Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

2.4.3

Throughput

127

2A3

THROUGHPUT

2.4.3.1 Maximum Throughput

It may be observed that maximum throughput is often the important aspect of va-

por jet pump performance rather than pumping speed. The value of the maximum

throughput determines the amount of power required to operate a given pump. Di-

mensionally, throughput and power are equivalent. For pumps of current designs

using modern pumping fluids, approximately

kw of power is required to obtain

a maximum throughput of 1.2 torr L/sec.

For systems that remain under high vacuum for long periods of

time,

the maxi-

mum throughput is of little value. In such cases, provisions can be made to reduce

the power after the initial evacuation or to use lower-power heaters. Operation of

the pump at half power is possible without changes in pump design. Special de-

signs can be made for low-power (low-throughput) operation with appropriate at-

tention to corresponding reduction of forepressure tolerance.

For rapid frequent evacuation and for high gas load application, the value of

maximum throughput determines the choice of pump size.

If the graph in Figure 8 is replotted, interchanging the axis, a more customary

arrangement of dependent and independent variables is obtained, Figure

11.

Then

Fig.

11.

10''

10'

10"

(?) 10

10^

rTr~T 1 T--|

1 V >. Ny

r "^ ^v \

1 ^ N. \

I N )

^ ^

r ^"^

LIMITS OF

\- DIFFUSION PUMP

OPERATION

I 1 1 1

T ••"• I

H I

< H

< 1

to J

Q 1

< J

o 1

1 1 .

10*2 10'

I02 10^ 10"*

THROUGHPUT, TORR LIT/SEC

Input pressure vs. throughput.

128 Chapter 2.4: Vapor Jet Pumps [Diffusion Pumps)

it is simpler to see the maximum throughput and the pressure stability limits and

the concept of overload when those limits are exceeded. The more convenient

way of looking at this is to keep in mind that for a given system gas load presented

to the pump there is a resulting inlet pressure. This will help in selecting the re-

quired pump size and in distinguishing between the requirements of evacuating a

chamber and maintaining a desired operation pressure at a given process gas load.

2A4

TOLERABLE FOREPRESSURE

2.4.4.1 Discharge Pressure

A vapor jet pump is designed for high-vacuum application. Thus, its boiler pres-

sure usually is 1-1.5 torr. This implies that the maximum pressure difference the

pump can be expected to produce is 1.5 torr. In addition, vapor jet pump working

fluids cannot be boiled at high pressures because the resulting high temperatures

will decompose the fluid at a unacceptable

rate.

Thus, such pumps must always be

backed by another pump that produces a pressure of generally less than 0.5 torr at

the discharge of the vapor jet pump.

Tolerable forepressure of a vapor jet pump is the maximum permissible pres-

sure at the foreline.

The pumping action of the vapor jet pump collapses completely when the tol-

erable forepressure is exceeded [2]. Essentially, the vapor of the discharge stage

of the pump does not have sufficient energy and density to provide a barrier for the

air in the foreline when its pressure exceeds a certain value (usually near 0.5 torr).

Then this air will flow across the pump in the wrong direction, carrying with it the

pumping fluid vapor.

Approximately half of the initial pressure is recovered in the form of tolerable

forepressure. When the tolerable forepressure is measured, a conventional non-

ultrahigh vacuum system usually does not reveal the dependence between the

discharge and the inlet pressures. This is because the maximum pressure ratio for

air is usually not exceeded unless the experiment is conducted under ultrahigh

vacuum inlet pressures.

2.4.4.2 Backing Pump Requirements

To select an appropriate backing pump for a given vapor jet pump, several ques-

tions must be considered. First, what is the size of the initial roughing pump, and

2.4.4 Tolerable Forepressure 129

is it to be used for both roughing and backing? Second, is the backing pump ex-

pected to perform at the maximum throughput of the vapor jet pump? Third, what

is the tolerable forepressure of the diffusion pump? Also, what is the volume of

the foreline ducts (or a special reservoir or ballast chamber that could be used in

the line as a buffer during roughing)? The nominal pumping speed of

the

required

backing pump for full load condition is obtained as follows:

S = QurJiTFP),

where gmax is the maximum throughput of the vapor jet pump and TFP stands for

tolerable forepressure. Setting aside consideration of safety factors and conduc-

tances of forelines, it should be noted that mechanical pumps often have reduced

speed when their inlet pressure is equal to the tolerable forepressure of the va-

por jet pump. A safety factor of 2 may be required to make sure that the toler-

able forepressure value is never exceeded even if mechanical pump and vapor

jet pump are not operating at their best. A numerical example follows. Assume

a pump with a maximum throughput of 4 torr L/sec and tolerable forepressure

(maximum permissible discharge pressure) of 0.5 torr at full load (at maximum

throughput). The required backing pump speed then becomes

4 torr L/sec

s = —— = 8 L/sec « 17 CFM

0.5 torr

A good choice for the backing pump would be a nominal 14 liters/sec (30 CFM)

pump, provided that the conductance between the two pumps are not severely

limited.

