XEI Scientific, Inc.
The EVACTRON® Anti-Contaminator and De-Contaminator Updated August 2007
Active Monitoring and Control of Electron Beam
Induced Contamination
National Institute of Standards and Technology,
Gaithersburg, MD 20899, USA1
Presented at SPIE Microlithography Conference Feb.
27-28, 2001
ABSTRACT
The vacuum system of all scanning electron microscopes
(SEMs), even in the so-called "clean" instruments, have
certain hydrocarbon residues that the vacuum pumps do not effectively
remove. The cleanliness of the vacuum and the amount and nature of
these residual molecules depends on the type of the pumps and also on
the samples moved through the system. Many times, the vacuum readings
are quite good but the electron beam still leaves disturbing
contamination marks on the sample. This means that in a CD-SEM,
repeated measurements cannot be done without extra, sometimes
unacceptably high measurement errors resulting from
"carry-over." During the time necessary for even one
measurement, the sample dimension can change, and the extent of this
change remains unknown unless a suitable contamination deposition
measurement technique is found and regular monitoring is implemented.
This paper assesses the problem of contamination of carbonatious
materials in the SEM and shows a possible method for its measurement
and presents a promising solution to the contamination deposition problem.
Key words: scanning electron microscope, CD, SEM,
contamination, measurement, CD-SEM, lithography, metrology, accuracy, linewidth
1.0 INTRODUCTION
The deposition of electron beam induced hydrocarbon
contamination is a pervasive problem in scanning electron microscopy.
Deposition of contamination on a sample is typically an unwanted and
negative effect (although it has been controlled and used positively
to manufacture atomic force microscope (AFM) probes). Many times the
vacuum readings appear quite good but the electron beam still leaves
disturbing contamination marks on the sample. This means that, in a
CD-SEM, repeated measurements cannot be done without extra, sometimes
unacceptably high measurement errors resulting from
"carry-over." Carryover is the increase in dimension size
due to the deposition of hydrocarbon contamination deposited by the
electron beam on the edges of the structure being measured. The
hydrocarbon moieties being deposited are from a variety of sources
including the vacuum system, stage lubricants and the sample itself.
During the time necessary for even one measurement, the sample may
change, and the extent of this change is unknown unless a suitable
contamination measurement technique is found and regular monitoring
is implemented. In an earlier report, some of the causes of
contamination in laboratory scanning electron microscopes (SEMs) were
reviewed and potential solutions were presented (Postek, 1996). This
work expands upon that earlier study and provides an additional
active monitoring and control approach to solving this problem.
2.0 MATERIALS AND METHODS
Throughout this study a Hitachi2 S-4700 laboratory
SEM, a Hitachi S6280-H CD-SEM and an S-806 tilt SEM were used. The
two latter instruments were equipped with 150 mm wafer stages. The
CD-SEM was equipped with two turbomolecular pumps and two Ebara dry
pumps. The tilt SEM had one turbomolecular pump and two rotary pumps.
The S-4700 laboratory SEM was equipped with the factory-installed
pumps: one diffusion and two rotary mechanical pumps. The samples
used in the study were clean wafers or diced chips from wafers made
with various processes used in silicon integrated circuit technology.
In this study the rate and amount of contamination deposited on these
samples were investigated and some of the possible methods of
reducing the effects of contamination, including a new anti-contamination
apparatus, called Evactron™, were also explored.
2.1 Evactron™. The Evactron2
is an automatic plasma cleaning and vacuum monitoring system. The
system can measure the vacuum level and by the use of valves control
the pressure of the gases introduced into the chamber needed for
plasma cleaning. It has a built-in power supply to drive a
plasma-generating head. Figure 1 shows the schematic diagram of the
Evactron cleaning head. The unit is mounted on the wall of the SEM
sample chamber and has a controller that can be configured to
automatically control the entire cleaning process. The cleaning
procedure is performed at much higher pressure (40-100 Pa) than the
normal operating pressure (10-3 - 10-4 Pa) of the specimen chamber.
