Accelerator Vacuum 101

Made Easy ???

 

 

 

 

 

 

Terry Anderson, P.E.

 

Resident Expert on Nothing

 

Fermi National Accelerator Laboratory

Accelerator Division

 Mechanical Support Department

 

 

 

 

MSDN - ME - 000069

 

 

 

 

11/10/2006


Introduction

 

This paper presents a condensed, simplified, and practical discussion of the principles, procedures, and operating parameters of particle accelerator vacuum systems as practiced at Fermilab.  It is intended to provide a basis for designers, builders, and operators of accelerator systems to communicate with each other about the needs and impact of the vacuum system.  Rigorous analytical development of the equations and concepts are not given.  It is assumed that the reader has some limited understanding of the subject.  Practical examples of real world experiences are used to illustrate the concepts outlined.  Examples of how to use this material is given in appendix 1 and references for further study are given in appendix 2.  The following advice is given for people who design, build, or operate accelerator vacuum systems:

 

A)  Keep it simple.

 

B)  Keep it clean.

 

C)  Establish guidelines and standard practices; then follow them.

 

D)  Always stop and think about what the outcome will be before you do something to the system.

 

E)  Test and certify everything you can before it goes in the system.

 

F)  Despite the abundance of Ōhot airĶ around physics laboratories, air is not the only gas we need to think about.

 

G)  There is no vacuum gauge on this planet that, in and of itself, gives you the real picture.

 

Vacuum can be a complicated subject, but on a base level it does not need to be.  Some may view this discussion as over simplified, but they should realize others donÕt have their level of understanding.  Others may find it complicated and they should realize that they need to have a base understanding in order meet operational goals.  All need to realize that they need to communicate with each other on some base level.  Most of the problems that arise in vacuum practice are a result of a lack of knowledge or communication.  Complicated technical issues can be addressed by physicists and experts.  Practical issues are usually addressed by engineers and technicians.  Having a base understanding by all involved is essential to ensure a successful outcome for the projects they work on. 

 

Why Vacuum?

 

Most all vacuum texts start out with a discussion on the ideal gas law (PV = nRT).  For this discussion it would be nice to avoid this, but it is simply too fundamental to ignore.  In particle accelerators the purpose of the vacuum system is to remove gas molecules from the path of the beam.  So, for accelerators it is more appropriate to think of the ideal gas law in terms of the number of moles in a volume.  Pressure is nothing more than a measure of the number of molecules that can interfere or interact with the beam. 

n = P V / R T              (Eq. 1)

 

Where:    n = Number of Moles

P = Pressure (Torr)

V = Volume (L)

R = Universal Gas Constant (62.3632 Torr-L/K-mol) )

T = Temperature (K)

 

The Vacuum World Accelerators Live In

 

Figure 1, below, is a graphic representation of the vacuum bounds associated with accelerator vacuum systems.  The information depicted is intended to be a guide or a starting point when thinking about accelerator vacuum systems.  Given a pressure range that a system needs to operate in, the chart gives a reasonable estimate of the out-gassing rate needed, the time scale that will be needed to achieve a given pressure, and the likely pump types that will be needed.

 

The chart shows 1(10)-3 Torr as the transition point between the viscous and molecular flow regimes.  This is not strictly the case, though.  Molecular flow can exist above 1(10)-3 Torr and there is the transition flow regime between viscous and molecular.  For all practical purpose in accelerator vacuum systems 1(10)-3 Torr is a good place to start thinking about molecular flow. 

 

For the systems encountered in accelerators the range between atmosphere and 1(10)-3 Torr is not of a lot of concern.  The roughing pumps and turbo molecular pumps (turbo) used usually pump the system down to 1(10)-3 Torr within hours if not minutes.  The main exception to this is insulating vacuum on cryogenic systems.  These systems can take many hours to days to reach 1(10)-3 Torr.

 

Figure 1. The Vacuum World Accelerators Live In

Note: Information depicted is intended for general reference and may or may not be representative of any real system.

 

The range between 1(10)-3 and 1(10)-6 Torr is usually achieved using turbos and can take hours or days depending on the system cleanliness and the amount of water vapor attached to the system walls.  If the system Ōhangs-upĶ in this range it is often necessary to bake the system to release the water vapor.  Ion pumps can operate at pressures above 1(10)-6 Torr, but it is usually not advisable.  The usable life-time of ion pumps operating above 1(10)-6 Torr is reduced significantly, which, may cause operational problems.  At Fermilab, the ion pump power supplies are generally set to trip off at 2(10)-6 Torr or lower and are often interlocked to vacuum sector valves.  If the ion pump trips, the valves close and interrupt the particle beam. 

