Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology
Подождите немного. Документ загружается.
54 Chapter 1.8: Surface Physics and Its Relation to Vacuum Science
Electron-stimulated gas desorption or atomic desorption was discovered during
the manufacture of electron vacuum tubes and was used in part to clean the metal
structures in the glass or metal envelope. The methodology was developed as a
production cleaning technique involving both thermal and electron bombardment
desorption. This was performed with little physical understanding of the physics
and chemistry of the desorption process. For example, electron-stimulated clean-
ing is still used in ultra-high-vacuum nude pressure gauges where this is the only
method of removing residuals from the metal structure.
Electron beam bombardment as well as photon absorption by polymers can re-
sult in decomposition of the material in addition to electron-induced desorption
of surface species
[17,19].
These materials are generally not used where they might
be exposed to high-energy particle environments. Energy near the bond strength
of a typical carbon-to-carbon bond is needed to break the bond. Under appropri-
ate conditions, the material can decompose in a dramatic fashion through coulom-
bic explosion [17]. In essentially the same fashion, electron bombardment can
break bonds, resulting in fragmentation of the molecule that can be released into
the vacuum.
Ion-stimulated desorption of adsorbed species is similar to electron beam-
induced desorption. Ion yield is larger than electron-stimulated desorption yields
as a result of the increased particle mass. This class of desorption event is related
to physical sputtering when the energy is great enough (see later discussions in
this book). Typically, the energy must be above the threshold energy for sputter-
ing (~10 electron volts, eV) to produce desorption. When the energy is higher,
molecular fragments are produced and can be released from the absorbate. This
mode is what is typically found in a static SIMS (secondary ion mass spectrome-
try) experiment employed in surface analysis of materials (see Part
5
of this book).
Ion bombardment is common when ion gauges are degassed to produce a pres-
sure rise in the vicinity of the gauge. Oxygen glow discharge cleaning is com-
monly used for substrate, vacuum chamber, and fixture cleaning. In this case oxy-
gen ions bombard the surface along with neutral bombardment.
1.8.13
GAS RELEASE FROM SURFACES
The mechanisms listed—vaporization, thermal desorption, photoactivation, elec-
tron- and ion-stimulated desorption, and outgassing through diffusion (or permea-
tion)—all produce a gas load on a vacuum system being pumped from atmo-
spheric pressure. O'Hanlon describes the rate-limiting steps in pumping a vacuum
chamber as volume, surface desorption, diffusion, and permeation. They occur in
sequence. Figure 11 shows a sketch of O'Hanlon's Figure 6.5 [1].
References
55
REFERENCES
1.
J. F. O'Hanlon, A User's Guide to Vacuum Technology (Wiley Interscience, New York, 1980).
2.
A. H. Turnbull, R. S. Barton, and J. C. Riviere, An Introduction to Vacuum Technique (Wiley,
New York, 1962).
3.
L. Ward and J. B. Bunn,
Introduction
to the Theory and Practice of
High Vacuum Technology
(Butterworth's, London, 1967).
4.
G. F. Weston, Ultra High
Vacuum
Practice (Butterworth's, London, 1985).
5.
G. L. Weissler and R. W Carlson,
Vacuum
Physics and Technology (Academic Press, 1979).
6. A. W Adamson, Physical Chemistry of Surfaces (Wiley Interscience, New York, 1994).
7. G. A. Somorjai, Introduction to Surface Chemistry and Catalysis (Wiley Interscience, New
York, 1994).
8. G. L.Weissler and R. W Carlson (eds.).
Vacuum
Physics and Technology, Vol. 14 of Methods of
Experimental Physics (Academic Press, San Diego, CA, 1979).
9. S. Roy Morrison, The Chemical Physics of Surfaces (Plenum Press, New York, 1978).
10.
R. Gomer, Field Emission and Field Ionization (Harvard Univ. Press, Cambridge, MA, 1961).
11.
J. H. de Boer, The Dynamical Character of Adsorption (Oxford Univ. Press, New York, 1953).
12.
S. Brunauer,
The
Adsorption of Gases and
Vapors,
Vol. 1. (Princeton Univ. Press, Princeton, NJ:
1945);
S. Brunauer, R H. Emmett, and E. Teller, 7. Am. Chem. Soc, 60, (1938) 390.
13.
For example, see, J. Crank, The Mathematics of Diffusion (Oxford Univ. Press, London,
1970);
D. Gupta in
Diffusion
Phenomena in
Thin
Films and Microelectronic Materials, edited
by D. Gupta and R S. Ho (Noyes, Park Ridge, NJ, 1988), Chap. 1.
