Transcript Beam-Specimen Interactions

First lab reports
Grading
Explanation of “soft windows” in upper right corner
and how mouse is used to change entities therein: 5
points
Adjustment of gun tilt and gun shift: 3 points
Need for diagram of sample locations: 2 points
Other details: 15 @ 1 point
25 point total
-1 for each incorrect statement
Average was 20
Only two people turned in prelabs for lab 2
Meeting place update
Monday classes: WEB 103
Friday classes: WEB 112
Currents in an SEM (Wfilament)
 Filament current: Current that heats a tungsten filament,
typically 2.6-2.8 A. Strongly affects filament lifetime. Similar
for Schottky FEG, but only heated to 1700 K
 Emission current: total current leaving the filament, typically
about 400 μA for W-filament, 40 μA for FEG.
 Beam current: Portion of emission current that transits the
anode aperture; decreases going down the column.
 Probe current: a calculated number related to the current on
the sample, typically 10 pA – 1 nA.
 Specimen current: the current leaving the sample through
the stage, typically about 10% of the probe current. Remember
that one electron incident on the sample can generate many in
the sample…a 20 keV electron can generate hundreds at 5 eV.
 FEI also defines a parameter called “spot size” which is
proportional to the log2(probe current); proportionality constant
depends on aperture size.
Surface Emissions
Pole Piece, etc
SE3
X-rays
Cathodoluminescence
 ≈ 1 nm for
metals up
to 10 nm
for insulators
Specimen current
Interaction volume
 The interaction volume
falls with beam energy
E as about 1/E5
 (dE/ds ~ [ln(E)]/E)
 The interaction volume no
longer samples the bulk
of the specimen but is
restricted to nearsurface regions only
 The signal is therefore
much more surface
oriented at low energies
than at high
Monte Carlo simulations of
interactions in silicon
What happens at low energy?
 At low energies the
electron range falls
from the micrometer
values found above
10keV to just a few
nanometers at
energies of 0.5keV
 This variation has
profound implications
for every aspect of
scanning microscopy
Range from modified Bethe
equation
Spatial resolution…..
 At high energy the SE1
signal typically comes from
a volume 3-5nm in
diameter, but the SE2
signal from a volume of
1-3µm in diameter
SE2 come from the full
width of interaction volume
 High resolution
contrast information
is therefore diluted
by the low spatial
resolution SE2
background
But at low energies…...
 ..the SE1 and SE2
electrons emerge from
the same volume because
of the reduction in the
size of the interaction
volume
 So SE1, SE2 and BSE
images can all exhibit
high resolution….
the interaction volume shrinks
Seeing is believing
 The sample is a 30nm film
of carbon on a copper grid
 At 20keV the carbon film
is transparent because it
is penetrated by the
beam.The SE signal comes
from the carbon film but
is produced by electrons
backscattered from the
copper
SE image of TEM grid 20keV
Electron range at low energy
 Carbon film completely
covers grid!!
 At 1keV -by comparison the carbon appears solid
and opaque because the
beam does not penetrate
through the film
 This variation of beam
range with energy is
dramatic and greatly
affects what we see in
the low voltage SEM
Same area as before but 1keV beam
Some consequences of low
energy operation
 The interaction volume decreases in
size and shrinks towards the top
surface as the energy falls
High Energy Images
 At high energies the beam
travels for many
micrometers giving the
sample a translucent
appearance
 The SE image information
is mostly SE2 and so
copies the BSE signal.
 The information depth is
~Range/3 and so is often
a micron or more
MgO cubes 30keV S900
Low Energy Images
 At low energy the beam
only penetrates a few
tens of nanometers.
 The image now only
contains information about
the surface and the near
surface regions of the
specimen
 The signal information
depth (SE1,SE2 and even
BSE) is only nanometers
0.1µm
Silver nanocrystals 1keV
as a result. . . .
the SE signal (in the
LVSEM can produce)
 high contrast
 nm resolution
 easy to interpret
surface images
from crystals &
nano-particles…
Silver Tin
Nanocrystals
1keV
Indium
Oxide (ITO)
and ..