 

In ranges below 1(10)-7 Torr system pumps are almost exclusively capture pumps (ion pumps (IP), titanium sublimation pumps (TSP), cryogenic pumps (cryo-pump), or non-evaporable getter pumps (NEG)).  These pumps capture the pumped gasses permanently or temporarily and the system is a closed system.  When TSPÕs, cryo-pumps, or NEGÕs are used some number of IPÕs are also used to pump the gasses not pumped by the TSPÕs, cryo-pumps, or NEGÕs.  In general TSPÕs and NEGÕs donÕt pump noble gasses, methane, or ethane.  Cryo-pumps will pump all gasses at some level depending on the temperature of the cryogenic surface.

 

At pressures below 5(10)-9 Torr it is almost always necessary to do an in-situ low temperature bake to remove water vapor from the system.  A useful low temperature water bake is time and temperature dependent.  The temperatures need to be at least 100C, and are often as high as 300C.  Bake durations run from 24 hours to a couple of weeks.  If the system is unusually water loaded, an in-situ low temperature bake may also be necessary to achieve pressures higher than 5(10)-9 Torr. 

 

There are two topics that are not explicitly shown on the chart, but that should be covered in this discussion.  They are connections and materials.  The materials used for vacuum systems are generally ceramics or metals, specifically stainless steel, aluminum, and copper.  There are synthetic materials that can be used, but most should be avoided at pressures below 1(10)-7 Torr.  At pressures lower than 5(10)-9 Torr it is usually necessary to degas metals to remove the trapped H2 in them.  This is another time/temperature dependent process and is done under vacuum.  For stainless steel this can be partially accomplished by baking at 500C for 24 hours or more.  The standard practice is to bake at 950C for 2 hours.  Care must be taken when heating and cooling between 600 and 850C.  In general, the heating and cooling in this range needs to be as fast as possible to prevent carbide precipitation to the grain boundary. 

 

It is almost always best to use welded or brazed connections in vacuum systems operating below 1(10)-8 Torr.  The exception is on components and devices that need to be removed periodically for maintenance or repair, in that case flanges should be used.  The seals used for the flanges should be metal (copper or aluminum).  Elastomer seals can be used at pressures above 5(10)-9 Torr, but they should be used sparingly.  Elastomers will permeate water and helium and have a significant out-gassing rate.  Helium permeation will interfere with leak checking by lowering the sensitivity and slowing clean-up time.

 

Basic Equations for Vacuum

 

Equations 2 through 5 constitute the basic equations for most all vacuum work.  Knowing these relationships and how to apply them is essential for understanding what is happening with regard to any vacuum systems.  In most cases accelerator vacuum work can be simplified to this level.  References [1] & [2] in appendix 2 are excellent sources for learning how to apply and understand these relationships.  In these sources, additional formulas for calculating the conductance based on an assortment of geometries and gasses are presented.  Figure 2 is presented here to illustrate the effect of geometry on the conductance for round tube.

 

For Molecular Flow the Following Apply:

 

Q = S P                    (Eq. 2) (Relationship between Flow, Pump Speed, & Pressure at the Pump) 

 

Q = C DP                 (Eq. 3) (Relationship between Flow, Conductance, & Pressure Drop)

 

C = k A a                (Eq. 4) (Relationship between Conductance, Gas Species, & Geometry)

 

1/Seff = 1/S + 1/C     (Eq. 5) (Relationship between Effective Pump Speed, Rated Pump Speed, & Conductance)

 

Where:       Q = Gas Flow (Torr-L/s)

S = Pump Speed (L/s)

P = Pressure (Torr)

C = Conductance (L/s)

k = Flow Constant for Specific Gas (L/s-cm2)

A = Cross-Sectional Aperture Area (cm2)

a = Transmission Probability

Seff = Effective Pumping Speed (L/s)

 

Figure 2. Conductance for Round Tube

Multiply by (29/Mm)0.5 to adjust for other gases. 

Where: Mm = molar mass (g/mol) of gas.

System Design

 

At this point in the discussion it is useful to outline the basic design parameters used in designing accelerator vacuum systems.  Figure 3 shows the basic layout for most accelerators.  In general there is some vacuum space (generally a stainless steel tube) with pumps connected to it at some defined spacing (Lp).  The spacing is often dictated by the length of the magnets used for steering the beam.  The aperture of the tube can be round, elliptical, or rectangular and is mostly dictated by the beam size and the magnet pole tip spacing.  The vacuum levels required are dictated by the beam and operational reliability.  Equations 6 through 8 [1] are the governing relationships for the system design.