14.
T. E. Madey, Radiation damage in auger electron spectroscopy and x-ray photoelectron spec-
troscopy, in Analytical Electron Microscopy, edited by D. C. Joy (San Francisco Press, San
Francisco, 1987), pp. 345-349.
15.
T. E. Madey and John T. Yates, Jr., J.
Vac.
Sci. Technol., 8 (1971) 525.
16.
D. A. King,
Surf.
Sci., 47 (1975) 384; M. W Roberts and C. S. McKee, Chemistry of the
Metal-Gas Interface (Clarendon Press, Oxford, 1978).
17.
T. A. Carlson, The coulomb explosion and recent methods for studying molecular decomposi-
tion, in Desorption
Induced
by
Electronic Transitions
DIET,
edited by N. H. Tolk, M. M. Traum,
J. C. Tully, and T. E. Madey (Springer-Verlag, Berlin, 1983).
18.
A. Berman, Water vapor in vacuum systems. Vacuum, 47 (1996) 327-332.
19.
T. Kelen, Polymer Degradation (Van Nostrand Reinhold, New York, 1983), Chap. 7.
CHAPTER
2.1
Technology of Vacuum
Pumps—An Overview
Marsbed Hablanian (Retired]
Varian Associates
2.1.1
VACUUM PUMP FUNCTION BASICS
Under normal environmental conditions, creation of vacuum requires a removal of
gases,
contained in atmospheric air, from a given vessel or a chamber. Thus, high-
vacuum technology involves the rarefaction of atmosphere by as much as 15 or-
ders of magnitude (from 10^ to 10"^^ mbar). This would be relatively simple if
the chamber walls did not have any internal gas content and any adsorbed gas
at the surfaces. The continuous evolution of occluded and adsorbed gases usually
necessitates continuous pumping even if the object is only to maintain a given
high vacuum level inside a chamber, such as the space between the double walls
of a thermos bottle. Even in this case, after evacuation the sealed space must con-
tain some residual pumping means, such as chemically active evaporated metallic
films.
There are a few basic ways to remove air (or another gas) from the space of a
vacuum chamber.
1.
Gas can be removed by an action of positive displacement devices, such as re-
ciprocating pistons, rotary pistons, rotary vanes, gears, cams, meshed screws,
orbiting scrolls, and so on.
ISBN 0-12-325065-7 Copyright © 1998 by Academic Press
$25.00 All rights of reproduction in any form reserved.
59
60 Chapter
2.1:
Technology
of
Vacuum
Pumps—An Overview
2.
Another way
is by
partial displacement
or
momentum transfer, such
as
fast-
moving blades
(fan or
turbine),
or a
high-velocity
jet
stream
of a
vapor
or
other motive fluid (ejector).
Pumps based
on
these methods
are
often called throughput pumps, because usu-
ally
the
gas evacuated from the vacuum chamber
is
compressed and exhausted
to
the atmosphere. Pumps based
on the
following removal methods
are
called cap-
ture pumps.
3.
Vacuum can also be created by refrigeration
to
temperatures
at
which the
gas
either becomes solid
or can be
adsorbed
on the
surface
of a
suitable porous
material.
4.
Gas can be removed from space by chemical reaction, which produces
a
solid
residue.
5.
Finally, the gas can be ionized and expelled from space either by the action of
suitable electromagnetic fields
or
embedded (implanted) into
a
solid surface
inside
the
pump
due to the
high velocity achieved
by
acceleration
in a
high-
voltage field.
The word
"pump,"
usually associated with liquids, is only of historical significance
in high-vacuum technology.
The
displacement devices could
be
called more
de-
scriptively "rarefied gas compressors" and capture devices, perhaps, "condensers."
Generally,
it is not
easy
to
design
a
single pumping device that would function
effectively under conditions
of
density changes involving 15 orders
of
magnitude.
All pumping devices have pressure
or
density ranges
in
which they have reason-
able performance, size, weight, noise level, cost, etc. Typically,
at
least
two dif-
ferent pumping devices
are
used
to
obtain high vacuum. They
can be
conve-
niently designated
as
coarse vacuum pumps
and
high-
(or
fine) vacuum pumps.
Coarse pumps are used to remove the bulk
of
the gas from
a
chamber (the gas that
was
in the
space
or
volume),
and
high-vacuum pumps
are
used
for
maintaining
vacuum
by
removing
the
gas evolving from
the
surfaces. Very roughly speaking,
these two types
are
associated with the conditions
of
viscous
and
molecular flow.