The cleaning cycle starts with the closing of the necessary valves to
separate the specimen chamber from the electron optics. In the best
case, the specimen chamber can be simply separated from the
turbo-molecular or diffusion pump. In other cases, depending on the
design of the vacuum system the procedure will vary. The next step is
to let gas (filtered, clean, room air or oxygen-argon mixture for
faster cleaning) into the specimen chamber and stabilize the pressure
at its optimal value of 80 Pa. After reaching this point, the high
(13.56 MHz) frequency power is applied to the cleaning head for a few
minutes. The power applied and time duration depends on the size of
the chamber. This procedure provides a gentle plasma cleaning. The
nascent oxygen present in the chamber quickly reacts with the
residues in the vacuum, and the products are pumped out. To
accelerate this step it is advantageous to use a clean N2 flush. In
the case of an oxygen - argon gas mixture this step is mandatory.
This procedure can easily be made fully automatic, thus the user only
has to start the unit and wait until it has finished and the SEM is
ready for its regular work schedule.
2 Certain commercial equipment is identified in this
report to adequately describe the experimental procedure. Such
identification does not imply recommendation or endorsement by the
National Institute of Standards and Technology, nor does it imply
that the equipment identified is necessarily the best available for
the purpose. EVACTRON is a trademark of XEI Scientific.
Figure 1. Schematic diagram of the Evactron cleaning head
3.0 RESULTS
Figure 2 illustrates the contamination deposition
behavior of two CD-SEMs. Each instrument examined the same UV
photoresist wafer. The instruments were programmed to go to a
specific isolated line and without moving the sample perform 50
(so-called static repeatability) measurements. While one instrument
was able to report the 50 repeated measurements to within less than 2
nm change in the linewidth, the values obtained with the other tool
shifted close to 8 nm. Typically, linewidth measurements are done
from repeated line scans or averaged images (which exhibit similar
behavior). Thus, the deposition of contamination can invalidate the
data since an SEM with a severe contamination problem is not capable
of measuring the line without changing the width. This may occur even
after a single scan. The case of a positive carry-over is especially
suspicious, because that may be a sign of serious contamination.
Precise measurements require the measurement and understanding of the
contamination rate. Without measuring the rate of contamination,
which in effect is a measurement error, valid dimensional
measurements cannot be done. Furthermore, a thick layer of
beam-induced contamination can act as a resist and those locations
that were measured with the SEM may not etch at the same rate as
similar undisturbed areas during subsequent processing.
It is suspected that the contamination layer formed by
the electron beam comes from two sources: the vacuum and the sample
surface. Reimer (1993) describes a process of drift of large
molecular weight molecules under the electron bombardment. The
deposition of the contamination layer is a dynamic process. Molecules
arrive at and leave the sample surface at the same time. The amount
of contamination deposited depends on the electron dose, (i.e. the
length of time the beam dwells on the sample). The longer the dwell
time, the thicker the contamination becomes. Figure 3 and Figure 4
illustrate this process. In the case of Figure 3, the (5 kV, 10 pA)
electron beam was left on a clean, etched silicon sample for two
hours at high magnification. Viewed at lower magnification, the very
heavy deposition of contamination is formed as a "sculpture"
with a peak at the upper left corner. The electron beam created the
frame of the structure because it dwelled longer at the edges before
the sync pulses arrived and initiated the horizontal and frame scans.
Furthermore, just before the frame scan began the electron beam
dwelled a bit more at the upper left corner, therefore the beam
remained stationary at that point for the longest time overall and
formed a somewhat taller contamination peak (not unlike an e-beam
deposited AFM probe). The structure appears distorted because the
beam and/or the sample stage drifted during the long irradiation
time. The presence of the vertical line (originally the left edge of
the frame) near the middle of the contamination-induced frame clearly
proves that once contamination has been deposited, the electron beam
cannot remove it. If viewed in real time at high magnification, the
operator does not readily see the contamination. The image that is
displayed on the viewing screen is about 20% smaller than this frame
hence, the electron beam typically over-scans a larger area on the
sample than the electron beam displays on the screen. Therefore, the
most disturbing part of the contamination deposited at high
magnification is that the effect of contamination is typically not
observed by the operator except when going back to lower magnifications.
Figure 3. Contamination formed on a silicon wafer
sample during 2 hours of continuous electron beam bombardment. The
left image was taken with 60° sample tilt and the 3-dimensional
structure of the contamination can be observed.
