation on the right are dark in the scanning electron case. The image looks “natural” because it matches a situation we encounter nearly every moment of the day.

Important result: If we reverse the ray paths, and exchange the source and detectorswe will get the same contrast.

eye

Surface Surface

LightSource

ElectronSource

detector

ShadowShadow

Illustrates a general principle: Reciprocity. If we interchange the source and detector and reverse the particle paths, we get identical results.

http://upload.wikimedia.org/wikipedia/en/0/00/SEM_SE_vs_BE_Zr_Al.png

Secondaries BackscatteredBackscattered electrons are higher energies:penetrate deeper, see less surface contrast

Fred Cosandey

Suppose your specimen thickness is much less than a few times λ ?

Then the large plume of scattered electrons below the surface does not exist !!

λ

The secondary electron resolution can then be as good as the probe size.

Zhu, Y., Inada, H., Nakamura, K. & Wall, J. Nature Mater. 8, 808–812 (2009).

Secondary Electron Image using Hitachi STEM at Brookhaven

Annular DarkField

Backsideparticle

Backsideparticle

Zhu, Y., Inada, H., Nakamura, K. & Wall, J. Nature Mater. 8, 808–812 (2009).

Secondary Electron Image using Hitachi STEM at Brookhaven

Atomic resolution with secondary (and backscattering) electrons

Backside atomvisible in AnnularDark Field

Backside atommissing in secondaries

Now, what happens when electrons produce xrays?

Large angle, relatively chaotic processes

Small angle, relatively well behaved processes

Image plane

Source

Optics

Apertures optimized for signal

Background due to Bremstrahlung (deceleration) radiation from beam electrons

Light elements

Typical Energy Dispersive X-Ray Spectrum

Multiple linesfrom singleelement

Apollo SDD - Silicon Drift Detector

Older design required liquid nitrogen for high performance, because the charge amplifierwas located away from the Si detector. This particular picture is actually of a new detector designed for performance under hard radiation.

In the new design, the charge amplifier is located on the silicon detector, lowering noise and allowingoperation using only a Peltier cooler. No liquid nitrogen is needed.

These detectors work by collecting electron hole pairs generated by incoming x-ray photons. The number of carriers is proportional to the energy ofthe x-ray, so a pulse is generated for each x-raywith a height proportional to the x-ray energy.A preamp and pulse height analyzer producesthe spectrum.

http://gsc.nrcan.gc.ca/labs/ebeam/images/sem6.jpg

We can use a single peak in the EDX spectrum

To produce maps for different elements.Here is a result for a natural mineral.

This can also be a high resolutiontechnique if the specimen is made verythin.

Specimen

Source

SEM TEM/STEM LEEMEnergy 2-10 KeV 50-200 keV 0-10 eVProbe Size 1-10 nm 0.1-1 nm 2-5 nm

Probe FormingOptics

Now look at theImaging Optics

Now, what happens when electrons interact with the specimen and howdo we collect them?

Large angle, relatively chaotic processes

Small angle, relatively well behaved processes

Image plane

Source

Optics

ConjugatePlanesConjugate

Planes

What changes hereis the intermediatelens

Diffraction Imaging

Crystallites in the material which are slightlyrotated.

Patterson Function Inversion Model

Recombined to form interferenceFred Cosandey

Reciprocal lattice RODS Making Satellitespots

Imaging a small step at the surface

Fred Cosandey

Spots for types a,b,c layers can be apertured to reveal thedomain behavior of that surface structure

Imaging using surface satellite spots

Fred Cosandey

Small Angle elastic scattering makesinterference patterns

Specimen

Source

SEM TEM/STEM LEEMEnergy 2-10 KeV 50-200 keV 0-10 eVProbe Size 1-10 nm 0.1-1 nm 2-5 nm

Probe FormingOptics

Imaging OpticsNow moving to the LEEM

http://www.research.ibm.com/leem/

The LEEM does typicalTEM things, but using <10eV electrons at the specimensurface. Here is an example of Dark Field Imaging

Si 2x1 reconstruction has 2 orientationsGreen Spot Red Spot

Scanning Transmission Electron Microscopy

What’s different: Strong dependence on

Analytical Information

Two types of signals:Elastic vs Inelastic Scattering Large angle,

relatively chaotic processes

Small angle, relatively well behaved processes

Source

Optics

Electron Energy Loss Spectroscopy

Annular Dark Field STEM

Atomic structure, EELS in nanoscale devicesBonding = 2 - 5eV changesConduction bands = 0.2-1eV changesStrain = 0.05-0.25eV changes P.E. Batson, Jan. 2004

Electron Energy Loss Spectroscopy

Bright Field (BF) and Annular Dark Field (ADF) Imaging

Small Angle Phase Contrast Large Angle Scattering Rutherfordand Thermal Diffuse Scattering

BF-ADF pair using same defocus – optimized for ADF

Interference Patterns Positive Definite Guassian Images

Aberration Correction!! Silicon [110]

Before After

120kV 25 year old instrument

SJ Pennycook, M. Chisholm, A Lupini, M Varela, K van Benthem, A Borisevich, M Oxley, W Luo, and S Pantelides, In Advances inImaging and Electron Physics - Aberration-Corrected Electron Microscopy, 153 327-384. Elsevier (2008).

The number of accessible latticereflections increases from 2-3 at 2 Angstromto as many as a dozenat 0.8 Angstroms

GaAs Example fromAR Lupini

[111]

[211]

[321][110]

Scherzer Defocusto balanceSpherical Aberrationwith Defocus

CTEM: Bright Field Phase Contrast

Jia, C.L. & Urban, K. (2004). Atomic-resolution measurement of oxygen concentration in oxide materials. Science 303, 2001.