 

 

Figure 3. Basic Accelerator Vacuum System Layout

 

Pm = Pp + DP              (Eq. 6) (Pressure between pumps)

 

Pp = qD B Lp / S           (Eq. 7) (Pressure at the pump)

 

DP = qD B Lp / (4C)     (Eq. 8) (Pressure drop in a beam tube from midpoint to pump)

 

Where:    Pm = Midpoint Pressure (Torr)

Pp = Pump Pressure (Torr)

qD = Specific Outgassing Rate (Torr-L/s-cm2)

B = Inside Tube Perimeter (cm)

Lp = Pump Spacing (cm)

C = Conductance Over Length Lp/2 (L/s)

 

Although the beam size and magnet geometry dominate the geometry for the vacuum system, vacuum considerations must be taken into account early in the magnet and tube size selection.  This is necessary to make sure the vacuum system will perform as needed.  To illustrate this, example 1 is given in appendix 1.

 

 

Component and Device Design

 

This brings us to the point in the discussion where we need to address components and devices that get placed where magnets arenÕt.  For the purpose of this discussion, components are any part of the vacuum system that is not the beam tube.  Components would include bellows, fittings, flanges, valves, and any other parts whose primary function is as part of the vacuum system.  Devices are components whose primary function is beam related.  Examples would be RF cavities, separators, collimators, Lambertsons, and instrumentation such as BPMÕs, IPMÕs, flying wires, etc.

 

Given that there is a basic system design, a system will have some gas load per unit length. 

If Q = S P, then S P / Lp equals the gas load/cm for the system.  Any component or device to be installed has to have a gas load, per unit length, equal to or less than the above load.  If the gas load is larger, additional pumping must be supplied.  In general the pressure at any additional device or component has to have a pressure equal to or less than the average in the system.  Devices and components installed in accelerators tend to have very large internal surface areas relative to their length.  In addition, the materials used are not always the best choice for use in vacuum systems.  Therefore, very high gas loads can be expected.  Examples 2 and 3 are given in appendix 1 to illustrate these basic concepts.

 

Someone knowledgeable in vacuum practice should always be involved in the design, manufacture, and assembly phases of these components and devices.  The best way to assure that these are built so that they will do no harm to the vacuum system that they will reside in is to test the components before they go into the assembly and certify the assemblies before they go in the system.  If components and devices are installed without certification there is a very real risk that the entire system can become contaminated. 

 

Figure 4 is a sketch of a test chamber for testing parts that will be used in devices.  The set-up would be the same for certifying an assembly, with the assembly replacing the chamber.  With a set-up like this, one can test for out-gassing rates, total gas load from a part, identify contaminants, and determine the gas composition in the assemblies.  Tables 1 & 2 are examples of the results obtained from tests done with this set-up.

 


Figure 4.  Vacuum Test Chamber Set-up


Table 1.  Out-gassing Rates (Torr-L/s-cm2) of Various Materials

Material

Totals                 (Torr-L/s-cm2)

H2

CH4 Methane

H2O

CO / N2

C2H6 Ethane

Ethyl Alcohol

O2

Ar

C3H6 Cyclo - propane

CO2

Stainless Steel (unbaked, no degas)

1.0E-10

5.0E-11

2.0E-13

5.0E-11

1.0E-12

5.1E-14

 

3.0E-16

 

1.0E-14

1.0E-13

Stainless Steel (baked, no degas, based on Small Test Chamber)

5.1E-11

5.0E-11

2.0E-13

8.0E-14

2.0E-13

5.1E-14

 

3.0E-16

 

1.0E-14

8.0E-14

Stainless Steel (baked & degassed, based on Recycler)

6.7E-13

6.2E-13

1.0E-14

1.3E-14

1.3E-14

5.4E-15

 

1.7E-16

4.8E-16

 

5.7E-15

Torlon (baked)

3.1E-08

4.2E-09

6.2E-11

2.5E-08

6.7E-10

9.6E-11

 

8.2E-11

2.6E-12

7.5E-12

9.1E-10

Armalon (baked) (Glass filled Teflon)

3.1E-10

3.0E-10

2.0E-12

6.0E-12

4.0E-12

3.0E-13

3.0E-14

2.0E-15

1.0E-14

4.0E-13

2.0E-12

Rulon (baked) (Glass filled Teflon)

6.7E-11

6.1E-11

1.7E-13

4.2E-12

1.4E-12

 

 

1.4E-15

 

