What is needed for High Resolution SEM?

A small probe size High beam currentA mechanically

stable microscope and a quiet lab environment

A skilled operator

Lens performance

The probe size is determined by the aberrations of the lens

The magnitude of the aberrations vary with the focal length of the lens - which is about equal to the working distance

Some lens’ designs are more capable than others at combining both high performance and good sample access

Lens performance cont.

Aberrations including spherical and chromatic are correctable to varying degrees

Corrections depend on a variety of factors including pole piece quality, aperture size, aperture, angle and electron wavelength

Pinhole Lens

The original SEM lens - designed to produce no magnetic field in the sample chamber

Good sample access Long focal length and a

big working distance so high aberrations

Poor EM screening Asymmetric SE

collection due to position of ET

ET

sample

Immersion Lens

Short focal length - so low aberrations

Good EM screening Very stable specimen

mounting in lens Symmetric SE

collection using the through the lens (TTL) detector system

Restricted to small samples (3mm disc)

det

magnetic field

B

Snorkel Lens

Short focal length - so low aberrations and high performance

Good EM screeningThe sample is

outside the lens so there is no limitation on the size of the specimen

TTL

ETMagnetic fieldprojected out of lens

S4700 Snorkel Lens

Up to 45 degrees of sample tilt even at short WD and permits EDS operation at WD of 12mm

Biased deflector plates optimize SE collection for either or both detectors

Improved magnetic screen and stronger stigmators can image magnetic samples at all WD

The lens also acts to filter the SE signal to the TTL

S4700 lens configuration Excitation - 1000 amp.turns

S4700 detector

Snorkel lens permits multiple detectors to be used

In-lens (TTL) detector gives a shadow free image with ultra-high topographical resolution. Super efficient

Lower (ET) detector gives SE images with material contrast information and high efficiency at high tilt angles

These detectors can be used separately or combined as desired for maximum flexibility

Snorkel lenses allow multiple detectors

Detector Flexibility

DRAM with both Upper and Lower detectors MO layer in BSE mode

(DRAM stands for dynamic random access memory, a type of memory)

Multiple imaging modes provide flexibility and problem solving power

What determines spot size? The spot size depends on

the beam energy, WD, and the final aperture convergence angle

Performance improves with higher energies

On the S4700 the aperture size is set automatically

Changing the CL (spot size) does not affect resolution much Variation of probe size with energy

and beam convergence for S4700

Working Distance

Working distance is the most important user controlled parameter

Always use the smallest WD that is possible for a given specimen

Note also that the image resolution is almost independent of the beam energy

Imaging

Microanalysis

Beam current

Typical contrast levels are 3-10% on most samples

Improving contrast lowers required IB , beam current, and improves resolution

Increase IB by raising the tip emission current from 10A to 20 or 30 A if necessary

Resolution

The pixel size is equal to the CRT pixel size divided by the actual magnification e.g a 100µm pixel at 100x gives 1µm resolution

Probe size only limits resolution at high magnifications Image at 1kx magnification

has 0.1µm pixel resolution

Image Content

SE1 - high resolution

SE2 - low (BSE) resolution

SE3 - tertiary signal, interactions of the BSE with the pole piece and chamber walls

ET sees 40% SE3, 45% SE2, 15% SE1

TTL sees 75% SE2 and 25% SE1

SE escape

Lens Detector

ET

TTL

SE1SE2 SE3

SE1

SE2

BSE

SE1/SE2 interaction volumes

The SE1 signal comes from a few nm area at all energies

The SE2 signal comes from an area that can be up to a few microns in diameter at high energies

Pixel size and SE2

At low and medium magnifications the pixel size ( a few µm) is comparable with SE2 interaction volume

So the image is mostly from the BSE generated SE2 component

The SE1 are not a significant contributor

SE2area

pixel

Medium magnification

Medium magnification images have a resolution limited by SE2 interaction volume

SE and BSE images will look similar but not necessarily identical Image at 20kx - 50Å

pixels

High magnification images

Field of view is about size of the SE2 interaction volume so that signal remains about constant as beam scans

The pixel size is about equal with the SE1 area so the SE1 component now provides the image detail

field of view

pixel

Pixels - a summary

High resolution requires the use of a high magnification to keep the pixel size at a small enough value not to limit the resolution

High resolution at high beam energies also requires a high magnification so as to separate the SE1 signal from the lower resolution SE2 signal

High resolution imaging On the S4700

imaging in SE mode with a resolution into the nanometer range is readily possible

What is the ultimate resolution limit?

Optical performance, signal origination, and current to establish sufficient signal quality

Imaging a 10nm thick oxide layer

How good is SE resolution?

The production of SE occurs over a finite volume of space

The initial SE event produces additional SE and so on, leading to a diffusing cloud of SE around the impact point

How far do they travel? Depends on the MFP (mean free path)

SE resolution

The diffusion effect is visible at the edges of a sample as the ‘bright white line' due to extra SE emission

The width of this line is a measure of the SE MFP

The presence of this SE1 edge effect sets an initial limit to the achievable SE image resolution

Molybdenum tri-oxide crystals Hitachi S900 25keVSE diffusion volume

Classical resolution limit

When the object is large its edges are clearly defined by the ‘white lines’

But as the feature reaches a size which is comparable with the edge fringes begin to overlap and the edge contrast falls

20nm

Width =

10 nm

Classical resolution limit

When the feature size is equal to or less than the edge lines overlap and the object is not resolved at all since it has no defined size or shape

This is Gabor’s resolution limit for SE imaging

The resolution in SE mode therefore depends on the value of

Particle contrast

5 nm

width =

High Resolution Imaging

On a high atomic number, very dense , material such as tungsten the SE MFP is only a nanometer or so

So a spatial resolution of about 1nm is likely to be possible

In fact ...

