First lab reportsGrading

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 103Friday classes: WEB 112

Currents in an SEM (W-filament)

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

Specimen current

X-raysCathodoluminescence

Pole Piece, etc

SE3

≈ 1 nm for metals upto 10 nmfor insulators

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 near-surface 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

High resolution contrast information is therefore diluted by the low spatial resolution SE2 background

SE2 come from the full width of interaction volume

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

Silver nanocrystals 1keV

0.1µm

as a result. . . .as a result. . . .

Indium Tin Oxide (ITO) Indium Tin Oxide (ITO)

the SE signal (in the LVSEM can produce)

high contrast nm resolution easy to interpret

surface images from crystals & nano-particles…

Silver Nanocrystals 1keV

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 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 interaction volumes

low voltagehigh

voltage

BS yields at Low Voltage

The BSE yield 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 is absent

•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.

At high energy tilting the sample puts more of the interaction volume in the SE escape zone

SE escape

But at low energy all the SE always escape

Beam penetration effects

At high energy the interaction volume fills features on the surface - SE2 emission leads to enhanced SE emission making objects look almost 3-dimensional

But at low energies the reduced interaction volume means that only the edges of features are enhanced

SE emission

High energy

Low energy

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

I bI b

I b

sc

Electrons cannot be created or destroyed so currents at a pointsum to zero (Kirchoff’s Law)

Where 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

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 thenThis is achieved when

1..)( ifeiII BB

0dt

dQ

This condition represents a dynamic charge balance

If ()<1 then negative charging will occur and

If ()>1 then positive charging will occur

The charge balance condition

The variation of the () yield curve is about the same for all materials

In most cases there are energies for which () = 1

These are called the E1 and E2 or ‘crossover’ energies

Total yield data for quartz (SiO2)

Positive charge

Negative charge

NEUTRAL

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 E2(keV) Material E2 (keV)

Resist 0.55 Kapton 0.4 Resist on Si 1.10 Polysulfone 1.1 PMMA 1.6 Nylon 1.2 Pyrex glass 1.9 Polystyrene 1.3 Cr on glass 2.0 Polyethylene 1.5 GaAs 2.6 PVC 1.65 Sapphire 2.9 PTFE 1.8 Quartz 3.0 Teflon 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.

substrate

SE BS

Thin film charging (E3)

SE Image of Chip covered by a 1m passivation layer imaged at 15keV -

above the E3 energyHow a thin metal film on top of an

insulator charges with 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

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

Tilting the sample reduces charging at all energies

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 TIME

Beam dwell time on pixel

Potential

Slow scan

Fast scan

Forget eliminating charge – stabilize it then live with it

IIBB=100pA=100pA

Vacc.Vacc. : : 1.5kV 1.5kV

Mag.Mag. : : x200kx200k

Scan stabilized imaging

Uncoated photoresist

Imaged at E2 and scanned

at TV rate

the choice of detector

Pure SE signal – Thru-lens upper detector

ET lower detector SE + BSE + scattered electrons

Single polymer macro-molecules

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 electron-hole 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 macro-

molecules 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

Poly2 with CoSi on Top

rougher

SiO2

Si substrate with CoSi2

smoother

What is this residue??

Missing CoSi!!

Side Detector - Topography In-Lens Detector – Chemistry Image

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 becomes a good choice

Uncoated Teflon S4700 ExB BSE image

Controlling Charging by Coating

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

ground plane

Field lines do not leak away from the surface

Charge in sample

Field deflects electrons

If you must use carbon..

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

Con

du

ctiv

ity

Thickness

Minimum useful thickness is about 10nm

Con

du

ctiv

ity

Temperature-150C RT

300M-ohm

1 M-ohm

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 energyExperimental 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 effects25keV 2.5keV 1.0keV 0.5keV

Kenway-Cliff numerical ray-tracing simulations of electron arrivals with a lens Cs=3mm,Cc=3mm, =7 m.rads

5nm

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

Vacc

VR

ee

Accelerating VoltageVacc

VR Retarding Voltage

Vacc VR Landing Voltage

(ex)

2000V – 1500V = 500V

Retarding system

Vacc

Normal Accelerating Voltage

AcceleratingVacc

Voltage

Keeping 2kV spot size and beam current condition, accelerating voltage of 500V condition is obtained.

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

Sample : Membrane Filter

0.1kV 0.1kV

Disadvantages of Retarding System

Sample

Electron beam without retarding

Electron beam with Retarding

1 2

3 4

5

Not usable for general Depth of Focus become shallow

(SE/BSE) Signal Control cannot be used

1

2

3

4

5

Sample edge area

Pre-Tilted sample

Rough surface sample

Tilting stage

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.

sample observation

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 frontier500eV 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

Because Cs and Cc decrease with the landing energy the imaging resolution is only limited by diffraction

500eV30eV

100nm

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