XEI Scientific, Inc.
The EVACTRON® Anti-Contaminator and De-Contaminator Updated August 2007
Reducing Oil Backstreaming in
by Ronald Vane, XEI Scientific
INTRODUCTION
Of the sources of contamination,
vacuum pump oil backstreaming is the easiest to attack. The original
version following report was written before I founded XEI Scientific
to provide Electron Microscopists more satisfactory methods stopping
contamination. Because of this the SEM-CLEAN Nitrogen purge system
and Evactron Decontaminator system are not mentioned as solutions.
Dry Pump technolgy has also advanced since 1990, and there are many
more models than mentioned below that are available from various
manufacturers. -R.V.
Some types of Electron Microscopes are standard
equipped with Oil Diffusion Pumps for High Vacuum and Oil Sealed
Rotary pumps for backing and roughing. Under some conditions oil
backstreaming can be a problem to the user. This oil can cause sample
contamination, oil films on detector windows, dirty apertures, and
other problems for the microscopist. A number of methods can be used
to attack this problem with a choice based on the cost, maintenance
schedule, reliability, and achievable vacuum.
BACKSTREAMING
To understand the solutions to the backstreaming
problem it is important to understand the types of flow found in
vacuum systems. In rough vacuums above about 10-2 mbar ( 1 Pa) we
find viscous flow conditions. In viscous flow the gas acts like a
fluid and moves from areas of high pressure to those of low pressure.
Each molecule moves with the molecules around it. If the forward flow
rate is faster than the backwards diffusion rate there can be no
backstreaming in viscous flow conditions.
At high vacuums below about 10-5 mbar ( 10-3 Pa) we
find molecular flow conditions. The mean free path between molecular
collisions becomes long enough that each molecule acts independently
of the movement of surrounding molecules. For this reason high vacuum
pumps act by capturing the molecules that come in and prevent them
from returning to the vacuum chamber. To pump efficiently under
molecular flow conditions wide pump openings are needed to capture
the maximum number of molecules. At pressures between the pure
viscous and molecular flow conditions there is a transitional flow
region (Knudsen flow) where flow characteristics have properties of
both. In this region the back diffusion rates become larger and
viscous flow becomes less effective at preventing backstreaming as
the pressure drops.
As pressure drops, backstreaming (basically a
molecular flow phenomenon) becomes more of a problem for pumping
oils. The prevention of backstreaming basically becomes that of
stopping the back migration of oil molecules to the sample chamber
while the system is in transitional or molecular flow conditions.
Surface creep of pump oil along the surfaces of the vacuum system
must also be stopped. The basic ways to do this includes traps,
changing types of pump oils, operational methods, foreline pressure
modification, or the use of dry vacuum pumps.
Once a vacuum system is contaminated with oil,
all surfaces which are oily will become secondary sources of oil
within the system. Installing a corrective measure on the primary
source of oil will not correct any secondary sources in the system.
These secondary sources must be removed by cleaning.
VACUUM PUMP OILS USED BY HITACHI
Hitachi microscopes come from Japan with Lion S oil in
the diffusion pump and Neovac MR-100 in the rotary pump. Because of
the cost and difficulty involved with obtaining large quantities of
these oils from Japan, HII has for some time now used Santovac 5 D.P.
Oil and TKO-19 (Kurt J. Lesker company) rotary pump oil as
replacement pump fluids. Both of the D.P. fluids have vapor pressures
in the 10-10 torr range but are very different materials. Santovac 5
is a polyphenylether fluid a with very high viscosity at 25 C and
Lion S is a Alkyl-naphtalene with much lower viscosity. In addition
Lion S oil works best if the cooling water in the DP does not rise
above 25 C and Santovac prefers a slightly warmer cooling water
temperature because of its extreme viscosity at 25 C or below.
The rotary pump oils have vapor pressures of about
10-6 mbar at 25 C. If the DP fluid becomes contaminated with a
significant amount of rotary pump oil, then the RP oil will distill
out of DP oil during pumping and the ultimate vacuum of the DP will
be that obtainable by the RP oil and not that of the low vapor
pressure DP oil! Thus keeping the DP oil free of RP oil becomes
important in obtaining a low pressures and a clean vacuum. The oil
contamination found in several Hitachi microscopes has been analyzed
and identified as rotary pump oil and not diffusion pump oil.
There are two possible ways for this oil to reach the
sample chamber. One is by distillation out of contaminated diffusion
pump oil, and the other is by direct backstreaming through the
roughing line of the sample chamber or the exchange chamber.
Backstreaming takes place by either in the vapor phase by direct
evaporation or by surface creep of the liquid film. We will discuss
methods to stop the oil from reaching the sample chamber.
