Short description Beam parameters
influence on image
e - gun components
filaments lenses
Beam-sample interaction
electron scattering
Image formation
Scanning Electron Microscopy (SEM)
Goldstein, Scanning Electron Microscopy and X-ray Microanalysis
Scanning Electron Microscope (SEM)
V-shaped Filament Extractor
Deflecting Plates
Backscattered Electrons
e
-Detector Primary
e
-Beam
Sample
Image Display
Electron Column
Field of view:
5x 5 mm
2– 500 x 500 nm
2Resolution: down to 1 nm
Scan quadrupole
Beam accelerator
Optical axis
How to sweep an electron beam
First coil deviate beam from optical axis
Second coil brings beam back at optical axis on the pivot point
Image formation point by point
collecting signal at each raster point
L = raster length on sample W = working distance
S = raster length on screen Magnification = S/L
L
S
L = 10 m, S = 10 cm X
m
x m
M x 10000
10 10
10 10
6
2
M depends on working distance
Effect of beam parameters on image
V
0= beam voltage i
p= beam current
p= beam convergence angle
d
p= beam diameter at sample
High resolution mode Noise on signal
Effect of beam parameters on image
i
p= 1 pA, d
p= 15 nm
i
p= 320 pA, d
p= 130 nm
High current mode Resolution too low
i
p= 5 pA, d
p= 20 nm Good compromise
i
p= beam current d
p= beam diameter
Resolution
Depth of focus
Effect of beam parameters on image
If
pis small, d
pchanges little with depth, so features at different heights can be in focus
p= 15 mrad
p= 1 mrad
Effect of beam parameters on image
V
0< 5 kV, beam interaction limited to region close to surface, info on surface details V
015 - 30 kV, beam penetrates into sample, info on interior of sample
V
0= beam voltage
Electron energy
Electron column
e
-are produced and accelerated
Beam is reduced to increase resolution
Beam is focused on sample
Filament
e
-are accelerated to anode and the hole allows a fraction of this e
-to reach the lenses
Wehnelt: focuses e
-inside the gun Controls intensity of emitted e
-Grid connected to filament with variable resistor
e
-exit filament following + lines
The equipontential line shape
has focussing effect and
determines
0and d
0Filament
Electron column
Filament head
The equipontential line shape
has focussing effect and
determines
0and d
0Equipotential lines
Electron beam
Filament types
Tungsten hairpin (most common)
Lanthanum hexaboride (LaB
6)
0.120 mm Tungsten wire
LaB6 crystal 0.20 mm
Operating principle: thermionic electron emission
Filament types
Tungsten hairpin Lanthanum hexaboride (LaB
6)
E
w= 4.5 eV
J
c= 3.4 A/cm
2at 2700 K Lifetime 50-150 hours
Energy width 0.7 eV Operating pressure 10
-5mbar
E
w= 2.5 eV
J
c= 40 A/cm
2at 1800 °K Lifetime 200-1000 hours
Energy width 0.3 eV Operating pressure 10
-6mbar
T K
E c
c
Be
wT A
J 2
thermionic electron emission
A
c= 120 A/cm
2K
2E
w= work function
To reduce filament evaporation operate the electron gun at the lowest possible temperature
Materials of low work function are desired.
Filament types
Thermal Field Emission
W-Zr crystal 0.20 mm I = 1 10
4A/cm
2at 1800 °C
Lifetime > 1000 hours Energy width 0.1 eV
Small source dimension (few nm) Operating pressure 10
-9mbar
Operating principle:
thermionic electron emission +
Tunnelling
E gun brightness
Tungsten hairpin Lanthanum hexaboride (LaB
6) Thermal Field Emission
= 10
5A/sr cm
22 2 2 2 2
4 4
angle solid
area
current beam
p p
p
p p p
d i d
i
Brightness is conserved throughout the column
22 2
2
2
1
2 angle 4
solid
pp p
p
d d
R
A
R
pd
p = 10
6A/sr cm
2 = 10
8A/sr cm
2Beam current changes throughout the column
d
p: 30 – 100 m d
p: 5 – 50 m d
p: 5 nm
Electromagnetic Lenses
) ( v B F e
Demagnification of beam crossover image (d
0) to get high resolution (small d
p)
Beam focussing High demag
needed
d
0: 5 – 100 m
for filaments d
0: 5 nm for TFE Low demag
needed
coils Fringe field
radial
parallel
Electromagnetic Lenses
f = focal length
the distance from the point
where an electron first begins to change direction to the point where it crosses the axis.
