Electron Energy-loss Spectroscopy

and Energy Dispersive X-Ray Analysis

 Nano-Analysis
Low energy, 100-1000 eV

Techniques
• Inner-shell ionization – an emission of
characteristic x-ray quanta or Auger
electrons
• Electron energy-loss spectroscopy
– The electronic structure Resolution
– The elemental composition The volume
– Efficient for low-Z elements irradiated by the
electron

The volume from Nanodimension

which the stimulated
signal arises
 Microanalysis in EM
Technique Input Output Depth Lateral Elemental Sample Detection
Signal Signal Sampled Resolution Range Limit (at%)
(nm) (nm)
Wavelength electron Photon (X- ~ 1000 nm ~ 1000 nm Z4 Flat polished 0.005% (50
dispersive X-ray Ray) surface ppm)
spectroscopy
(WDX, WDS)
Energy dispersive Electron Photon (X- ~ 1000 nm ~ 1000 nm Z5 Flat polished 0.1%
X-ray spectroscopy ray) surface
in SEM (EDX, EDS)

Energy dispersive Electron Photon (X- ~ 10 -100 nm ~ 2 nm Z5 TEM foil 0.1%
X-ray spectroscopy ray) (Sample
in TEM (EDX, EDS) thickness) Accuracy: 5%

Electron energy Electron electron ~ 10 -100 nm ~ 1 nm Z3 TEM foil 0.1%

loss spectroscopy (Sample
in TEM (EDX, EDS) thickness)

Auger electron Electron electron 0.5-2 nm 400 nm Z2 Surface 0.1%

spectroscopy (AES)
 Inelastic scattering of electrons
Through smaller angles than for elastic scattering
The cross section varies linearly with atomic number

5-30eV, ~ 100 nm,

dominant

< 1 eV; ~ 10o, ~ 1um, heat

50eV – keV: EDX, EELS
Interaction volume and information
volume

Limited range

20 kV, Cu
Generation - escape

Depends on materials and beam energy,

Kanaya-Okayama range (empirical)

0.0276 AE 01.67
R
Z 0.889 
 X-Ray Generation
Beam current

• Bremsstarhlung: electrons being decelerated by the nucleus and iZ E0  E 

I
electrons in the atoms in the materials - not useful, background E

• Characteristic X-rays: The energies/wavelengths of the Maximum energy is

the beam energy
emitted X-rays

Low energy X-rays

absorbed by the
specimen
 Characteristic X-rays
• Sharp and element-dependent: Useful for analysis

The allowed transitions: Moseley’s Relation:

n 1; l=1; j= 1 or 0 • X-Ray wavelength:
B

Angular Z  C 2
momentum
• X-Ray Energy:
hcZ  C 
2
E
eB

B, C = constants
depending on the
X-ray line

Transition to

K An outer shell electron falling

into the vacancy (~ 10-12 s)
Transition is from..

Remove an electron from an inner shell: vacancy

 EDX and WDX
EDX: WDX:
• Detector: pn junction – e-h •Use crystal monochromators to disperse
generation ~ the energy of the X-ray the emitted X-ray spectrum in terms of
photon Bragg diffraction angle and hence
• Fast electronics: separate the pulses wavelength
•Detectors collect X-rays in a near- • Slow to acquire (30 min)
parallel fashion (1 min) • Increased spectral resolution: 10 eV
• Spectral resolution: 100-150 eV • Sensitivity to all elements (especially
• Common attachment light elements)
• Confined to dedicated analytical SEMS
(Electron Probe Microanalysers, EPMAs)
EDX: poor energy resolution • Detection limit: ~ 0.005%
• Peak overlap
• no chemical-state information can
be extractive (< 10 eV)

Florescent yield

X-ray emission is very inefficient for low Z (and

absorption), EEELs is more efficient technique for
analysing low Z elements
 EDX
Cooled by liquid nitrogen to
• Reduce thermal e-h pair
• Prevent the Li atoms from diffusing,
• Reduce noise in the FET preamplifier

Collect e-h pair

Charge is not completely detected

Detector:
• reversed p-i-n (p-type, intrinsic, n-type)
• e-h generation
3.8 eV is required to form an e-h (not all energy creats e-h holes)
A Cu Ka, 8040/3.8=2300 e-h holes ~ 10-16C – small signal – preamplifier (FET)
•Be(7-12um) > Na
•Ultrathin (<100nm polymer) > C
•Windowless (in vacuum)> Boron
 Interpretation of X-ray Spectra
Experiment
• Beam energy > 2 X highest peak energy
• Reproducible spectrum from the same
area
• Counts: enough to recognize peaks; not
too high (<3000 counts/sec; dead time<
50%) to minimize sum and escape peaks