TRAPS AND BAFFLES
One of the oldest ways of preventing backstreaming is
to use a trap or baffle to prevent the oil from backstreaming. The
oil condenses into the trap which prevents it from moving further up
into the system. The many types of traps. On electron microscopes
traps are found between the sample chamber and sometimes between the
diffusion pump and on the foreline to the rotary pump.
Traps have the disadvantage that eventually oil
molecules will find their way past especially if the trap is not
cleaned or renewed regularly. Some traps will have a longer hold down
time than others but all traps need renewal regularly. The trapping
material in foreline traps eventually becomes saturated with oil or
water, and then it will act as simply as a secondary source for vapor
migration into the system. Traps slow backstreaming but without
renewal eventually all the surfaces of the vacuum system will be
coated with oil, and the lowest pressure obtainable in the system
will be determined by the vapor pressure of the most volatile oil in
the system. Trapping materials must be constantly maintained to be
effective. Maintenance consists of either baking the material,
cleaning, or replacement.
If a foreline trap is not being effective at stopping
oil contamination in a system, the first thing to do is to check the
condition of the trap and check the oil installed in the vacuum
pumps. If the trap is saturated with oil the trapping material should
be replaced and the rotary pump checked to make sure it is contains
clean oil of the correct type and that oil is not being sucked back
into the foreline. If the type of oil in either pump is unknown it
should be replaced.
A new or renewed foreline trap will not clean up
a contaminated vacuum system. Oil from previously contaminated
surfaces will continue to creep and evaporate within the vacuum
system after the trap is installed. This oil will eventually be
captured by the trap, but this is a process governed exponential
decay and has a long half life. A system cleaning is recommended to
remove oil contamination.
Foreline traps that require baking for renewal are not
suited for most electron microscopes because during the bake cycle
the released gases and oils must be prevented from condensing
elsewhere in the foreline. This requires either that the whole
foreline be baked and there be a bakeable valve to seal the system
from the trap, or that dry nitrogen be bled through the trap during
baking to carry the released gases to the pump. Foreline traps for
electron microscopes should have a replaceable or cleanable filter material.
There are number of types of foreline traps available.
We will discuss the common traps and their advantages and disadvantages.
Molecular Sieve Traps:
These traps work by adsorbing oil into a molecular sieve material
which is very porous and has a very large surface area. To work
effectively the molecular sieve must be baked out under vacuum to
remove accumulated gases, water, and oil. This baking must be done
frequently maintain the effectiveness of the trap. These traps prefer
to trap water vapor over oil, and when they are full of water they
are no longer effective for oil. Hitachi does not recommend molecular
sieve traps because they must be baked to be refreshed and because
water vapor makes them ineffective.
Copper Gauze Traps:
Copper gauze reacts chemically with the oil vapor to trap it. In
addition the gauze provides a high surface area and an optically
dense path against molecular flow. When the copper gauze turns green
and gunky it should be cleaned or replaced. The use of a transparent
trap housing such as the Varian 346 is recommended because the
condition of the copper gauze can be monitored from the outside by
the user. This type of housing should be mounted so that any oil
droplets that pass will puddle in the bottom of the trap. Depending
on the system this trap will require maintenance every 2 to 6 months.
Replacement copper gauze is easily found in the form of copper pot
scrubber pads in the super market. These should be cleaned with a
solvent to remove oil before use.
Stainless Steel Gauze Traps:
These traps work only by providing a high surface area and a
optically dense path against molecular flow. It is most effective
against simple aerosols rather than the oil vapors. Unlike the copper
gauze they do not react chemically with the oil vapor. Therefore
their capacity is less than that of the copper gauze trap. Therefore
these traps will need maintenance every 1 to 2 months to clean or
replace the gauze. This type is not recommended.
Micromaze Trap
(Kurt J.
Lesker Company): This is a very effective foreline trap that uses a
proprietary ceramic sheet material with many pores and a high surface
area and acts like a molecular sieve in its ability to trap large or
polar molecules such as water and hydrocarbon oils. To be effective
it must be baked at 200 C for several hours before use and when it
needs to be refreshed. Because it is baked in-situ this trap is not
recommended for most systems.
In the mid 1990s experiments by David Joy of the
University of Tennessee suggested that a continuously heated
Micromaze trap would stop backstreaming. A trapping device using this
arrangement (heated micromaze) as then sold by Hitachi as an
accessory to their electron microscopes users to solve the
backstreaming problem. The additional heat makes in hard for
molecules to be adsorbed on the micromaze surfaces and slows down the
migration of oil through he trap. In the long run, however, this
arrangement did not prove satisfactory. David Joy no longer
recommends this device.