Focusing process
e
-interacts with B
rand B
zseparately -e (v
zx B
r) produces a force into screen F
ingiving e
-rotational velocity v
inv
ininteracts with B
zproduces a force toward optical axis F
r= -e (v
inx B
z)
The actual trajectory of the electron will be a spiral
The final image shows this spiraling action as a rotation of the image as the objective lens strength is changed.
Electromagnetic Lenses
I = lens coil current N = number of coils V
0= accelerating voltage
Lens coil current and focal length
Increasing the strength (current) of the lens reduces the focal distance
NI 0 2
f V
the focal length will become longer at higher accelerating voltages for the same lens current
Comparison to optical lenses
q p
f
1 1
1
p M q ion Magnificat
q m p ation Demagnific
Beam crossover
d
0= tungsten diameter = 50 m
Scaling from the figure, the demag factor is 3.4 so d
1= d
0/m = 14.7 m
CONDENSER LENSES: the aim is to reduce the beam diameter
Demagnification of beam crossover image (d
0) = object
Objective Lenses
Scope: focus beam on sample
• Pinhole
• No B outside
• Large samples
• Long working distances (40 mm)
• High aberrations
They also provide
further demagnification
• Immersion
• Sample in B field
• Small samples
• Short working distances (3 mm)
• Highest resolution
• Low aberrations
• Separation of secondary from backscattered e
-• Snorkel
• B outside lens
• Large samples
• Separation of secondary from backscattered e-
• Long working distances
• Low aberrations
They should contain:
Scanning coil Stigmator
Beam limiting aperture
Effect of aperture size
Aperture size: 50 – 500 m
Decrease
1for e
-entering OL to
a
adetermines the depth of focus
Determines the beam current
Reduces aberrations
Effect of working distance
Increase in WD increase in q
m smaller larger d lower resolution but longer depth of focus
q p
f 1 1 1
q
m p
Effect of condenser lens strenght
Increase in condenser strenght (current) shorter q larger m and smaller d
Also it brings a beam current reduction, so a compromise between current and resolution is needed
Weak Strong
q m p
NI
0
2f V
Higher I
beamLower I
beamLower d
pHigher d
pDecrease q
1and increase p
2 larger m
Gaussian probe diameter
2 2 2
4
p p
p
d i
Distribution of emission intensity from filament = gaussian with size d
Gd
G= FWHM
2 2
4
p p G
d i
4
2 2 2
G p p
i d
With no aberrations, keeping d
Gconstant would allow to increase i
pby only increasing
pUnderstand how probe size varies with probe current
Calculate the minimum probe size and the maximum probe current
Knowing emitter source size, d
Gmay be calculated from the total demagnification
Spherical aberrations
Origin: e- far from optical axis are deflected more strongly
2
3s C s
d
So at the focal plane there is a disk and not a point
e
-along PA gives rise to gaussian image plane No aberration
e
-along PB cross the optical axis in d
sSpherical aberration disk of least confusion
C
s= Spherical aberration coefficient
f
Immersion and snorkel C
s~ 5 mm Pinholes C
s~ 20-30 mm
So one need to put a physical aperture to limit aberrations x nm
d
sx 0 . 16
2
10 64 10
5 2
) 10 4 ( 10
5
6
3 3
6
9
x nm
d
sx 2 . 5
2 10 10 5 2
) 10 10 ( 10
5
6
3 3
6 6
Aperture diffraction
eV
61 .
0 d d
To estimate the contribution to beam diameter one takes half the diameter of the diffraction disk
E 24 .
1
nm rad
nm 10 24 . 1 10
@ KeV
2nm 89 . 10 1
4
10 24 . 1 61 . 0
3
2
x
d
dOrigin: initial energy difference of accelerated electrons
For tungsten filament E = 3 eV
Chromatic aberrations
Chromatic aberration disk of least confusion
E
0C E d C C
At 30 KeV E/E
0= 10
-4At 3 KeV E/E
0= 10
-3C
s= Chromatic aberration coefficient f
2 2
2
2
s d C
p d G d d d
d
Origin: machining errors, asymmetry in coils, dirt
Astigmatism
Result: formation ow two differecnt focal points
Effect on image:
Stretching of points into lines
Can be compensated with octupole stigmator
Astigmatism
Beam-sample interaction
Backscattered e
-Silicon V
0= 20 KV
TFE, = 1 10
8A/sr cm
2d
p= 1 nm
I
b= 60 pA Simulation of e
-trajectories
Main reason of large interaction volume:
Elastic Scattering
Inelastic scattering
Beam-sample interaction
Elastic scattering cross section
2-2 0 2
20
tan 2 10
62 .