….organics such as polymer resists
The best approach - try a wide range
of energies and modes
CNT with intercalated iron
Some consequences of low energy
operation
 Spatial resolution is improved in
all image modes
Low Voltage BSE imaging
 At a WD of 1.5 or
2mm high resolution
BSE imaging is readily
possible and is very
efficient
 ‘Z’ contrast may be
less evident at low
energies than at high
Ta barrier under copper seed
Some consequences of low
energy operation
 Changes in SE and BSE generation
lead to differences in image detail
and interpretation
SE yield variation
 The rapid change in the
incident electron beam
range causes a large,
characteristic variation in
the SE yield
 Typically the yield rises
from ~0.1 at 30keV to in
excess of 1 at around
1keV, and as high as 100
for some materials
Experimental SE yield data for Ag
Why the SE yield changes
 SE escape depth is ~ 3-5nm
 At high energies most SE are
produced too deep to escape so
the SE yield d is low
 But at lower energies the
incident range is so small that
most of the SE generated can
escape so the SE yield rises
rapidly
 At very low energies fewer SE
are produced because less
energy is available so the SE
yield falls again
high
voltage
low
voltage
interaction volumes
BS yields at Low Voltage
 The BSE yield h varies with
energy as well as with
atomic number
 Above ~2keV the yield
rises steadily with Z
 But at low energies the BSE
yield for low Z elements
rises, and for high Z
elements it falls
 Below 100eV the situation
is more complex
Experimental BSE yield data
Do high and low kV SE images look the
same?
•No..compared to
the high energy
norm…
•The image looks
less 3-D
•Highlighting
absent
is
•Surface junk is
more visible
•Interpretation is
essential
Device images at 20keV and 1keV
Origin of topographic contrast
 Topographic contrast weakens
and ultimately disappears as
the beam energy is reduced.
SE escape
At high energy tilting the sample puts more of the
interaction volume in the SE escape zone
But at low energy all the SE always escape
Beam penetration effects
SE emission
High energy
Low energy
 At high energy the interaction
volume fills features on the
surface - SE2 emission leads
to enhanced SE emission
making objects look almost 3dimensional
 But at low energies the
reduced interaction volume
means that only the edges of
features are enhanced
Some consequences of low
energy operation
 Less charge is deposited in the
sample
 This is the real advantage of a
FEG over a W-filament: FEG has
almost as much resolution at 1 kV
as at 15 kV
 FEI now has landing energies as
low as 50 eV!!!
The LVSEM and charging
 When electron beams impinge
on non-conducting samples a
charge can build up inside the
specimen which can make SEM
imaging unstable, difficult, or
even in extreme cases,
impossible
 By operating at low beam
energies this problem can
often be minimized or
eliminated
 Low voltage SEM has now
become the norm for many
users because of this effect
Pathological charging artifacts
Charge Balance
Electrons cannot be created or
destroyed so currents at a point
sum to zero (Kirchoff’s Law)
I B  (h  d ) I B  I SC 
I
h I
b
b
dI
b
dQ
dt
Where h, d are the BSE and SE yields respectively, and Q is the
charge on the specimen at some time t. For a conductor this
equation is always balanced by Isc

sc
Working with Conductors
 If the sample is a conductor then it cannot
charge and Q=0 at all times
 In this case at high energies where electron
yields are small excess current flows to ground
as specimen current ISC
 At low energies where yields are high current
flows from ground to make up the deficit
 But the charge is always balanced and stable
imaging is possible
..but in an insulator
ISC is zero
If the sample is not to charge then
This is achieved when
I B  I B (h  d ) i .e .