Activated Alumina:
This
material is very effective in preventing the back migration of oil
into the system and in trapping water and acid vapors before they
reach the pump. It has a long life and needs to be renewed only every
six months. Hitachi used activated alumina traps in the forelines of
Turbo Molecular Pumps (TMP) in the early 0s. For renewal the
activated alumina is poured out and replaced. Activated Alumina also
actively absorbs water and acid vapors which may interfere with
obtaining a good vacuum. Wet alumina will not pump down properly and
must be dried and regenerated. Wet alumina may be regenerated by
baking at 200 C outside of the vacuum system in an oven. Activated
Alumina is a caustic material and there are certain handling precautions.
Liquid Nitrogen: Liquid
Nitrogen traps are very effective for freezing out oil vapor. But
they require constant refilling to stay effective. LN2 traps are best
used in short term pumping situations where the pump is turned off or
valved off when the pumping is complete. For foreline use the a LN2
trap requires either a regular filling schedule of an automatic
filling system.
Traps and baffles for the diffusion pumps:
Hitachi microscopes with diffusion pumps are equipped with a water
cooled baffle mounted over the diffusion pumps. This baffle is meant
to condense diffusion pump oil and return it to the D.P. with a small
decrease in pumping speed. On some Hitachi Microscopes there is an
additional LN2 cryotrap mounted over the diffusion pump. These traps
need constant refilling to keep them effective.
FORELINE PRESSURE
Backstreaming in the foreline of the diffusion pump is
the result of the pressure in the foreline being so low that
molecular flow is possible for the oil molecules against the flow of
the pumped gasses. Higher foreline pressures bring viscous flow
conditions which act to keep the R.P. oil in the pump. If a rotary
pump is only used for roughing and not as a forepump, the pump can be
simply turned off to prevent backstreaming when a rough vacuum is
achieved. Where the R.P. is used as a forepump this is not possible
so other methods must be used to keep the fore pressure up.
One way to achieve higher foreline pressure and
viscous flow conditions is by the use of a Nitrogen leak into the
foreline which adjusts the pressure to one which causes viscous flow
but which is low enough for the D.P. to work against. The nitrogen
leak may be either a calibrated leak or an adjustable leak valve. Dry
nitrogen is used rather than air to avoid the introduction of oxygen
or water vapor into the system. This method is used in many types of
process equipment to reduce foreline backstreaming.
A detailed discussion of this technique was published
by Michael Postek of NIST. An Approach to
the Reduction of Hydrocarbon Backstreaming in Scanning Electron
Microscopes, Scanning Vol. 18, 269-274 (1996). Reprints
of this Article are available from XEI Scientific.
CROSSOVER PRESSURE
Related to the question of foreline pressure is that
of crossover pressure. Crossover pressure is the pressure at which
the roughing valve to the sample chamber is closed, and the DP valve
to sample chamber is opened. The crossover pressure can be adjusted
by field service engineers for most microscopes.
If the cross over pressure is too low, then the RP
will be connected for too long to the chamber and conditions
favorable to direct backstreaming of RP oil into the chamber will
result. If the crossover pressure is too high, then when the main
valve is opened into the DP from the chamber, the resulting turbulent
flow will upset the upper most jets of the DP and DP oil can come
into the chamber. For microscopes with a single pump for both
roughing and backing the DP there is the additional problem that at
high crossover pressures the contents of the roughing line can be
sucked into the DP foreline to contaminate the DP after crossover.
Therefore you may make a small adjustment upwards of
the crossover pressure to reduce the direct backstreaming of RP oil
into the chamber during roughing, but you must exercise care not to
raise the crossover pressure too high or the pumping action of the DP
will be disrupted releasing oil from the DP to the chamber when the
main valve opens.
ROTARY PUMP OIL SELECTION
Backstreaming from the RP can also be reduced by using
a pump oil with a lower vapor pressure. In addition if the Rotary
Pump oil and the diffusion pump oils are the same, the contamination
problem of the DP oil by the RP oil is eliminated. The diffusion pump
oil must have adequate lubrication properties to be used in rotary
pumps. Highly refined hydrocarbon oils meet this requirement.
Silicone oils such as those made by Dow Corning have very poor
lubricating properties and should not be used in RPs. Other oils such
as ethers, esters, and fluorinated hydrocarbons depend on individual characteristics.
Santovac 5 oil has a very high viscosity and should
not be used in an RP.
Lion S oil can be used in a RP successfully and has a
vapor pressure in the 10-10 mbar range. Unfortunately it is available
in only small quantities (in 100 ml bottles) and has limited
availability since it is imported from Japan.
Diff Oil 20 (Kurt J. Lesker Co.) has a vapor pressure
of 10-7 mbar and is suitable for both mechanical pumps and diffusion
pumps according to the distributor. It is of reasonable price and
readily available. While it will not give as low a ultimate vacuum as
uncontaminated Santovac 5 it should lower the overall pressure and
oil backstreaming in most Hitachi DP pumped systems by an order of
magnitude because RP oil contamination is eliminated.