Analysis A Si K X-ray escape from Two X-rays arrive at the same

the Si detector time
• Prior knowledge of the sample: likely elements
• Be aware of stray irradiation peaks: Fe Cr Ni Cu Zn Al Pt Mo
• Confirm elements by looking for other peaks from that
elements (KLM)
• Work from high energy to low energy identifying peaks (at
high energy, few peaks and better resolution of neighbouring
peaks)
• Peak shapes and energies (0-20 kV)
•K series – B (Z=4) to Ru (Z=44)
• Z > 16(S) look for K;
• L series – Cl(Z-17) up
• Z> 42(Mo) look for L
• M series - ~ Ag(Z=47) up
• Peak overlaps:
• S K 2.31keV Mo L 2.29 keV Pb M 2.35
keV
• N K 0.39 keV Ti L 0.45 keV
 EELS

• Compositional analysis: efficient for light

elements
• Chemical analysis: shapes of energy loss edges
depend on local bonding and oxidation state
 EELS Spectrometer
Typical energy loss: < 1 kV, a spectrometer can separate this.

Curved faces of the magnet:

focus the electrons

Bends the electrons through 90o

with a magnetic prism.
Inelastic scattered
electrons: Small Parallel detection: a diode array with
scattered angle < 1o 1024 or more diodes to collect the
whole spectrum in parallel
Additional quadrupole and sextupole
lenses: fine tune the focus
 Typical Energy Loss spectra
Zero Loss: Unscattered, • Width of Zero peak: energy spread of the microscope: 0.3 eV (FEG) YBa2Cu3O7
elasticlly, phonon-
scattered electrons
• Phonon: thermal diffuse scattering: < 1eV; ~ 50 mrad Log Scale
Plasmon Losses: • Plasmon:
low-loss (5-50 eV) •oscillation of conduction band electrons
•Characteristic plasmon energy
• A few mrad
• Thick sample: multiples of the plasmon energy
 t 
I p  I 0 exp   
  
 p 

Core Losses •Excitation of a core level electrons

• Edge: excited to the lowest empty state or higher energies
• Each shell has its own edge
•It can be obtained from a much smaller volume than EDX

Single electron
excitation: background
~ 30 eV up - SEs

100 kV Electron Energy 99 kV

 Quantifying EELS
• Thin specimen
– to avoid multiple scattering which obscures the edges
– the Plasmon peak < 1/10 the zero loss
• Edge: the intensity is spread over large range of loss rather than in a sharp
peak… the edge area is difficult to measure… fit the background above the
edge
– Fit the background using I = AE-r above the edge and extrapolate this
background under the edge
– Subtract the background from the edge
– Calculate the edge area within a window: C
– Calculate the ionisation cross-section  and integrate over same window
(collection angle-the objective aperture, microscope voltage) : not accurate ~
10%
– Repeat for other edges and the compositions
CA
XA A

XB C B
A
 EDX and EELS

Techniques Signal to noise Low Z Detection limit Spatial Artefacts Compositional

resolution accuracy
EDX high Boron (5) 0.1 at% 2 nm Escape, stray, Better < 1%
sum