 

1.3E-13

Microwave Absorber Material (MF190) (baked)

6.5E-11

4.0E-11

2.5E-13

1.1E-11

6.0E-12

1.1E-12

 

3.8E-14

3.3E-15

5.7E-14

6.2E-12

 

Table 2.  Out-gassing Rates (Torr-L/s) of Various TeV IPM Flex Circuit Components

Sample Material

Total

H2

CH4 Methane

H2O

CO / N2

C2H6 Ethane

Ethyl Alcohol

O2

Ar

C3H6 Cyclo - propane

CO2

Test Chamber Baseline

2.3E-08

2.2E-08

1.2E-10

2.2E-11

6.8E-10

1.2E-11

 

 

5.3E-13

3.7E-13

3.6E-11

Flex Circuit Sample 1

1.0E-07

8.7E-08

6.9E-11

1.9E-09

1.3E-08

5.0E-10

1.2E-10

4.9E-12

3.1E-11

1.3E-10

6.2E-10

Flex Wires

2.3E-08

2.2E-08

1.7E-10

1.5E-11

9.9E-10

2.5E-12

 

 

 

 

3.9E-11

Flex Circuit Sample 2

1.7E-08

1.6E-08

7.1E-11

4.0E-11

9.7E-10

2.1E-12

1.2E-11

 

 

 

3.4E-11

Flex Circuit Sample 3

3.9E-08

3.1E-08

4.6E-10

7.0E-10

5.6E-09

5.4E-11

 

3.9E-11

 

 

5.6E-10

Peek Connector, before bake

1.3E-04

3.6E-05

1.1E-07

8.5E-05

2.0E-06

2.2E-07

 

6.4E-06

6.4E-08

1.1E-09

3.3E-06

Peek Connector, after bake

7.2E-07

4.0E-07

1.4E-09

2.7E-07

2.4E-08

 

 

8.1E-09

 

 

1.7E-08


A Cautionary Tale

 

Earlier the potential for contamination was mentioned, Figure 5 is a residual gas analyzer (RGA) scan of a device (Flying Wire) that was installed in FermilabÕs Recycler.  The black peaks are from a scan taken before installation.  This device was not properly certified prior to installation.  There was no low temperature bake done to get an accurate picture of the assembly and the RGA was only on for a couple of hours.  RGAÕs need to be on for many hours to days before reasonable scans can be taken.  The scan seen here (black peaks) is typical of an RGA being on for only a few hours.  The clusters of peaks that repeat about every 12 to 14 mass units are hydrocarbons.  If there is no contamination they are the result of the RGA filament heating up and degassing.  On a clean system, with a clean RGA, this can be seen to clean up over several hours.  On this particular device these clusters were observed to be decreasing over a period of a couple hours. 

 

Figure 5.  RGA Scan of Recycler Flying Wire

 

There was a push to install this device during a shutdown and the decision was made to go ahead with the installation without a proper certification.  The device was subsequently installed and baked in-situ at 150C.  The bake appeared to go well and the system came down to 1(10)-9 Torr after the bake.  After the shutdown was over and the machines started running it was noticed that the vacuum was degrading in this sector.  Over the course of a week the pressure degraded to 1(10)-8 Torr and stabilized.  Another access was made and the TSPÕs were reactivated and the pressure again was 1(10)-9 Torr.  Within a week the pressure was again 1(10)-8 Torr and stabilized.  This went on for several months with periodic accesses to leak check and other attempts to find the problem.  Eventually an RGA was installed on the system and the scan showed something very similar to the red peaks in Figure 5. 

 

It turned out that a bearing in the rotary feedthrough on the device had a grease lubricant with a high vapor pressure.  The original bearings specified for the feedthrough were dry, but the supplier had sent some that had grease.  To make matters worse, the initial 150C bake had spread the contamination through the entire sector and the change in tunnel temperature while running (about 15C) would accelerate the out-gassing by an order of magnitude.  The TSPÕs would spoil in about a week.  Figure 6 shows the effect of the small temperature rise.  Ultimately the entire sector had to be replaced.

 

Figure 6.  Change in Out-Gassing Due to Small Temperature Rise

 

 

This tale is told to illustrate several points that have been made in this discussion:

 

1)      Establish guidelines and standard practices, then follow them.

 

2)      Always stop and think about what the outcome will be before you do something to the system.

 

3)      DonÕt guess – test and certify everything you can before it goes in the system.

 

4)      Someone knowledgeable in vacuum practice should always be involved in the design, manufacture, and assembly phases of  components and devices that will reside in the vacuum system. 