2.4.4.3 Safety Factors

The most important rule of vapor jet pump operation is: do not exceed tolerable

forepressure! In other words, in an operating pump, the maximum permissible

discharge pressure should not be exceeded under any circumstance. Observance

of this most basic requirement will eliminate most of the difficulties encountered

with vapor jet pumps, especially problems with noticeable backstreaming of the

pumping vapor into the vacuum system. High-vacuum systems should be designed

with interlocks, fail-safe valve arrangements, or clearly marked instructions to

preclude the possibility of exceeding the tolerable forepressure. As in most en-

gineering considerations, a safety factor should be included in establishing the

maximum permissible discharge pressure. A factor of 2 is a good general recom-

mendation. As much as 25% reduction of tolerable forepressure can be expected

near maximum throughput operation (full load), and some reduction can be ex-

pected from low heater power. The selection of the mechanical backing pump

must be made with these points in mind. Also, it should be noted that the desired

130

Chapter 2.4: Vapor Jet Pumps [Diffusion Pumps)

discharge pressure must be maintained near the vapor jet pump to avoid errors

caused by limited conductance of forelines and the reduction of mechanical pump

speed at lower pressures.

2A4.4 Tolerable Forepressure and Pressure Ratio

Tolerable forepressure (maximum permissible discharge pressure) and the maxi-

mum pressure ratio achievable with a given pump should not be confused. Toler-

able forepressure is directly related to the boiler pressure, while the pressure ratio

has a logarithmic dependence on vapor density in the

jets.

This distinction is in-

dicated in Figure 12. For air the maximum pressure ratio is usually so high that it

cannot be measured under normal circumstances. Only when inlet pressure is less

than

10 ~ ^^

torr can a dependence between discharge and inlet pressure be ob-

served. However, for hydrogen and helium the dependence can be observed in the

high-vacuum range.

The effects of boiler pressure variation on pump performance are summarized,

qualitatively, in Figure 12. Due to the strong dependency indicated in Figure 12,

small changes in vapor jet density may cause large variations in the maximum

pressure ratio. In applications where stable pressure is required with lighter gases,

this may require attention. Ordinary pumps often display inlet pressure variation

Fig.

12.

NOZZLE

TEMPERATURE

lOOX

HEAT INPUT

Relation of forepressure to boiler pressure and the logarithmic dependence of pressure ratio on

vapor density

jets.

2.4.4 Tolerable Forepressure

131

exceeding 5% for helium. This is unacceptable for some applications, such as

highly sensitive mass spectrometer leak detectors and analytical mass spectrome-

ters where helium is used either as a tracer or a carrier gas. Thus, special pump

designs having very stable boiler pressure with power control and stable pumping

action may be required.

In regard to pumping speed, it can be observed that the curves for helium and

hydrogen can have a rather steep slope near the nominal heat input value (100%).

Therefore, if variations of vapor density occur, they may result in noticeable pump-

ing speed variations for these gases.

2.4.4.5 Tolerable Forepressure for Various Gases

The tolerable forepressure for various gases is approximately the same as far as

the complete collapse of the vapor jet (last stage) is concerned. However, in re-

gard to the appearance of lighter gases at the inlet of the pump, when they are in-

troduced in the foreline the maximum pressure ratio effect becomes noticeable.

The results of pressure ratio measurements are shown in Figure 13. The data

were obtained by introducing helium into the foreline of the pump and observing

Fig. 13.

0.2 0.4 0.6

DISCHARGE PRESSURE , TORR

Results of pressure ratio measurements.

132 Chapter 2.4: Vapor Jet Pumps [Diffusion Pumps]

the resulting effect on the inlet pressure. The upper curve represents an older 5-cm

pump. The comparison between the two pumps indicates a thousandfold improve-

ment in maximum pressure ratio, which can be sustained across the pump.

For the older pump in Figure 13, it is difficult to establish a concept of fore-

pressure tolerance for helium, at inlet pressures below

"^ torr. The newer pump

shows a behavior that can be interpreted as if a forepressure tolerance of 0.6 torr

for helium is present. This pump had a forepressure tolerance for air of about the

same magnitude.