EELS low Better- Li and 0.05 at% (for Better (1 nm) No stray etc. Poor 10%
He (2) some
elements)
 ELNES and EXELFS
• ELNES: Energy loss near edge
structure Al K, from Ni3Al, background subtracted
– The probability of the electron
ending up at a particular energy
level in the conduction band
depends on the conduction
band density of states (DOS)
– Due to the selection rule, the
near edge structure doesn’t
show the full DOS
– DOS depends on the bonding
state of the atom
• EXELFS: Extended energy loss fine
structure
– 50-200 eV beyond the edge
– The structure depends on local
arrangement of atoms around
the excited atom
– To get nearest neighbour
distances and radial distribution
functions
– Low noise spectra are needed
 EFTEM
• Energy-filtered Microscopy
Normal EFTEM:
– Collect an image from a given energy loss slice // xy plane,
– Two ways: low energy
• STEM: focused probe; collect an energy loss resolution
spectrum through a slit at the desired energy
loss; scan to create an image
– collecting spectrum at each image point
– Useful for the detection of low concentrations of
an element
• TEM: Use extra lenses after the energy selecting
slit to allow the original image to be reformed STEM Spectrum
(GIF, Gatan imaging filter) Imaging: sample drift
– collecting an image at each energy loss and distortion
– Applications
• Zero loss filtering
– Allow thick regions to be examined
t  I unf   I 
• Thickness determination: t/ map  ln   ln1  loss 
– Thickness map: a zero loss image (I0) and an   I0   I0 
unfiltered image (Iunf) or a loss image (Iloss),  :
total inelastic mean free path …. Non-uniform
composition?!
• Core-loss mapping: Elemental map
– Jump-ratio image: dividing the iamge beyond the
edge by an image below the edge – thickness
variations are removed (at least approximately)
– 3 Window method (can be quantified): 2 pre-
edge and one post-edge, the two pre-edge images
are used to estimate the background under the
post-edge image and the background can be
subtracted … the thickness variations!
– Image spectroscopy: a series of images is Si 110 CBED at 100 kV
collected both before and after the edge of
interest – to get a better background (longer time
larger dose)
Appendix
 Inelastic Scattering
• Electron-Specimen Interactions with Energy
Loss
• Differential Cross Section for Single-Electron
Excitation
• Bethe Surface and Compton Scattering
• Approximation for the Total Inelastic Cross
Section
 Electron-Specimen Interactions with Energy Loss
• Elastic preserves: Kinetic energy and momentum
• Inelastic conserves total energy and momentum
– Energy conversion: atom-electron excitation
– Energy loss: the primary beam
Excitation Mechnisms:
• Oscillation in molecules and phonon excitations in solids:
• E ~ 20meV-1 eV
• Monochromator
• Low beam intensity – low spatial resolution
• Infrared
• Intra- and interband excitation of the outer atomic electrons/ plasmons
• Broad maxima: E ~ 3-25 eV
• Concentration of C/V band electrons; chemical bonds, band structure
• Visible and ultraviolet
• Ionization of core electrons
• Edge E
• Spectrum ~ eV beyond E
• low loss – less localized ( excite atom from ~ 10 nm – small angle)
• For low atomic number: total inelastic cross section > total elastic cross section
• Thick specimens: chromatic aberration due to energy loss
• Energy - heat
 Differential Cross Section for Single-
electron Excitation
• Energy Transferred
– E – Ei→Ef
– Selection rules: l=1
– Scattering vector: q '  k f  ki

• Conservation of momentum, energy k n2  k02  q'2 2k0 q' cos

E 
2 2
2m

k0  k n2 
2
2m

2k0 q' cos  q'2 
2
2m
2k0 q' cos 
 2 k0 q'
m

cos

d i  f m2 k f
 f V r  i    
Inelastic cross section: 
2
     
d  4 ki  i  ais (rj ) exp iki  ri  f  a fs (rj ) exp ik f  ri
d i  f  me 2   (q' )
   
2
     
  exp  ik f  ri a*fs (rj )V ri , rj ais (rj ) exp iki  ri d 3ri d 3rj
 
 f V r  i
2 2
  
2 
d  2 0   q'
4

 (q' )   a*fs (rj ) exp  iq 'rj ais (rj )d 3rj   a*fs (rj )1  iq 'rj  ...ais (rj )d 3rj  q '2
2      2      2    2  2
a fs u  rj a is  q '2 xif
2 2 2 2
d i  f me 2   (q' )
2 2 2
 me 2  q ' xif  me 2  xif
  
2 
  
2 
  
2 
q'2  k0   q' cos   k02  2   E2
2 2
 
d  2 0  q'  2 0   2 0  q'
4
q '4 2

2 2 2 2 2
d i  f
2 2 2 2
 me 2  xif  me 2   2  xif 2  1 xif    xif
2
xif 
     2  2 2    2   2 2    2
d 2 
 2 0  q'
2 2  2 2
 2

 4 0  k0    E  aH k0     E  2  aH    E  aH   E
2

q' cos 2
2 2 2 2
 q' cos   mE   mE   E 
 2
E       2 2    2    
k02  k0    k0   p0   2 E0 
 Differential Cross Section for Single-
electron Excitation
• Generalized Oscillator strength (GOS)
 2mE   (q' )
2
 2mE 
f if q'  
2
   xif
   q'   
2 2 2

f if q' f if q' f if q' f if q'

2
d i  f
2 2
   xif    2 2 4 4 e4
   2     2 2  
d  aH     E  aH
2
 
2

  2mE   2   2  aH 2mE    E
2
 2 2 2

p aH 2mE    E 2
 
(4 0 ) EE  2   E2
2

E
  
2

• Generalized Oscillator strength (GOS) per unit energy loss

d i  f e4 1 df if q' , E 

ddE 
(4 0 ) 2 EE  2   E2  dE

• Bethe Surface
df if q' , E 
dE

• Bethe ridge – Compton scattering