 

Keep It Clean

 

If I have learned anything in my 18 years at Fermilab, it is that everything that goes in the vacuum system needs to be clean.  In my opinion, this is the single most important factor affecting the quality of a vacuum system.  The equipment will run better, last longer, and be more reliable if the system is clean.  The gas composition will be more acceptable to the beam and the desired pressure levels will be achievable.  I have this sign on my office wall (Figure 7) and it is the single most important guiding principle when I look at vacuum systems. 

 


Figure 7.  TerryÕs Five Rules for Good Vacuum

 

The obvious next question is; what constitutes a clean vacuum system?  And, the answer is; that depends.  The better question is; what is a contaminant?  And, that answer is easy.  Any gas in the system that does not need to be there, that interferes with the beam, or prevents the system from reaching the required ultimate pressure is a contaminant.  Figure 8 is an RGA scan of a very clean system.  H2 is over 95% of the gas composition with the next highest component being CO at about 1%.  N2 is slightly less that than CO and CH4, H2O, C2H6, and CO2 are all some fraction of 1%.  These are all contaminants, but it may not be possible to do any better.  If the pumping is sufficient the system should be able to reach a desired ultimate pressure and at that pressure the beam/gas interaction should be acceptable.

 

Figure 9 is an RGA scan of a dirty system, and is the one discussed earlier with the grease in the bearing.  The hydrocarbons in that system did not need to be there, they interfered with the beam (caused large losses and short life times), and they prevented the system from reaching the required ultimate pressure.

 

It is always asked what the best cleaning procedure is?  The best cleaning procedure is the one that works best for a given application.  There are any number of procedures that work, so be flexible when specifying cleaning procedures.  An understanding of the processing that a part has been through is a prerequisite for determining the proper cleaning procedure.  Specify the vacuum performance you want, and then choose the cleaning procedure that works.  Once a part or device is clean, great care should be taken to keep it clean.  Clean assemblies should always be vented with dry N2 and clean parts should always be stored under vacuum or in hermetically sealed UHV grade containers, back-filled with dry N2.  Clean parts should always be handled with clean latex (or equivalent) gloves.  Transfer of solvents, grease, and dirt by handling is the number one way a system is contaminated.


Figure 8. What a Very Clean System Looks Like

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Figure 9. What a Dirty System Looks Like

 

The most common procedure used is ultra-sonic cleaning with an acid or alkaline based detergent.  This is followed by several thorough clean water rinses (usually distilled or de-ionized water).  If it is a metal part and needs to be cleaner than the above procedure can achieve, follow it with a vacuum bake at 400 to 500C, and that will be about as good as it gets.

 


Vacuum Quality and Operation

 

To this point the discussion has centered on how a system is designed and a little bit about how that design is achieved.  But ultimately, what really matters is how the system performs and how easy it is to maintain and operate.  The only performance that really matters is the interaction (or lack of) with beam.  This goes back to how much gas and of what species is in the vacuum space.  Two different gasses (say H2 and H2O) at the same pressure can have dramatically different effects on the beam.  It basically comes down to the size of the molecules; larger molecules present a bigger target for the beam to hit.  Physicists use a number, Zeff (Eq. 9), as a measure of the molecular size of the gas mixture that the beam encounters.  This is really nothing more than the effective atomic number of the gas mixture.  Higher values for Zeff  present a greater potential for beam/gas interactions than smaller values, thereby increasing losses and decreasing beam lifetimes. 

 

Zeff = S(xnniZi/nn)       (Eq. 9)             Where: ni = The number of atoms of the ith species in the molecule.

                                                            Zi = The atomic number of the ith atom in the molecule.

                                                            xn = The fractional concentration of the nth gas.

                                                            nn = The number of atoms of the nth molecule.

 

So, when one talks about accelerator vacuum quality, it is really the combination of the system pressure (a measure of the number of molecules in the space) and the species make-up of the gasses remaining in the system. 

 

Operationally this means that there is some standard (quantitative and qualitative) that has to be maintained for any given system.  Most of this is addressed in the design and building phase of a system through proper selection of materials, processing methods, and careful construction.  But, at some point the system has to operate and invariably will need modifications.  Additions and modifications to a system must follow the same rules that were used to design and build it in the first place. 

 

Maintaining the system operationally requires an understanding of what was done before and what will happen in the future if something is done now.  More precisely this means that the people responsible for maintaining the system need to know what the system design parameters are, what the vacuum quality (both quantitative and qualitative) needs to be, and what has been done over the course of time that could have changed the vacuum quality.  At a basic level this means good record keeping. 