Uniformity ofcontrast is resultof high pass filter

Aberration Corrected

Defocus delocalization

CTEM: Bright Field Phase Contrast

Now, we look at EELS analytical signalsusing transmitted electrons.

Large angle, relatively chaotic processes

Small angle, relatively well behaved processes

Source

Optics

Electron Energy Loss SpectroscopyAnnular Dark Field STEM

Gatan GIF/Enfina/Tridiem/Quantum/HREELS

Enfina:

CCD Array

HREELS:

HREELS: 60-250meV, depending on exposure timeHigh throughput P.E. Batson, Jan. 2004

Multipole correction at entrance

x500

x5000

Typical results for cold field emission system – 0.3 eV. Intensity at the Si L23 edge is larger than that in the direct interband region. P.E. Batson, Jan. 2004

Core Loss: composition,chemistry, electronic structure

Optical Excitations

Direct Band Gap HARD!!

Gatan/Phillips Tecnai Performance

Si L23 edge at ~100eV2 nm probe

Zandbergen Lab in Delft

These are very exciting results, because they are obtained from commercial equipment.In the past all high resolution results were on home built machines.

P.E. Batson, Jan. 2004

Processing required to convert measured data

into 2p3/2 intensity for comparison with theory :

1) Background subtraction2) Deconvolution 3) Removal of 2p½ part

Allows reproducible comparisons

Si L2,3

P.E. Batson, Jan. 2004

Using DOS defined

A(E-Eth)1/2(Emax-E)

And a scattering theory That includes the

core exciton,We can fit the data

In terms of the model DOS

P.E. Batson, J. Electron Microscopy 45, 51–58 (1996).P.E. Batson, Jan. 2004

ELNES Si L3 Interpretation

Dipole selection rules are important

Features in EELS corespond to Regions of high DOS

Brillouin Zone Center is mostly p-like and so is dipole forbidden and does not show strongly in data

P.E. Batson, Jan. 2004

GeSi Alloy SeriesMost conduction band BZ special points canbe followed as we go from Si to Ge

P.E. Batson, Atomic Resolution Electronic Structure in Silicon-Germanium Alloys, J. Electron Microscopy 45, 51-58 (1996).

P.E. Batson, Jan. 2004

We can use thisCB information tofollow band offsetin special cases:

In GeSi alloys, theSi core loss is constant and is usedas a zero in energyto get CB position.

ħ γ = ħ vf / R

Damping follows the form:

where vf is the “Fermi” velocity for Sigives vf = 2.05x108 cm/s

Size LimitedConductivity

Drude form for the Conductivity:

σ = n e2 τ / m

with τ= 1/γ

Providing a link between theLocal conductivity and

damping parameter.P.E. Batson, Jan. 2004

J. Nelayah, et al., "Mapping surface plasmons on a single metallic nanoparticle," Nature Physics 3, 348-353 (2007).

Mapping Surface Plasmon Scattering Using EELS Signals

This is the supporting materialfor the result of the next slide.

Spectra were acquired from 400-850 eV.

Ti L23, Mn L23 and La M45 absorptionedges were acquired for each pixel of the area shown in (B).

The Ti L23 shows pronounced finestructure changes as you go into theLaxSr(1-x)MnO3 from the SrTiO3 .

ADF imaging shows disorder at interfaces,

FFT of image shows 0.98A spot.

Quantification for elements, result in the mapson the previous sklide.

Example of Core Loss Imaging:EELS atomic-resolution chemical mapping

D. A. Muller, L. Fitting Kourkoutis, M. Murfitt, J. H. Song, H. Y. Hwang, J. Silcox, N. Dellby, O. L. Krivanek, Science 319, 1073-1076 (2008).

Example of Core Loss Imaging:EELS atomic-resolution chemical mapping

Pulsed laser-depositedLa0.7Sr0.3MnO3/SrTiO3multilayer

40 mr illum. half-angle0.7 nA beam current~1.2 Å probe>70% eff. EELS coupling

64x64 pixel map7 msec per pixel, i.e. 29 sec total acquisition time 10 sec additional processing time

i.e., <1 min total time5 Å

D. A. Muller, L. Fitting Kourkoutis, M. Murfitt, J. H. Song, H. Y. Hwang, J. Silcox, N. Dellby, O. L. Krivanek, Science 319, 1073-1076 (2008).

Ti (L)La (M)

RGBMn (L)

Kazu Suenaga & Masanori Koshino, Nature, 2010.

MAADF imaging of single layer BN with impurities

60 keV MAADF image, 6 x 106 e- / Å2, with probe tails and high spatial frequency noise removed.Image courtesy Matt Chisholm, ORNL, image processing by olk.B and N atoms are readily identifiable by their MAADF intensities.C and O substitutional impurities are identifiable in the line profiles.

O.L. Krivanek, M.F. Chisholm et. al., Nature 464 (2010 ) 571-574.

Histogram analysis of MAADF image (in which probe tails have been removed)shows that B, C, N and O can be identified unambiguously in monolayer BN.

The experimentally worked out dependence of image intensity on Z goes as Z1.64.

BN monolayer with impurities: histogram analysis

Result of DFT calculation overlaid on an experimental image

Cx6

Na adatom

O

N

Longer bonds

C ring is deformed

B C

C

O

MAADF imaging of single layer BN with impurities