2.4.5

ULTIMATE PRESSURE

2.4.5.1 Pressure Ratio Concept

Two distinct observations can be made regarding the ultimate pressure of

pump.

Ultimate pressure may be considered to be a gas load limit or a pressure ratio

limit. Both are of significance in practice, the latter usually only with light gases.

The pumping action of vapor jets does not cease at any pressure, however low.

The ultimate pressure of the pump depends on the ratio of pumped versus back-

diffused molecules, plus the ratio of gas load to the pumping speed. In addition,

the pump itself can contribute a gas load through backstreaming of pump fluid

vapor and its cracked fractions and the outgassing from its parts. Thus, in prac-

tice,

the observed total ultimate pressure is a composite consisting of several ele-

ments. In practice, the conmionly observed first limit is due to the pumping fluid,

although with the best fluid some system degassing (baking) may be necessary to

observe this limit when it is below 1 X

10 ~^

torr.

When liquid nitrogen traps are employed, the limiting pressure is usually given

by various gas loads in the system, even if the system consists only of the mea-

suring dome and a pressure gauge. Below 5 X 10"^ torr thorough baking is usu-

ally required.

For helium and hydrogen, a true pressure ratio limitation can be observed in

most pumps, although special designs can be made to improve the limit beyond

observable level.

2.4.5.2 Baffle and Trap Effects

Room temperature baffles at the inlet to the pump do not affect the ultimate pres-

sure except by shielding a gauge from a direct entry of pumping fluid vapor into

the gauge. Water-cooled baffles suppress the rate of reevaporation of condensed

2.4.5 Ultimate Pressure 133

or intercepted fluid, thereby reducing the density of vapor in the space between

the baffle and the trap. For substances such as vapor jet pump fluids, each 20°C

temperature change near room temperature will result in about an order of magni-

tude change in vapor pressure and hence the rate of evaporation. The lowered

vapor density will reduce the possibility of intercollisions and consequent bypass

through the trap without touching a refrigerated surface.

Cryogenic or refrigerated traps have basically two effects. They act as barriers

for the flow of condensable vapors from pump to system, but they also act as cryo-

pumps for condensable vapors emanating from the system. The latter may be the

primary effect on the ultimate pressure in many cases. In the high-vacuum region

and for unbaked systems using modern low-vapor-pressure pumping fluids, the

reduction of pressure (when traps are cooled) is primarily due to water vapor

pumping. In unbaked systems after the initial evacuation, water may constitute

90%

of the remaining gases and cooling of the trap simply increases the pumping

speed for water vapor (usually by a factor of 2 or 3).

2.4.5.3 Pressure Ratio for Lighter Gases

As noted previously, the pressure ratio can be sufficiently small for the light gases

to reveal the dependence of inlet pressure on the discharge pressure. Measure-

ments of pressure ratio for various gases have been reported as follows: hydrogen

3 X 10^-2 X 10^ helium 10-^-2 X 10^ neon 1 or 2 X 10^ CO and argon 10^,

oxygen and krypton 3-5 X 10^, and hydrocarbons

(C„H2n+2)

7 X 10^. In

modern pumps the helium pressure ratio is closer to 10^, and it can be increased

even as high as

10 ^^

by doubling the heat input. In practice, even an ion gauge op-

erated in the foreline can produce sufficient hydrogen to cause an increase of the

inlet pressure (the same occurrence has been observed in a turbomolecular pump

system).

As far as ultimate pressure is concerned, hydrogen can be a substantial part of

the residual gas composition due to its presence in metals, in the pumping fluid,

and in water vapor. This can be an important consideration for ultrahigh vacuum

work where some diffusion pumps may need a second pump in series. The same

consideration may apply to helium in leak detectors, mass spectrometers, molec-

ular beam experiments, etc.

2.4.5.4 Pumping Fluid Selection

A variety of

"organic"

liquids have been used as motive fluids in vapor jet pumps.

Criteria for the selection of the fluid are low vapor pressure at room temperature,

good thermal stability, chemical inertness, nontoxicity, high surface tension to

Si U

«3 O

'fc

3 53 w >»

C/3 ^3 cd 73

C/3

•^ §

2 i

4^ w

o <o in

V ^

^ ^ '^'

O O rn

O O ON

r\| »o —

«Ot^--Ni^o Oo iTso

«o m

ON —< vO rf

-H CN (N (N

O o

^ in

o ;3 u 2 u 2

- '^ °^ X °^ X

^^ m '^

—<

- V

-^ Tf rj- NO

ON 00 r^ Tt

C<^

Tt lO »0

? ?