 

They also need to know what the affect will be if something is done that changes the systemÕs equilibrium temporarily or permanently.  An example of this is the simple act of letting-up a system to atmospheric pressures.  Figure 10 is a plot of a system that was first baked at 200C, and then vented by two different methods.  The first method was venting with clean dry N2 and then pumping back down without a bake.  The second method was to vent to outside air and remain exposed for two hours, then pumping back down without a bake.


Figure 10.  What Happens When a System is Vented?

 

It can be seen that there is a difference in both total pressure and the time taken to reach any given pressure.  Zero time in all cases is the start of initial pump down.  The bake was approximately 4 days, but could have been as short as two days with about the same results.  As far as total pressure goes, there is about a factor of two between the subsequent processes.  From a strictly pressure point of view this is not very significant.  But, if the gas mixture composition is looked at (Figure 11) the difference is significant. 

 

There is a significant increase in the carbon containing molecules (CH4, CO, C2H6, and CO2), H2O, and N2.  The increase in the carbon compounds is probably driven by the hot filaments on the ion gauge and RGA.  This will probably clean up in time and has a large affect because the vessel is small relative to the number of filaments.  What is significant though, is the increase in H2O and N2.  For the system vented with air there is an increase in the level of H2O of almost two orders of magnitude.  This is due to mono-layers of H2O accumulating on the surfaces from the moisture in the air.  Without a bake this will take a very long time (months to years) to pump away.  There is also an increase in the H2O level as a result of venting with clean dry N2.  This is because the vent line and exterior of the system valve had all been exposed to air for a very long time.  Additionally, there is always a slug of air in the vent line; unless it is permanently connected to the system, purged of air, and baked to remove the H2O.  The N2 level has increased in both cases and will continue to remain high for some time, but will eventually pump away.  Below the chart are values for Zeff in the three cases. 


 

Figure 11.  Vacuum Quality After Venting

Zeff for Baked System = 1.1, Zeff for N2 Vented System = 1.57, Zeff for Air Vented System = 2.75

 

 

Gauges

 

To this point, the discussion has really said nothing about gauges except a statement in the introduction that; there is no vacuum gauge on this planet that, in and of itself, gives you the real picture.  We could probably leave it at that and go on with life in blissful ignorance.  But, if you really want to understand what a vacuum system is doing an understanding of this is required.  In practice, gauges are the only indicator of a systemÕs health.  And, that is all they are; indicators.  Almost all gauges used in the vacuum regimes that are of interest to accelerators are calibrated for N2 as the only gas in the system.  As was shown earlier this is almost never the case.  In addition, most of the gauges used (primarily ion gauge and cold cathode) are total pressure gauges.  So they treat all the gas species as one gas.  The relationship between the gauge pressure and actual pressure for a specific gas for a Bayard-Alpert type ion gauge is given in equation 10. 

 

PgasX =             (Pind / rgasX)(KN2, cont / KN2, gauge)          (Eq. 10)

 

Where:             PgasX = The pressure of gas X.

                        Pind = The N2 equivalent pressure indicated by the gauge.

                        rgasX = Gas sensitivity for gas X.

                        KN2, cont = Controller sensitivity for N2.

                        KN2, gauge = Gauge sensitivity for N2.

 

Values for KN2, gauge, KN2, cont, and rgasX can be found in the literature and manufacturersÕ catalogs.   KN2, gauge is typically between 10 and 25 /Torr.  If the controller and gauge are calibrated the same the K terms drop out.  Typical sensitivity values for some gasses are listed in Table 3.

 

Table 3.  Gas Sensitivity Table

Gas

H2

CH4

H2O

CO

N2

C2H6

O2

Ar

C3H6

CO2

Sensitivity, r

0.46

1.4

1.12

1.05

1.0

2.6

1.01

1.29

3.6

1.4

 

To get a better picture, a gauge that differentiates between the various gasses is needed.  At first glance an RGA would seem to do this, but there are also calibration issues there.  An RGA differentiates between the masses of the various gasses, but thinks everything is still N2 when it comes to the magnitude of the pressure.  In addition, there is always a question of how well any gauge (ion gauge, cold cathode, or RGA) is calibrated for N2.  As a general rule, I treat all high or ultra-high vacuum gauges as only being accurate to with in a factor of two.  Most RGAÕs are moved from system to system as needed.  This is mostly because they cost so much.  Even when they are permanently installed on a device or system, it is usually only the analyzer that is installed and the head and electronics are moved from analyzer to analyzer.  In this case calibration becomes a real slippery slope, and can be off by orders of magnitude. 