if
dQ
dt
h d 1
This condition represents a dynamic charge balance
If (hd)<1 then negative charging will occur and
If (hd)>1 then positive charging will occur
0
The charge balance condition
 The variation of the
(hd) yield curve is
about the same for all
materials
 In most cases there
are energies for which
(hd) = 1
 These are called the E1
and E2 or ‘crossover’
energies
Positive charge
NEUTRAL
Negative charge
Total yield data for quartz (SiO2)
E1 and E2 values for pure elements
 E1 and E2 both increase with
atomic number Z
 E2 may also depend on the
density (e.g diamond, graphite,
and dry biological tissue have
very different E2 values)
 A few elements never reach
charge balance (e.g Li, Ca)
 Low Z elements need low keV.
Since these elements so important
the goal has been to make SEMs
work at 0.5 - 2keV
Computed E1 and E2 energies
E2 values
Material
Resist
Resist on Si
PMMA
Pyrex glass
Cr on glass
GaAs
Sapphire
Quartz
E2(keV)
0.55
1.10
1.6
1.9
2.0
2.6
2.9
3.0
Material
E2 (keV)
Kapton
Polysulfone
Nylon
Polystyrene
Polyethylene
PVC
PTFE
Teflon
0.4
1.1
1.2
1.3
1.5
1.65
1.8
1.8
Determining E2 in the SEM
Negative E>E2
Positive E<E2
Charging in Complex materials
 In the case of complex
materials (e.g. layered) then
the charge balance must be
considered separately for each
component
 If a beam penetrates a layer
then it will charge positively
(net electron emitter). The E3
energy at which this first
occurs is typically <1keV for
3nm of hydrocarbon, and a few
keV for a 250nm thick
passivation layer.
SE
BS
substrate
Thin film charging (E3)
How a thin metal film on top of an
insulator charges with energy
SE Image of Chip covered by a 1mm
passivation layer imaged at 15keV above the E3 energy
Imaging non-conductors
 On a new SEM this will be the
lowest available energy
 On older machines you must
decide how low to go before
the performance becomes too
poor to be useful for the
purpose intended
 The goal is to avoid implanting
charge deep beneath the
surface. If this is allowed to
occur then stable imaging may
never be achieved.
Step #1 - Set the SEM
to the lowest
operating energy
Failure to follow this advice...
 If a poorly conducting
sample is irradiated with
a high energy beam then
the implanted charge
may prevent a low energy
beam from reaching the
surface at all
 In that case it acts as a
mirror giving a birds’ eye
view of the inside of the
SEM
Mirror image of sample chamber in an SEM
Next……...
 If the sample is charging
positively (i.e. a dark scan
square) then E1< E<E2 or
E>E3. Increase the beam
energy and proceed to image
 If sample is charging
negatively (i.e. bright scan
square) then E>E2.
 Since we cannot reduce the
beam energy any further we
go on to step 3.
Step #2 - Determine
the charging state of
the sample using the
scan square test
Step 3
Tilting the sample reduces charging at
all energies
 Tilt the sample to 45 degrees
and repeat the usual scan
square test
 Can E2 be reached now?
 E2() = E2(0)/cos2
so tilting by 45 degrees raises
E2 by a factor of 2x
 But ..because E2 varies with
the angle of incidence the ‘no
charge’ condition can never be
satisfied everywhere on the
surface at the same time and
charging will always occur
So does charge balancing help ?