1 electron events atom/cm
E x Z
Q
Z = atomic number;
E = e
-energy (keV);
A = atomic number
N
0= Avogadro’s number;
= atomic density
Elastic Scattering
(cm)
0
N A Q
Elastic mean free path =
distance between scattering events
Silicon
= 2.33 g/cm
3Z = 14
A = 28
N
0= 6.022 10
23
nm Si
atom/cm electron events
x Si
Q
keV keV
2 . 1 )
(
10 66 . 1 )
(
1 5
2 15
1 5
0 0
0
nm x
Si
atom/cm electron
events x
Si Q
keV keV
10 08 . 1 )
(
10 84 . 1 )
(
3 30 5
2 18
30 5
0 0
Beam-sample interaction
Inelastic scattering energy loss rate
J E
AE Z N
ds e
dE i
i
166 .
ln 1 2
4 0
Inelastic Scattering
Z = atomic number A= atomic number
N
0= Avogadro’s number
= atomic density
E
i= e
-energy in any point inside sample J = average energy loss per event
9 . 76 58 . 5
0.19 10
3 Z Z x
J
E
b= 20 KeV
The path of a 20 KeV e- is of the
order of microns, so the interaction volume
is about few microns cube
Beam-sample interaction
Simulation
Energy transferred to sample
Interaction volume 20 KeV beam incident on PMMA
with different time periods
Influence of beam parameters on beam-sample interaction
Beam energy
10 KeV 20 KeV
30 KeV Fe
E ds
dE E Q
1 1
2
Longer
Lower loss rate
Elastic scattering cross section
Inelastic scattering energy loss rate
(cm)
0
N A Q
Incidence angle
Influence of beam parameters on beam-sample interaction
45°
60°
Fe
Smaller and asymmetric interaction volume
Scattering of e
-out of the sample
Reduced depth Same lateral dimensions
surface su rf
ac e
10% to 50% of the beam electrons are backscattered
They retain 60% to 80% of the initial energy of the beam
Atomic number
C (Z=6)
C, k shell
Fe (Z=26)
Influence of sample on beam-sample interaction
Fe, k shell
V
0= 20 keV
Reduced linear dimensions of interaction volume
ds Z dE
Z Q
2Elastic scattering cross section
Inelastic scattering
energy loss rate
Atomic number
Ag (Z=47)
Ag, k shell
U (Z=92)
Influence of sample on beam-sample interaction
U, k shell
V
0= 20 keV
More spherical shape of interaction volume
Backscattered electrons
Signal from interaction volume (what do we see?)
Secondary electrons
Backscattered e
-Backscattered electron coefficient
60°
B BSE i
BSE
i
i
Relationship between
and a sample property (Z)
This gives atomic number contrast
If different atomic species are present in the sample
i
C
i i
C
i= weight concentration
BSE dependence
Monotonic increase
Incidence angle
60°
( )
ncos
BSE dependence
n= intensity at normal
Line length: relative intensity of BSE
Strong influence on BSE detector position
Energy distribution
BSE dependence
The energy of each BSE depends on the trajectory inside sample, hence different energy losses
Region I: E up to 50 %
Becomes peaked with increasing Z
Lateral spatial distribution
Region good for
high resolution Gives rise to loss in lateral resolution
At low Z the external region increases
Sampling depth
BSE dependence
Sampling depth is typically 100 -300 nm for beam energies above 10 keV
Fraction of maximum e
-penetration
(microns) Percent of
R
KOdefines a circle on the surface (center in the beam) spanning the interaction volume
Energy distribution of electrons emitted by a solid
Signal from interaction volume (what do we see?)