“Lattice” fringes

In this image by Kuroda et al (J.Elect.Micro 34,179, 1985) fringe structures with a spacing of 1.4nm are clearly visible in the SE image

This resolution is consistent with the diffusion model for SE production with =1nm

Image recorded at 20keV on an Hitachi S-900 FEGSEM

The probe size for this image was about 0.9nm

SurfaceSurfaceConfigurationConfiguration

In other samples...

When an object gets small enough to be comparable with then it becomes bright all over and the defining edges disappear.

For low Z, low density materials, this can happen at a scale of 5-10nm

Carbon nanotubes

edge brightness

no edges

The resolution limit

The resolution of the SEM in SE mode is thus seen to be limited by the diffusion range of secondary electrons, especially in low Z materials

In addition the signal to noise ratio is always worse for the smallest detail in the image

Improving the resolution

Improving SEM resolution therefore requires two steps:

minimizing or eliminating the spread of secondary electrons

improving the signal to noise ratio so that detail can be seen

Improving the S/N ratio

Use a metal coat as all metals give more SE than carbon

SE yield tends to rise with Z value

But high Z materials are denser and cause more scatter

Usually consider Cr, or Ti as best choices but W, Pt are also good

Computed SE1 yield at 2keV

Particulate Coatings

Au produces very big particles (30nm)

Au/Pd and W make much smaller (3nm) particles

These have a very high SE yield

Can be deposited in a sputter coater

Coatings are stable Good below 100kx

3nm of Au/Pd at 100kx

Decoration

In some cases the sputtered particles decorate active features on a structure, making them more visible

High Z materials, such as tungsten also permit BSE imaging

Tungsten decorated T4 polyheads 25nm ring diameter 30keV Hitachi S900

Bypassing the limit

Since metals have much lower than carbon, and a higher SE yield, a thin metal film coating on a low Z, low density sample effectively localizes all SE production within itself. The resolution now is a function of the film thickness only and not of

Works even with very thin metal films (few atoms thick)

Can exploit this effect to give interpretable contrast at high resolution

Low SE yield

High SE yield

width film evenwhen <

Mass thickness contrast

The SE1 yield varies with the thickness of the metal film

This effect saturates at a thickness equal to about 3

The conformation of the film to surface topography thus provides contrast 1nm 2nm 3nm

Film thickness

SE

Yield

bulk value

mass thickness variation

Metal builds contrast

The SE localization in the film provides edge resolution

The mass thickness effect gives extra contrast enhancement

The feature is now truly ‘resolved’ since its size and shape are visible

5nm low Z object2nm metal film

Beam positionSE profile with metal film

SE profile without metal

SE

Cr coatings

Cr films are smooth and without structure even at thicknesses as low as 1nm

The mass thickness contrast resolves edges and make the detail visible down to a nanometer scale

The high SE yield of the Cr improves the S/N ratio

However these coatings are not stable - so use Cr coated samples immediately after they have been made

AIDS virus on human cells 500kx 2nm Cr at 20keV Hitachi S900

Coating Summary

Coatings are an essential part of the technique of high resolution SEM because they generate interpretable contrast, improve resolution, and enhance the S/N ratio

Thin coatings are better than thick coatings - do not make your sample a piece of jewelry

Below 100kx particulate coatings are superior because of higher SE yields

Above 100kx use chromium or titanium MRC lab uses Au/Pd coatings on most samples Carbon is a contaminant not a coating

Getting the most from your SEM

Alignment is crucial. Check aperture alignment every time you change areas or imaging conditions and ensure that the stigmators are properly balanced

Minimize vibrations by choice of SEM location. Move pumps away etc.

Keep the room quiet, noise dampening material on the walls.

Check for stray fields. Remove fluorescent lights and dimmer controls.

Keep computer monitors away - use flat screens

Beware of ground loops

Clean Power

Many cases of ‘jaggies’ are due to dirty mains lines not EM pickup

Check waveform at your wall plug

Use clean power from a UPS for critical electronics

Avoids surges

zerocrossings

AC line andEM Field

raster issynchronizedwith field

switching spikes

raster is nownot synchronizedwith field

zerocrossings

Operating tips

Allow the SEM to thermally stabilize and the cold finger to cool down before attempting high resolution - this may take > 1 hour (seldom used at MRC)

Use the stage lock - but don’t forget to turn it off before unloading sample

Use the beam shift rather than stage motion - but remember to recenter the beam before taking a critical image

Look for the scan speed which minimizes ‘jaggies’ when viewing the image live

Getting the best image

Whenever possible take a single slow speed scan rather than accumulating multiple high speed scans

This eliminates blurring due to drift, and distortions in the video amplifier chain and usually produces a higher signal to noise ratio and better contrast32 high speed

framessingle 20

second scan