 

To combat all these inaccuracies and uncertainties I try to only use an RGA in conjunction with an ion gauge.  The RGA is used only to get a qualitative measure of the gasses in the system and a sense of the relative concentrations of the gasses.  An ion gauge is then used as the total pressure measurement and the RGA data is normalized to that.  As an example assume that the RGA scan in Figure 8 indicates the relative partial pressures of the gasses as shown in Table 4.  The ion gauge on the system near the RGA shows a total pressure of 2.7(10)-9 Torr.  The fractional contribution of each gas to the total is also shown in the table.  Setting both KÕs in Equation 9 equal to 1 and making use of the fact that the sum of the partial pressures is equal to the total pressure an equation can be written to correct for the gauge calibration.

 

PTact = PTIG / S((Pi / PTRGA) ri)                        Eq. 10

 

Where:    PTact = Actual total pressure (Torr).

PTIG  = Indicated ion gauge pressure (Torr).

PTRGA = Total indicated RGA pressure (Torr).

Pi = Partial pressure of the ith gas (Torr).

ri = Gas specific sensitivity for the ith gas.

 

It can be seen that the actual total pressure is 5.6(10)-9 Torr, which is about two times the indicated 2.7(10)-9 Torr.  The individual partial pressures are then the fractional contributions of each gas times PTact. 

 


Table 4.  RGA Analysis Normalized to an Ion Gauge

 

Totals

H2

CH4

H2O

 N2

CO

C2H6

CO2

RGA Partial Pressure (Torr)

1.54E-08

1.50E-08

1.00E-10

3.90E-11

3.50E-11

1.70E-10

1.10E-11

2.10E-11

Fractional Contribution to Total

1.00

0.976

0.007

0.003

0.002

0.011

0.001

0.001

Gas Specific Sensitivity (r)

 

0.46

1.40

1.10

1.05

1.05

2.60

1.40

Adjusted Partial Pressure (Torr)

5.64E-09

5.51E-09

3.67E-11

1.43E-11

1.28E-11

6.24E-11

4.04E-12

7.71E-12

 

 

 

 

Conclusion

 

The topics discussed here are by no means complete.  Anyone needing an in-depth knowledge of the subject is encouraged to pursue further study with the references suggested.  This discussion does, however, provide a core of knowledge for those working with accelerator vacuum systems.  The intention was to condense and simplify the basic concepts, design, and operating parameters that are encountered in most accelerator vacuum systems.  This discussion should be of particular use to those that design and build components and devices that reside in the vacuum system, but who typically have no involvement in the design, building, or operation of the vacuum system itself. For those that are actively involved in the design, building, and operation of a vacuum system, this discussion provides a base point to build on.  The primary purpose of this discussion is to give all those that are in some way contributors to the vacuum system, whether directly or ancillary, a common means to communicate with each other.


Appendix 1

 

Examples

 

Example 1.

 

Assume the following:  An accelerator is to be built using a new room temperature super-conductor, so cryogenic magnets are not needed.  The required aperture is determined to be 2.5 cm in diameter.  The desired magnet length is 10 m.  The required vacuum level is 4(10)-10 Torr and it is preferred that a bake is not needed.

 

First Pass:          qD = 5(10)-11 Torr-L/s-cm2                                                                                              (from Figure 1)

 

S = 20 L/s ion pump                       

Because of the small tube size there will be very little conductance on the port, so a larger pump will be effective.

 

C (pump to tube) = 13 L/s                                                                          (from Figure 2)

Assume a 4Ķ long port 2.5 cm in diameter from the beam tube to the pump.

 

Seff = 1/((1/20)+(1/13)) = 8 L/s                                                        (from Eq. 5)

 

B = 3.14 x 2.5 = 8 cm

 

Pp = 5(10)-11 (8) (1000) / 8 = 5(10)-8 Torr                                       (from Eq. 7)

 

Therefore --        NOT GOING TO WORK!  Clearly the gas load is too high to reach the required pressure even at the pump.  If the beam tube is H2 degassed and there is an in-situ water bake, the out-gassing rate would drop to 5(10)-13 Torr-L/s-cm2.  So, letÕs try that.

 

Second Pass:      qD = 5(10)-13 Torr-L/s-cm2                                                                                              (from Figure 1)

 

Beam Tube C = .4 L/s                                                                      (from Figure 2)

This is the conductance from the mid-point between pumps to a pump.