Figure 4 shows an etched silicon "grass"
sample at 50 000 times magnification. This sample was bombarded with
a 5 kV, 10 pA electron beam at 500 000 times magnification. The area
shown on the screen of the SEM at this magnification is marked close
to the center of the image. The 500 000 times magnification image is
inserted at the lower right corner. The typical frame-like
contamination mark is not visible any longer; the dynamic process
formed only a circle at this very high magnification. The electron
beam was left on the sample for 10 minutes.
Figure 4. Etched silicon "grass" sample at
50 000 times magnification, which was bombarded with a 5 kV, 10 pA
electron beam at 500 000 times magnification
Figure 5 shows two images taken with CD-SEM. In this
case, the (800 V, 3 pA) electron beam bombarded the sample for 6
minutes. During this time the sample/beam drift and the contamination
resulted in the enlargement of the corner section by about 0.15
micrometer. Similar effects can take place in shorter times, in
severe cases even in a few seconds.
The extent that deposited contamination contributes to
measurement errors in CD-SEMs will remain unknown unless the operator
is aware of this problem and regular correct examinations and
measurement of the contamination rate is conducted.
Figure 5. X and Y enlargement of the corner region due
to contamination and drift in a CD-SEM. 800 V, 3 pA electron beam
bombardment for 6 minutes.
3.1 Anti-contamination Devices. Clearly,
electron beam induced contamination is a real problem in scanning
electron microscopy and no instrument is totally free of this
problem. Even dry pumped instruments can deposit contamination
(generally at a much lower rate). Contamination also results from
hydrocarbons brought into the proximity of the electron beam by the
specimen itself. Regular contamination rate measurements may reveal
the extent of the problem, but if the errors due to contamination are
too high, something has to be done to lessen the problem. One
possibility with certain SEMs is the use of a liquid nitrogen
anti-contamination device. An anti-contamination device is
essentially a small, flat metal piece located above the sample
surface that is kept at liquid nitrogen temperatures. The
anti-contamination device, by working essentially as a getter pump or
cold trap, collects the good part of the contaminants. Therefore, the
localized sample contamination rate, at the sample, is reduced. It
should be noted however, that once the device is allowed to warm-up
the trapped contaminants are released. Figure 6 shows an etched
silicon "grass" sample viewed at 50 000 times magnification
in an instrument equipped with a liquid nitrogen anti-contamination
device. This sample was bombarded two times at two different
locations at 100 000 times magnification for 10 minutes (@ 5 kV, 10
pA). The left image of Figure 6 was collected without liquid nitrogen
added to the device, and the right image was taken with the same
conditions but with the anti-contamination device fully charged with
liquid nitrogen. The amount of contamination is certainly reduced,
but it is still remains high.
Figure 6. Two locations of a silicon "grass"
sample irradiated for 10 minutes without (left) and with (right) the
application of liquid N2 to cool the anti-contamination device (50 000x).
3.2 Evactron Cleaning. Figure 7 shows two images of
the same silicon "grass" sample viewed in the same scanning
electron microscope as Figure 6. For the purposes of easy comparison,
the left image is the same as the image on the left of Figure 6 but
the right image was taken after the Evactron unit was turned on and
the cleaning procedure applied for ten minutes. On the right side of
Figure 7, where the Evactron anti-contamination device was used, the
amount of contamination is greatly reduced. The amount deposited is
also far less than even with the liquid nitrogen-cooled anti-contamination
device.
The SEM that was used to take the images in Figure 7
was a cold field emission gun instrument. It was equipped with a
liquid nitrogen cold trap above the water-cooled baffle on top of the
diffusion pump. This instrument was also equipped with a gaseous nitrogen
leak system where needle valves in the fore lines of
the mechanical pumps are set to about 2 Pa. This intentionally
inhibits the rotary pumps from reaching their ultimate vacuum
(Postek, 1996). This leak is small enough keep the pump efficiently
backing the diffusion pump or pump-down the specimen exchange chamber
with no compromise to the ultimate chamber vacuum. But, the
continuously streaming of nitrogen molecules into the line minimizes
the potential of backstreaming of oil from the rotary pumps. This
simple system provides an effective mechanism for reducing
instrument-induced contamination but in many cases , as shown here,
may not be sufficient.