Compared to the ordinary compression work near atmospheric pressures, high-
vacuum gas flow conditions are characterized
by
very high volumetric flows,
ex-
tremely small mass flows, rather high pressure ratios, and very small pressure
dif-
ferences.
The
onset
of
molecular flow produces additional complications. First,
under molecular flow conditions, pressure differences
are
very small (although
significant density gradients usually
are
present). Thus, individual molecules
do
not travel into
the
pump following
the
direction
of
reducing pressure,
but
enter
the pump
as a
result
of
their normal thermal motion. Therefore, when molecules
rebound from
a
surface they
do not
"know" which
is the
downstream direction,
and they have
an
equal chance
to
move toward
or
away from the pump.
An
addi-
tional complication
of
molecular flow
is the
separation
of
flow behavior
of
differ-
ent gases
in a
mixture. Because
of
the absence
of
gas
to gas
collisions, different
2.1.2 Gas
Transport:
Throughput 61
gases may be pumped with different pumping effects; that is, different volumetric
flow rates and different compression ratios.
The absence of pressure gradients produces lower values for maximum pos-
sible volumetric flow through a restriction, such as an orifice. In viscous flow the
volumetric flow rate is nearly twice as high as that for molecular flow (for air,
near 20 L/s cm^, versus 11.6 L/s cm^ at room temperature). In general, if the
orifice diameter is less than or equal to the mean free path, the 11.6 L/s figure ap-
plies [1]. Also, relatively very high pressure ratios may exist in molecular flow
across a rather short flow duct. For example, an ordinary elbow may produce 50%
reduction in volumetric flow compared to a few percent in viscous flow. There-
fore,
the conductance of ducts in high-vacuum system design becomes important.
The concept of conductance (instead of resistance or friction coefficients in con-
ventional fluid mechanics) is described in detail in Section 2.4 and Part 1.
The units of conductance are usually given in "liters per second" (L/s) for sim-
plicity. However, the correct units are "torr L/s per torr" or, in other words, torr
L/s of throughput per torr of pressure difference.
2.1.2
GAS
TRANSPORT:
THROUGHPUT
The actual amount of gas flowing through a duct or into a pump cannot be deter-
mined by volumetric flow unless its density, 8, is also known. The more important
quantity is the mass flow expressed in units such as grams per second. Usually,
the conversion from volume flow to mass flow can be easily made using the well-
known gas law relationships
PV=RT
(1)
Q=V'8
where P = pressure, R = gas constant, T = absolute temperature, Q = mass flow,
and 8 = density.
However, these relationships may lose their usual meaning when the gas ap-
proaches molecular beam conditions, that may exist on occasions in high-vacuum
systems. It is not a simple matter to convert flow in grams per second into liters
per second for a gas that does not have isotropic molecular velocity distribution
(non-MaxweUian gas). For this reason, great precision and accuracy is often not
possible for considerations of design of most high-vacuum systems.
Because mass flows are very low, it is customary to use "throughput" instead of
mass flow. The assumption is made that heat transfer effects are negligible and
the gas quickly acconmiodates to the temperature of the vessel or pumping ducts.
62 Chapter
2.1:
Technology of
Vacuum
Pumps—An Overview
When temperature is constant, the "mass flov^" can be measured in units of
throughput, torr liters per second. Throughput then is simply the product of pres-
sure and the volume flow of
a
gas (Q = Sp) for any given location (cross-sectional
plane) in the flow passages.
Most analytical discussion of molecular flow found in textbooks on vacuum
technology is associated with noncondensable gases and steady-state flow con-
dition. Surface effects that govern the transient flow of condensable gases have
little effect on O2, N2, etc., at normal ambient temperatures and thus are often
neglected. (A detailed discussion is presented by Lewin [2].) Usually vapors of
higher molecular weight are troublesome, but water vapor and even such "vola-
tile"
substances as acetone or alcohol are difficult to work with at room tempera-
ture.
For example, if we attempt to measure the pumping speed of a high-vacuum
pump for water vapor, the measurements of the flow rate must be done by care-
fully considering the available surface area for evaporation from the liquid, heat
transfer necessary for evaporation, the temperature of the tubes leading the vapor
into the pump, etc. Even then it is difficult to separate the amount of water vapor
actually pumped by the pump and that which is adsorbed inside the cold inlet
areas.
For example, pumping-fluid vapor in a baked system may take a week to be
noticed by a high-vacuum gauge with a 4-cm-long, 2-cm-diameter entrance tube,
or a mass spectrometer can retain a strong "memory" of acetone for two or three
weeks if its ion source is 30 cm away from the pumps (2-cm-diameter tube). In
such cases, baking is indispensable.