 In some cases - yes
 But because the E2
‘charge balance’
condition can never be
simultaneously
satisfied everywhere
on a surface with
topography - hence
charging will always
be present
Phase Shift Lithography mask slow
scan imaged at E2
A better strategy
 Go to E2 and then scan at
high rates
 The sample acts like a leaky
capacitor which charges more
quickly than it discharges
 At slow scan speeds each pixel
charges and then discharges
before the beam reaches it
again
 this fluctuating potential
affects SE emission, signal
collection, scan raster etc
 At high scan speeds (TV)
there is less time to discharge
so the potential stabilizes
P
o
te
n
ti
al
Fast
scan
Slow
scan
Beam dwell time on
pixel
TIME
Forget eliminating charge – stabilize it then live with it
Scan stabilized imaging
IB=100pA
Vacc. : 1.5kV
Mag. : x200k
Imaged at E2
and scanned at
TV rate
Uncoated photoresist
the choice of detector
Single polymer macro-molecules
Pure SE signal – Thru-lens
upper detector
ET lower detector SE
+
BSE
+
scattered electrons
makes a difference
Uncoated Teflon tape adhesive BSE image at 2keV
..so does reducing IB
 the charging varies
directly with IB so
reducing the current
cuts the charge
 Use a smaller
aperture, or reduce
the gun emission
current
 Reduces the S/N
ratio so longer scan
times may be required
..and lowering the magnification
 This minimizes Dynamic
Charging (internal charge
production from electronhole pairs). The magnitude
of this depends on the
dose and hence on the
magnification
 Dynamic charging is worst
when E0 is close to E2
 Limits resolution by limiting
magnification
Choosing a detector
 The choice of detector
can have a significant
effect on the apparent
severity of charging
 The conventional ET
(Everhart - Thornley)
detector sees more
topography but is much
less sensitive to charging
than...
Individual polymer macromolecules on Si at 1.5keV Lower (ET) detector
Upper detector
 …a through the lens
detector. This is because
TTL systems act as simple
SE spectrometers and
preferentially select low
energy electrons
 Note however that
charging can be a useful
form of contrast
mechanism when
properly employed
Same area as before, TTL detector
Comparing upper and lower detectors
In-Lens Detector – Chemistry Image
Side Detector - Topography
Poly2 with CoSi on Top
rougher
What is this residue??
SiO2
Si substrate with CoSi2
smoother
Missing CoSi!!
BSE imaging to avoid charging
 Backscattered electrons are
less affected by charging
and offer the same
resolution as SE at LV
 Newer technologies such as
conversion plates, and ExB
filters, for BSE actually
improve in efficiency as the
beam energy is reduced, so
using this mode to avoid
charging problems
Uncoated Teflon S4700 ExB BSE image
becomes a good choice
Controlling Charging by Coating
Field deflects electrons



The oldest method for controlling
charging is to put a coat of carbon or
metal on the surface of the sample
Coatings do not make the sample a
conductor except in the limiting case
when the surface is buried by a thick
layer of metal
Instead the coating forms a ground
plane - a localized equipotential region.
In this area the free electrons in the
metal re-arrange so as to eliminate the
external field. The sample remains
charged but incident and emitted
electrons are unaffected
Charge in
sample
ground plane
Field lines do not leak away from the
surface
If you must use carbon..
Minimum
useful
thickness is
about 10nm
Thickness
1 M-ohm
Conductivity
 Carbon is not an ideal coat
because it must be quite thick
before it becomes a good
conductor and has a low SE
yield. Do not use evaporated
carbon as this contains a lot of
filler, instead use ion deposition
 Thickness - probably 10 to
20nm minimum
 How to check - shadow on the
filter paper is light to dark
grey
300M-ohm
-150C
Temperature
RT
How effective is coating?