Secondary electrons
Energy: 5 – 50 eV
Probability of e
-escape from solid
e z
p
= e
-mean free path
Origin: electron elastic and inelastic scattering
Secondary electrons
SURFACE SENSITIVE
SE
1= secondary due directly to incident beam
SE
2= secondary generated by backscattered electrons
Carbon: SE2 /SE1 = 0.18 Aluminum: SE2 /SE1 = 0.48 Copper: SE2 /SE1 = 0.9 Gold: SE2 /SE1 = 1.5
Low backscattering cross section
High backscattering cross section
Beam resolution
BSE resolution
SE Intensity angular distribution: cos
Image formation
Backscattered e
-Secondary e
-Volume sensitive Surface sensitive
e z
p
Sampling depth
~ 100 -300 nm
Image formation
Many different signals can be extracted from beam-sample interaction
So the information depends on the signal acquired, is not only topography
The beam is scanned along a single vector (line) and the same scan generator is used to drive the horizontal scan on a screen
A one to one correspondence is established between a single beam location and a single point of the display
For each point the detector collects a current and the intensity is plotted
or the intensity is associated with a grey scale at a single point
Signals to be recorded
Image formation
Magnification M = L
CRT/L
sampleBut the best way is to calibrate the instrument
Image formation
Pixel = picture element
Pixel is the size of the area on the sample from which information is collected
Actually is a circle
PE SAMPLE
PE N
D L Length of the scan on sample
number of steps along the scan line
The image is focused when the signal come only from a the location where the beam is addressed
At high magnification there will be overlap between two pixel Digital image: numerical array (x,y,Signal)
Signal: output of ADC Resolution = 2
n8 bits = 2
8= 256 gray levels 16 bits = 2
16= 65536 gray levels
Considering the matrix defining the Dimension of Pixel Element
Image formation
For a given experiment (sample type) and experimental conditions (beam size, energy) the limiting magnification should obtained by calculating the area generating
signal taking into account beam-sample interactions and compare to pixel size 2
2 BSE
eff d B d
d
beam Area producing BSe
-V
0= 10 keV, d
B= 50 nm
on Al, d
BSE= 1.3 m d
eff= 1.3 m on Au d
BSE= 0.13 m d
eff= 0.14 m
There is overlapping of pixel signal intensity
10x 10 cm display
Different operation settings
for low and high magnification
Depth of field
Depth of field D = distance along the lens axis (z) in the object plane in which an image can be focused without a loss of clarity.
To calculate D, we need to know where from the focal plane the beam is broadened
2 tan /
D
r
The vertical distance required to broaden a beam r
0to a radius r (causing defocusing) is
For small angles
2 tan /
D r
D 2 r
Broadening means adjacent pixel overlapping
Depth of field
On a CRT defocusing is visible when two pixels are overlapped r = 1 pixel (on screen 0.1 mm)
But 1 pixel size referred to sample depends on magnification
To increase D, we can either reduce M or reduce beam divergence
2 mm . 0 M D
1 mm . 0 r M
D 2 r How much is r?
Beam divergence is defined by the beam defining aperture
W
D
AP R
Depth of field
W
D
AP R
Optical SEM
Detector
Everhart-Thornley Secondary + BSE
Grid negative: only BSE
solid angle acceptance: 0.05 sr Geometric efficiency: 0.8 %
Grid Positive: BSE+SE
The bias attracts most of SE
Topographic contrast
Intensity of SE and BSE depends on beam/sample incidence angle () and on detector/sample angle ()
BSE coefficient increase with BSE emission distribution ~ cos SE emission distribution ~ sec
Detector position and electron energy window are important
Topographic contrast
Negative bias cage to exclude secondary e
-High contrast due to orientation of sample surfaces
- Detector is on one side of sample anysotropic view - Small solid angle of acceptance small signal
- High tilt angle
Analogy to eye view
Dierctional view
Topographic contrast
Positive bias cage to accept secondary e
-Contributions:
Direct BSE+SE
SE distribution intensity I ~ sec
Variation in SE signal between two surfaces with different dI = sec tan d
So the contrast is given by dI/I = tan d
The SE are collected from most emitting surfaces since the positive bias allows SE to reach the detector
Analogy to eye view
High resolution imaging
High resolution signal if selected in energy
High resolution signal generated by BSE
1, SE
1Separation of signal is necessary to obtain high resolution
SE
1: e
-directly generated by beam
BSE
1: low energy loss (<2%) e
-from beam SE
2: e
-generated by BSE into sample
BSE
2: higher energy loss e
-from beam
Silicon V
0= 30 KV
TFE, = 1 10
8A/sr cm
2d
p= 1 nm
I
b= 60 pA
SE
1- BSE
1width = about 2 nm Beam penetration depth = 9.5 m Emission area = 9.5 m
Scan width at 10000 X = 10x10 m
2image 1024x1024, pixel width 10 nm
Low mag
Scanning at low M means field of view larger than SE
2emission area
So there is large overlap between pixel
And the changes are due only to SE
2variations
Scanning at high M means field of view smaller than SE
2emission area
So as the beam is scanned, no changes in SE
2but changes are due to SE
1SE
2gives large random noise
Scan width at 100000 X = 1x1 m
2image 1024x1024, pixel width 1 nm
High mag
FWHM = 2 nm
Carbon nanotubes Ag NP on glass
TiO
2on silicon
SEM in FOOD
Schematic representation of gaseous SED the role of imaging gas in VP-SEM
B. James / Trends in Food Science & Technology 20 (2009) 114