Figure 7. Two locations of a silicon "grass"
sample irradiated for 10 minutes before (left) and after (right) the
use of Evactron anti-contamination device. 50 000x
CD-SEMs are typically equipped with
turbomolecular pumps and oil-free fore line pumps to provide better,
cleaner vacuum. Even with these pumps, many SEMs are not clean enough
and disturbing contamination deposition occurs. Regular monitoring
and if necessary, instrument cleaning has to take place. For
monitoring two possible ways can be followed: 1) measure the rate of
contamination with the Evactron, which shows the cleanliness of the
vacuum, or 2) measure the effect on dimensional measurement by
measuring the change occurring during repeated measurements. The rate
measurement is more advantageous in general, but it may not be as
applicable for all samples. This is because different samples made of
different materials are prone to contaminate at different rates and
the samples themselves are sources of the contaminating molecules as
well. Another possibility for sample dimensional change that must be
recognized is that the sample under test can become distorted by the
electron beam exposure itself (Postek et al., 1989). So, the
distinction between change due to contribution from contamination
deposition and electron beam induced change must be properly assessed.
After several Evactron treatments, the vacuum begins
to stay clean for longer periods of time and the frequency of
necessary Evactron treatments becomes less. Nevertheless, since the
cleanliness of the vacuum depends also on the nature and cleanliness
of samples going through the system, regular monitoring of the
contamination rate can indicate when another cleaning cycle must take
place. Figure 8 shows an image taken with a tilt wafer SEM. Prior to
the Evactron cleaning, it was impossible to work without severely
contaminating the sample. The instrument underwent a series of 10-minute
Evactron cleanings. After the cleanings, only a small amount of
contamination developed under the electron beam. This is illustrated
by the light contamination mark on the clean, freshly etched
polysilicon sample.
Figure 8. Tilt wafer SEM. Minimal contamination mark
on a polysilicon sample after treatment, 15 minutes dwell time at 30
000x magnification, @5 kV.
3.3 Sample Cleaning. Once contamination has been
deposited on a sample it is possible to remove it to some extent with
in situ oxygen plasma cleaning using the Evactron. Figure 9
illustrates this cleaning procedure on a silicon "grass"
sample. Both of two locations of the sample were irradiated for 10
minutes. The left image of Figure 9 is the untreated sample. The
image on the right of Figure 9 was taken after a 60-minute treatment
of the sample with the Evactron anti-contamination device. It is
important to point out that long treatment of the sample may also
alter the "clean" areas as well.
Figure 9. Silicon grass sample that was irradiated for
10 minutes. The left image was taken after contamination deposition
and is shown untreated. The image on the right was taken after a
60-minute in-situ treatment of the sample with Evactron
anti-contamination device. 50 000x
4.0 CONCLUSIONS
Contamination of various samples under the electron
beam has been demonstrated to cause measurement errors in laboratory
and CD-SEMs. The extent of the error is unknown unless regular
monitoring of the contamination rate is implemented. Depending on the
severity of the problem, removal of the contaminating molecules must
take place. This paper described a new cleaning method using an
active plasma system that was found to be effective in cleaning the
vacuum of the specimen chamber of laboratory, and production
metrology SEMs.
5.0 ACKNOWLEDGEMENTS
The authors would like to thank International SEMATECH
and the Office of Microelectronics Programs for providing partial
funding for this work.
6.0 References
Postek, M. T. 1996. An approach to the reduction of
hydrocarbon contamination in the SEM. SCANNING 18:269-274
Postek, M. T. 1989. Scanning Electron Microscope-based
Metrological Electron Microscope System and New Prototype SEM
Magnification Standard. Scanning Microscopy 3(4):1087-1099
Reimer, L.: Image Formation in Low-Voltage Electron
Microscopy SPIE Opt. Eng. Press; Volume TT 12 (1993) ISBN 0-8194-1206-6
RF Plasma Cleaning Systems for Electron Microscopes
and High Vacuum Systems
Stops Artifacts and Removes Hydrocarbons and Organics.
András E. Vladár, Michael T. Postek and
Ronald Vane*
*XEI Scientific Redwood City, CA 94063
Proc. SPIE Vol. 4344 (2001), 835.
Contribution of the National Institute of Standards
and Technology, not subject to copyright.
Figure 2. The results of 50 repeated line width measurements with two
different CD- SEMs with the same UV photoresist wafer.