If the DP and RP oils are replaced in this manner, the
whole vacuum system (Pumps, valves, foreline, and chambers) must be
cleaned of the old RP oil before replacing the pump oils. If you do
not do this, the old oil will continue to evaporate from the surfaces
of the vacuum hoses, valves, pumps and chambers and the contamination
will continue.
DRY VACUUM PUMPS
One of the cleanest ways to eliminate the
backstreaming problem is to use a dry vacuum pump. There are several
types available, but the type compatible with roughing sample
chambers and backing oil diffusion pumps is a molecular drag pump. A
drag pump is similar to a turbo pump but is designed to operate at
higher pressures than a TMP (Turbo Molecular Pump). Because there is
no oil in the pump, there is no backstreaming and contamination. The
diffusion pump is them able to reach a pressure limited by the
pressure of the diffusion pump oil vapor pressure alone.
The pump approved by Hitachi in 1990 for use on
electron microscopes is a Alcatel DRYTEL 30. The Drytel 30 is
molecular drag pump backed by a diaphragm pump. For roughing the
system the Drytel pumps is slower than the standard rotary pump. Some
cusomers may find it unacceptably slow. HII has available a DRYTEL
pump conversion package available containing this pump with a
conversion kit for connecting it to Hitachi electron microscopes.
This pump has been placed through extensive reliability tests to
insure that it does not fail though many S.C. and S.E.C. roughing
cycles on a SEM.
The Alcatel MDP 5011 molecular drag pump used in
the Drytel 30 may be backed by a rotary pump instead of the diaphragm
pump. This will significantly improve roughing speeds. In this case
the drag pump acts like an active trap on the foreline to stop
backstreaming. Unlike a passive foreline trap it does not need the
trap material renewed on a frequent schedule.
TURBO MOLECULAR PUMPS (TMP)
Turbo molecular pumps are the most expensive solution
to the oil backstreaming problem. The TMP replaces the diffusion pump
in the system. The Hitachi TMP package consists of a Seiki-Seiko mag-lev
TMP with activated Alumina Traps for the foreline and roughing line.
The TMP will prevent oil from backstreaming through it. But the
activated Alumina traps are necessary to prevent oil backstreaming
during the roughing cycles. Thus it is necessary to keep the traps
fresh and maintained on a regular cycle in order to keep the vacuum
system dry.
For the ultimate in clean vacuum a TMP should be used
in combination with a Molecular drag pump for roughing. In this way
there is no oil in the system to cause contamination.
CRYOPUMPS
Currently there are no cryopumps approved for use with
Hitachi electron microscopes because of vibration or reliability
problems. A suitable cryopump may exist but has not been submitted
yet to Hitachi for test and approval.
Since a cryopump is a pure capture type pump which
does not require continuous backing, on a cryopump equipped
microscope, clean vacuum operation could be achieved by operating the
rotary pump only during roughing operations and avoiding low pressure
molecular flow conditions. A cryopumped system would require
considerable change in the evacuation control logic and circuits to
function properly.
SUMMARY- STOPPING BACKSTREAMING
There is no one method for reducing oil backstreaming
in electron microscopes. The choice depends acceptable cost,
maintenance level, and degree of vacuum cleanliness desired. The
following methods are available for reducing oil backstreaming:
CHANGE CROSSOVER PRESSURE: Cost very low. No maintenance.
Vacuum quality - small improvement.
ACTIVATED ALUMINA TRAP: Cost low. Needs renewal every
6 months.
Vacuum quality - medium
COPPER GAUZE TRAP: Cost low. Needs renewal every 2 months
With transparent trap it can be visually checked.
Vacuum quality - medium
CALIBRATED FORELINE LEAK: Cost low - medium. Need
source of dry nitrogen. Low maintenance. Vacuum quality - medium.
CHANGE TO DIFF OIL 20 IN BOTH PUMPS: Cost low. Low
maintenance. No cross contamination of oils. Vacuum quality - medium-high.
MOLECULAR DRAG PUMP:
Cost medium - low. Acts as a
active foreline trap to stop backstreaming. With rotary pump produces
fast roughing but system is not totally oil free. Less maintenance
than a passive trap. Generates some noise. Vacuum quality - high-medium.
DRYTEL 30 PUMP: Cost medium. Replaces rotary pump.
Needs more time for roughing. Needs PM to replace diaphragm every 6
months. Generates some noise. Vacuum quality - high.
TURBOMOLECULAR PUMP: Cost high. Replaces diffusion
pump. Low maintenance. Vacuum quality - high.
SYSTEM CLEANUP IS REQUIRED to make any of the above
methods effective. Residual oil contamination in the vacuum system
will act as a secondary source of oil contamination on samples even
after primary corrective action is taken.
RF Plasma Cleaning Systems for Electron Microscopes
and High Vacuum Systems
Stops Artifacts and Removes Hydrocarbons and Organics.
Electron Microscopes