In molecular flow environment, gas molecules rarely collide and, therefore,
move independently of each other. An outgassing or evaporating molecule can
move as easily away from the pump as well as into it. There exists a density (num-
ber of molecules) gradient toward the pump, but for condensable species, even
this cannot be assured, as temperature distribution may influence the flow more
than the presence of a pump.
It is often assumed that residual collisions between gas (or vapor) molecules
are completely negligible in high vacuum. This is not always true. Sometimes the
remaining collisions may be the predominant method by which certain molecules
pass through barriers such as cryogenic traps. Also, in some cases, introduction
of one gas into the system may affect flow conditions of another. For example, in
turbomolecular pumps, an improvement of density ratio for helium can be ob-
tained by introducing a certain amount of argon into the pump [3].
2.1.3
PERFORMANCE PARAMETERS
Performance of a vacuum pump or a compressor can be characterized, measured,
and displayed in several ways, depending on the interest in a particular quality.
2.13 Performance Parameters
63
Basic parameters are volumetric flow (known as pumping speed in high-vacuum
technology), mass flow (usually displayed as pressure-volume flow: throughput),
compression ratio, and pressure difference (between discharge and inlet). The re-
lationships between these parameters can be shown in several ways. For example,
we can plot pressure versus flow or flow versus pressure, where the flow can be
either volumetric or mass flow and the pressure term can be in the form of a pres-
sure ratio or pressure difference. The emphasis can be on inlet pressure or dis-
charge pressure, etc.
The most common traditional representation of performance has been in the
form of pumping speed versus inlet pressure. This is a somewhat unfortunate
choice because it is a reverse of the common practice in other engineering fields.
It tends to produce a conceptual confusion between the dependent and indepen-
dent variables, and it does not clearly distinguish between transient and steady-
state range of
the
operation. High-vacuum pumps are usually rated by their pump-
ing speed, the maximum permissible discharge pressure, the best obtainable
vacuum level (lowest possible absolute pressure), maximum obtainable compres-
sion ration, maximum mass flow capacity (throughput), and some secondary
characteristics related to the possibility of system contamination emanating from
the pump.
When the performance of a vacuum pump (or a compressor) is presented in the
form of a plot of volumetric flow (pumping speed) versus inlet pressure, the basic
limitations of the performance may be represented as shown in Figure 1. If this
Fig.l.
/ \ ^^
^ //
f //
//
;^
\ o 1
o 1
f ARRIVAL RATE AND \
CAPTURE LIMIT
Vv
-^ 1
V^'
INLET PRESSURE
Basic limitations of pump performance.
64 Chapter
2.1:
Technology of
Vacuum Pumps
— An Overview
graph were simply tilted 90 degrees to show pressure versus pumping speed, we
should be immediately suspicious of this double-valued function, which does not
resemble the usual representations of pump and compressor performance found
in the fields of mechanical engineering, fluid mechanics, or pneumatic machinery,
where the power limit section of the curve is normally omitted because it is not
really a part of intended performance; in other words, the performance range has
reached its power limit. However, in high-vacuum practice, the initiation of cross-
over from roughing to high-vacuum pumps can be made in this region as long as
the maximum flow capacity of the high-vacuum pump is not exceeded.
2.1.4
PUMPING SPEED
It is customary in high-vacuum technology to express gas flow in liters per second.
This is convenient in basic computations if the temperature is assumed to be con-
stant. There exist two distinct geometries in which the flow parameters are associ-
ated (Q = pS). One can be called orifice geometry (Knudsen geometry); the other,
pipe flow geometry. The two are not identical. In the case of the pipe
flow,
the pres-
sure and speed are associated with the same cross-sectional plane: In the case of
the orifice, the pressure is measured upstream from the orifice. The orifice geome-
try is shown in Figure 2. The definition of orifice conductance, C, is derived from
Q = C{P, -
P2)
(2)
where P| and
P2
are upstream and downstream pressures and Q is throughput. If
P2
is much smaller than Pi, we may speak of the orifice speed
S = C=Q/Pi when P2«^1 0)
Note that in this
case,
the Pi is not associated with the vicinity of the orifice, but with
an upstream location where the gas conditions may be considered to be isotropic.
As derived from kinetic theory of gases and confirmed by experiments, that for
air at room temperature, each square centimeter area of an orifice (having "zero"
thickness) will have a maximum pumping speed of approximately 11.6 L/sec.