8 8

.—4 >-•

•S E a <u

•as §.i

^ § .y S

J2 Si **^ -

U w H C

•/^ 1^ ,? ,5^

u H di PM

CO U

o "^

o fr S t^

O Q < Q

2.4.5 Ultimate Pressure 135

minimize creep, high flash and fire points, reasonable viscosity at ambient tem-

perature, low heat of vaporization, and, of

course,

low cost. A list of the presently

popular fluids and their properties is given in Table 1. The selection of the fluid

should be made giving due consideration to its operational stability in the pump

boiler. A breakthrough in the selection of fluids was made with the use of five-

ring polyphenyl ethers consisting of chains of carbon atoms interbonded by oxy-

gen

[3].

This fluid offered exceptional thermal and chemical stability and enabled

reaching ultimate pressures of 10"^ torr (approaching its vapor pressure at ambi-

ent temperatures) with only water-cooled baffles.

Operational characteristics of another low-vapor-pressure silicone fluid (DC-

705) were discussed by Crawley et al. [4]. Ultimate pressures of 10"^ torr with

water-cooled baffles and

10 ~^^

torr with a baffle at -20°C were reported [5].

Whatever fluid is used, its vapor may pervade the pumped system, dependent

on its vapor pressure, the pump design, and the type of trapping used. The vapor

can be broken down by the presence of hot filaments and bombardment by charged

particles. The polymerization of silicone fluids resulting from bombardment by

charged particles may cause an insulating film to be produced on electrode sur-

faces,

changing the characteristics of the electronic instrumentation. Octoil or

polyphenyl ethers are usually reconmiended to eliminate this problem in applica-

tions where mass spectrometers and other electron optical devices are used.

2.4.5.5 Residual Gas Analysis

With condensable gases of high molecular weight, it is extremely difficult to cor-

relate the ion currents indicated by the mass spectrometer with the rate of back-

streaming through the baffle. Generally, the residual gas analysis of a trapped

dif-

fusion pump system cannot be reliably performed with spectrometers having

poorer detectability than

10 ~^

torr. When the spectrometer tube and other parts of

the system are baked, the results can be very misleading. It may take weeks be-

fore equilibrium conditions are established.

Qualitative measurement can be made, however, with no more difficulty than

the ultra-high pressure measurements made with total pressure gauges such as an

ionization gauge.

A typical mass spectrum from a baffled vapor jet pump system, unbaked, oper-

ating in the ultra-high-vacuum region, with DC-705 motive fluid, is shown in Fig-

ure 14. The mass numbers 16, 19, and 35 are unusually high owing to the charac-

teristic properties of the particular spectrometer tube and the hydrogen peak is

not shown. The peaks 50, 51, 52, 77, and 78 are characteristic for the pumping

fluid. Crawley [4] also gives the residual gas analysis of a vapor jet pump system

using silicone DC-705 fluid under both unbaked and baked conditions. Wood and

136

Chapter 2.4: Vapor Jet Pumps [Diffusion Pumps]

Fig.

14.

I-

«o

»-

T r

III

I \ 1 T I •

••

T T I I 1 ' d

CHAMBER PRESSURE

X 10 '

TORR

111

4JI

III il,

30 40

50 60

MASS NUMBER, "^/e

Typical mass spectrum from a baffled pump. Chamber pressure is 2 X 10"^ torr.

Roenigk [6] and Cleaver and Fiveash [7] provide the mass spectra from which the

significant identifying peaks can be chosen for the detection of the pumping fluid.

2.4.5.6 Fluid Breakdown and Purification

The ultimate pressure in a vacuum system consists of the partial pressures of vari-

ous gases. Components arising from gas evolution from sources other than

the

vapor jet pump (including the inlet gasket) are not discussed here. From the com-

ponents arising

the pump, the major constituents are the vapor

the motive

fluid, the products of its decomposition in the boiler, the contaminant vapors from

other system parts, and back-diffused gases. In

system without the use of cryo-

genic trapping, the ultimate pressure due to the vapor pressure of the fluid at the

ambient temperature is the minimum achievable. In pumps that have a horizontal

ejector stage in the foreline,

addition to increasing the forepressure tolerance,

the ejector stage maintains low pressures

the region under the lower annular

stage. With this arrangement, the volatile components of the returning condensate

are more easily removed to the forevacuum region. Pumps with a strong side ejec-

tor stage are very little influenced by the addition of a booster vapor jet pump for

the improvement of ultimate pressures even in the

10 ~ ^^

torr range [8].