 Thin films of either Au-Pd
and Cr can effectively
eliminate charging up to 8keV
 Even at higher beam energies
charge-up is minimal
 Thin metal coats do not
degrade EDS analysis - they
improve it because they
stabilize the beam landing
energy
Experimental Charging Data from
Alumina (Sapphire)
Radio Shack Special
If you prefer too make a
ground line, or provide a
ground plane the Circuit
Writer, or Artic Silver, pens
which deposit a silver loaded
polymer work very well
Resistivity <0.1ohm.cm and
dries quickly
No vapor in vacuum
Building a real low voltage SEM
 There are several problems in achieving competitive
electron-optical performance at low energies
 Gun brightness falls linearly with energy. A FEG at
500eV is only as bright as a tungsten hairpin at 20keV
 It is increasingly difficult to shield the column against
outside electro-magnetic interference
 The electron wavelength gets larger so diffraction is
significant
 Depth of field decreases
 Chromatic aberration is the killer
Chromatic aberration effects
25keV
2.5keV
1.0keV
0.5keV
5nm
Kenway-Cliff numerical ray-tracing simulations of electron arrivals
with a lens Cs=3mm,Cc=3mm, a =7 m.rads
The energy spread of the beam causes a chromatic error in the
focus. Even with a cold FEG source (~ 0.3eV wide) this greatly
degrades the probe at 0.5 keV and below. Both the source and
the objective lens are important factors
Building a ULV CD-SEM
 Decelerating the electrons just
before they strike the sample
reduces the landing energy and
improves the optics
 If the beam voltage is E0, then
the landing energy is
Ef =E0-VB
and it can be shown that
Cc’ = -Cs = L.Ef/E0
 So if Eo=5000V, the landing
energy is 50eV, and L ~ 1mm
then Cs and Cc are reduced from
mm to micrometers
Retarding on the S4800
Retarding Field Operation can be used in two ways (a) to enhance the imaging
performance at an energy that is already available or (b) reach beam energies below the
lowest value available on the microscope
Normal Accelerating Voltage
Vacc
e
Retarding system
Vacc
e
Vacc
Vacc
Accelerating
Voltage
VR
Landing Voltage
(ex)
2000V – 1500V = 500V
Keeping 2kV spot size and beam current
condition, accelerating voltage of 500V
condition is obtained.
VR
Vacc
VR
Accelerating Voltage
Retarding Voltage
0.1kV
0.1kV
Sample : Membrane Filter
Mode (1) uses the retarding field effect to enhance resolution.
A retarding field image at 500eV has better resolution than a standard image at
500eV because the aberrations are smaller.
Here EL = 100eV => 1600 eV beam in - 1500 volts retarding potential
Disadvantages of Retarding System
Not usable for general
sample observation
1
Depth of Focus become shallow
2
Electron beam
retarding
without
Electron beam with
Retarding
3
4
Sample
(SE/BSE) Signal Control cannot be used
5
1 Sample edge area
2 Pre-Tilted sample
3 Rough surface sample
4 Tilting stage
5 Cross-section
Secondary electrons are accelerated by retarding voltage and
have the same energy level as backscattered electrons. So, it
becomes impossible to detect each signal separately. As a
result, always mixed signal of SE and BSE is detected and its
mixing ration cannot be controlled.
Mode 2 - ultra-low voltage use
 Retarding field can also be used
to reach ultra-low energies
 Below about 200eV SE and BSE
cannot be separated and so we
must consider them together
 The total yield (SE+BSE) varies
rapidly with beam energy as
shown but significant signal is
still present at energies <10eV
 Note that the total signal level
at 100eV is about the same as
that at 2keV so the signal to
noise ratio should be acceptable
Total yield data for Copper
The new frontier
500eV
25eV
14eV
Topographic contrast disappears at ultra-low energies but
strong shadow (detector) contrast remains visible.
Contamination is minimal. Many of these effects remain
to be explained
Resolution at Ultra-Low Energies
 The resolution can be
maintained at a very good level
using the retarding field
approach
 Down to energies of 30-40eV
the resolution is approximately
independent of the choice of
landing beam energy
 In this example images at up to
300kx are shown at 100eV
from a Hitachi S4800
Courtesy Bill Roth HHTA
Resolution at ultra-low energies
30eV
500eV
100nm
 Because Cs and Cc decrease with the landing energy the imaging
resolution is only limited by diffraction
Contrast changes with energy
As the landing energy is
reduced from 300eV the
contrast in this example
changes in many different
ways. For example, note
the change in contrast of
the ‘black dots’ below 60eV
- first they disappear then
they reappear in opposite
contrast.
Retarding Field ULV operation is a powerful new mode on
the S4800 microscopes