When computing the number of molecules bombarding a small (compared to the
chamber) unit area at a given pressure (measured far away), we can arrive at the
same amount of 11.6 L/sec. Note that this 11.6 L/sec is a conversion from num-
ber of molecules and is associated with the remote, upstream density.
The two geometries just mentioned occur in practice. Often the geometry rep-
resents a mixture of the two. For pipe flow and for a sequence of matched flow
elements connected in series, the association of pumping speed at a cross section
is more convenient. For a chamber with a small pumping port, the orifice associa-
tion is more helpful. This duality of approaches is responsible for the difference
2.1.5 Pumpdown Time
65
Fig.
2.
Basic geometry defining the concept of orifice conductance.
between the older pumping speed measurement standards (American Vacuum
Society 4/1 and International Standards Organization 1607). The ISO standard
and the new AVS Recommended Practice tend toward the orifice treatment and
give 10-15% lower pumping speed values for a pump compared to the older AVS
standard. The two can be related, depending on the ratio of pumping port to cham-
ber diameter [4].
The net pumping speed of the pumping system or the pumping stack can be de-
termined by the use of
the
same standard measuring dome. To relate this measure-
ment to the vacuum chamber in order to obtain an effective speed for the chamber,
we should take into account a pressure difference existing between the center of
the chamber and the inlet into the pumping port. In practice, such considerations
are not of great importance, and simple approximate calculations can be made by
adding conductance in series as long as no abrupt changes in geometry are present.
In cases where pronounced anisotropy exists, pumping speed consideration
can become meaningless. Consider an imaginary molecular beam with parallel
molecular trajectories (without collisions) entering directly into a pump. A gauge
mounted on the wall in the vicinity of this beam may not know of
its
existence. In
such cases, the meaning oiQ— pS is not obvious. Situations of this nature may
arise in practice.
2.1.5
PUMPDOWN TIME
One of the basic considerations in designing high-vacuum systems is the time re-
quired to evacuate a vessel to a given pressure. Usually, this ranges from a few
66 Chapter
2.1:
Technology of
Vacuum
Pumps—An Overview
minutes to a few
hours.
However, in some cases, the desired time can be a few sec-
onds or a few days. The prediction of
time
and pressure relationship by theoretical
or experimental methods is very difficult because of the uncertainty associated with
gas evolution rates from inner surfaces of the vessel (outgassing or virtual leak).
Gas evolution from a given material depends on temperature, surface finish,
previous history of exposure to a variety of atmospheric conditions, and cleaning
methods. Thus, even approximate prediction of evacuation time can be extremely
difficult when outgassing becomes significant. Using well-known relationships
applicable to higher pressures, the designer can predict a pressure-time curve
from atmospheric pressure to about 10"^ torr. From steady-state (long time) out-
gassing data and from characteristics of pumping devices, the final (ultimate)
pressures can sometimes be predicted. The region of greatest interest often is be-
tween these two points. The following discussion attempts to develop at least a
qualitative appreciation of significant effects.
2.L5-1 Constant Speed Case
In the pressure region between 760 and 10 torr, the pumping speed of the fore-
vacuum pump is usually nearly constant. A 10% reduction at 10 torr compared to
the speed near atmospheric pressure is common. Referring to Figure 3, the evac-
uation process can be represented by the following relationship:
-Vdp/dt = Sp (4)
where V is the volume of the vessel, dp/di the rate of change of pressure with re-
spect to time, and S the pumping speed. The physical meaning of Equation (4) can
be understood by noticing that the left side represents the amount of gas leaving the
chamber (minus sign indicates pressure decrease), and the right side shows the gas
entering the pumping duct. It is important to note that a solution of this differen-
tial equation will depend on the location where the pressure is measured, Pi,
/?2»
or
P2,.
Otherwise, to be valid, the equation must have the same pressure value on both
sides of the equality sign. Thus, the equation can be used as an abstract definition
of a pumping speed for the given chamber, pumping system, and gauge location.
If this pumping speed is plotted versus the inlet pressure using log-linear coor-
dinates, a straight line will be obtained in sections where speed is constant. The
slope of this line represents the pumping speed, as related to the pressure reduc-
tion rate in the chamber (5 = A In
p/^t).
The evacuation time from Equation (4) for constant volume and speed is
t = (V/S) In {Po/P\ or t = 23(V/S)\ogiPo/P) (5)
where
PQ
is the initial pressure and P the final pressure. This equation gives ade-
quately accurate values after a few seconds from the start of evacuation and until
approximately 10 torr is reached. At lower pressures the outgassing becomes