Greg Pilgrim - University of Rochester Department of Chemistry
OPT 407 Spring 2011
Electron mirrors are the result of charge buildup on an electrically
isolated sample. That sample becomes a mirror when an electron beam
with less energy (a lower accelerating voltage) than is stored on the
mirror is directed at it. The beam electrons are repelled by
the negative charge on the sample and sent in another direction where
they interact normally with another
“sample”. That sample can be another
actual sample on the stage or the inside of the SEM chamber.
Such a mirror adds a degree of freedom to the imaging process allowing
one to (for example) image around corners. That's all well
and good but significant problems arise when trying to use electron
mirrors for useful imaging. Specifically the shape of the
mirror, any imperfections on its surface, and even the added working
distance from inserting the mirror into the beam path will cause
distortions in the image. (Figure 1) Here attempts are made to qualify
and quantify that distortion.
1: A micrograph acquired via
an electron mirror of the stage and SEM chamber interior. The
mirror used was a Teflon sphere and the image displays a characteristic
“fish eye” distortion. (Inset: a
photograph of the stage with stubs for comparison.)
Three different types of electron mirrors were used to generate
micrographs: Teflon (Polytetraflouroethylene also known as PTFE)
spheres, Teflon “holes” (dilled out spheres) and a
High Density Polyethylene (HDPE) “plate” (a flat
piece of plastic). Several of these mirrors can be loaded at
once. In this manner one mirror can be imaged by way of another.
All the mirror types were charged for several minutes at a
low scan speed (~1 scan/2 seconds) with a 20kV electron beam.
The power of the beam was then dropped to a lower
accelerating voltage (8eV 9eV or 10eV as noted on the micrographs).
As soon as the voltage has dropped and stabilized a new
(reflected) image appears on the raster. It should be noted
that although all the mirror materials used are insulators, mirrors can
be formed with conductors as well as long as they are isolated from
ground. Most micrographs were collected with the SE2
detector due to it's sensitivity compared to the in lens.
The in lens detector does not us a Faraday cage but rather
only collects electrons directly incident on the lens.
Imaging was also done using the back scatter detector which
will be discussed later.
2: The small ball mirror
imaged via the large ball mirror. Colorized is for artistic
Distortion and Focal Points
A standard sample of 300 mesh copper TEM grid was imaged using each
type of mirror. The grids were placed on a special 45 degree
stub as the reflected electron beam does not lend itself to imaging
samples on flat stubs. The resulting grid images were then
measured using ImageJ and compared to a image taken without a mirror in
the beam path. Significant differences in scaling were
observed (Figure 3). In addition to issues with scale focal
points are also an issue. Increases in working distance as
seen in these experiments serve to improve depth of field however the
sample geometry works against it. The plate image (Figure 8)
illustrates this particularly well. The right side (as
viewed) of the image is in focus, but the rest of it is out.
This is because the samples were oriented such that the plate-reflected
beam came across the breadth of the grid stage rather than hitting it
at normal. An attempt was also made to use 1um gold sputter
coated polystyrene balls as a standard. It ended
poorly. (Figure 9) Stigmation correction is also an issue.
Many times the stigmators would be maxed out and the stigmation
still not corrected.
3: Table of measured grid
widths and working distance increases associated with each mirror type.
||Grid Width (um)
||Increase in Working
4: Grid with no
Grid via large ball
6: Grid via small
Grid via plate
8: Grid via
Polystyrene Balls (1 um)
Electron mirrors serve to massively increase the working distance of
the electron beam. Figure 1 shows a normal Sample/Beam
interaction with the working distance highlighted. Figure 2
displays the increase in that working distance when a mirror is
inserted into the path of the beam.
10: Normal (non mirror)
working distance Figure 11:
Mirror working distance
Increases in working distance improve depth of field but at the cost of
resolution. The loss in resolution comes from the increased
effect of spherical aberrations as seen in Figure 12 below.
Example A has a spherical aberration causing the beam to have multiple
focal points. Example B does not have a spherical aberration
and therefore has only one focal point. This extended working
distance is what makes imaging at higher resolutions nigh on impossible.
12: Effect of working
distance on focal point
Sample mirror geometry illustrating differences in working distance.
Since, as previously
mentioned the grid is placed on a 45° slant the interaction
volume of the reflected beam (10kV for this experiment) is tilted as
well. In the actual experiment the beam and interaction volume
would be rotated 90° clockwise. (Figure 14) As the copper is
very thin most of the interaction is actually in the carbon tape layer
with a bit leaking into the stage. There is no interaction volume
on the mirror as the electrons repel the beam before it can reach the
14: Interaction volume of a
10kV beam with 600A of
copper and 5800A of carbon tape on an aluminum stub
- Mirrors for BSD troubleshooting
A practical use for electron mirrors within the confines of scanning
electron microscopy is aiming a mirror at the backscatter
detector. This detector (at least in the Zeiss Auriga model)
has four components (Figure 15) that when activated change color
(Figure 16) when viewed through that same backscatter
detector. If the detector isn’t working one of
these sections will fail to light up. By imaging the detector
a technician can tell which component is malfunctioning and replace it.
16 and 17: The
back scatter detector. Figure 17 was taken by combining the
signal from the SE2 detector with that of the back scatter detector.
While electron mirrors are stable enough to use for extended periods of
time and have some practical application the distortions associated
make imaging challenging. This is especially true at high
magnification where the combination of increased working distance and
uncorrectable mirror astigmatism combine to destroy
resolution. More stable (perhaps powered, with stigmators)
and finely machined mirrors could help alleviate these issues but for
the time being electron mirrors remain a curiosity rather than a widely
useful imaging tool.
Thanks first and foremost to Brian McIntyre for his guidance and
patience with this project
To Greg Savich for allowing me to continue with his work and for his
help getting me going
To Zack Lapin and Mike Theisen for their time and assistance as TAs in
1) Rasband, W.S., ImageJ, U. S. National Institutes of Health,
Maryland, USA, http://imagej.nih.gov/ij/, 1997-2011.
2) Savich, Gregory,
3) Smith College Microscopy, http://126.96.36.199/microscopy/sem.html
4) F. CROCOLO, C. RICCARDI, Passive Mirror Imaging through a
Solid-State Back-Scattered Electron Detector, Microscopy Today, March
5) M Milani et. al. ˝Rear window˝ : looking at charged particles
hitting a charged target in a FIB/SEM Microscopy: Science, Technology,
Applications and Education 2010