 

Pp = 5(10)-8 x 5(10)-13 / 5(10)-11 = 5(10)-10 Torr                              (from Eq. 7)

 

DP = 5(10)-13 (8) (1000) / (4) (.4) = 2.5(10)-9 Torr                         (from Eq. 8)

 

Pm = 5(10)-10 + 2.5(10)-9 = 3(10)-9 Torr                                           (from Eq. 6)

 

Therefore --        NOT GOING TO WORK!  The pressure drop is too large given the pump spacing.  Remember the note at the bottom of Figure 2, the values for conductance and pumping speed used thus far have been for air or N2.  The previous calculation showed that the tube needed to be degassed and baked in-situ.  Under these conditions and pressures H2 is almost always 90% or more of the gas load.  The RGA scan of a very clean and baked system illustrated this.  Also, not divulged in the discussion is the fact that ion pump speeds are given for pumping N2, but their pumping speed for H2 can be 3 times that.  So, use H2 values and multiply the conductance by 3.8 and the pump speed by 3.

 

Third Pass:

 

Seff = 1/((1/(20 x 3)) + (1/(13 x 3.8))) = 27 L/s                                (from Eq. 5)

 

Pp = 5(10)-13 (8) (1000) / 27 = 1.5(10)-10 Torr                                (from Eq. 7)

 

DP = 5(10)-13 (8) (1000) / (4) (.4 x 3.8) = 6.6(10)-10 Torr               (from Eq. 8)

 

Pm = 1.5(10)-10  + 6.6(10)-10  = 7.5(10)-10 Torr                               (from Eq. 6)

 

Therefore --        Not Quite, but we can work with this! Change the pump spacing to 5m. 

 

Fourth Pass:

 

Pp = 5(10)-13 (8) (500) / (27) = 7.5(10)-11 Torr                                (from Eq. 7)

 

DP = 5(10)-13 (8) (500) / (4) (.8) (3.8) = 1.6(10)-10 Torr                 (from Eq. 8)

With the shorter pump spacing the conductance from the mid-point between pumps to a pump becomes 0.8 L/s.

 

Pm = 7.5(10)-11 + 1.6(10)-10 = 2.4(10)-10 Torr                                  (from Eq. 6)

 

Therefore --        This system can work.  Without going through this kind of a calculation the people building the magnets would not have known to build in a bake-out system or that there needed to be pump ports every 5 meters.  This example also highlights the assertion in Figure 1 that if pressures below 5(10)-9 Torr are needed a bake should be planned for.  From the second, third, and fourth pass calculations it should be evident that the beam tube is Ōconductance limitedĶ.  This means that regardless of how much pumping is provided the pressure at the mid-point can not be less than that calculated from equation 8.

 

Example 2:

 

A device is stated as having the equivalent internal surface area as a 10 meter length of beam tube.  The materials are degassed stainless steel or are such that the out-gassing rates are equivalent.  The device takes up a slot length of 1m in the beam line.  The vacuum parameters are the same as in Example 1 and will require a low temperature bake.

 

Since we know that the gas load is larger than that of 1m of beam tube, additional pumping will be required. 

 

The total gas load for the device is;

Q = 1000 S P / Lp = (1000) (27) (7.5(10)-11) / (500) = 4.1(10)-9 Torr-L/s

The pumping speed used is the Seff from the previous example and the pressure at the pump is also from the previous example.

 

The effective pumping speed needed is;

Seff = Q/P = 4.1(10)-9 / 4(10)-10 = 10 L/s                                                       

4(10)-10 Torr is used because that is the design pressure.

 

If a 20 L/s ion pump is used the effective pumping speed would be;

Seff = 1/((1/60) +(1/49)) = 27 L/s                                                                                (from Eq. 5)

Remember that the material is degassed and baked so the dominant gas is H2.  The values for conductance and pumping speed are for H2. 

 

Example 3:

 

If we now consider that in addition to the equivalent 10 meters of beam tube there is a small piece (1mm x 1mm x 2.5cm) of Torlon used as an insulator in the device.  The gas load will change significantly.  In addition, gas coming off of the insulator will not be dominated by H2 any more (see Table 2 for Torlon).  In this case the gas is dominated by H2O and the calculations need to reflect that.

 

The Torlon has a surface area of 1.02 cm2,  and an out-gassing rate of 3.1(10)-8 Torr-L/s-cm2.

 

The gas load from this 1 piece is Q = qd As = 3.1(10)-8 x 1.02 = 3.2(10)-8 Torr-L/s

 

The effective pumping speed needed is;

Seff = Q/P = 3.1(10)-8 / 4(10)-10 = 78 L/s for H2O

 

The actual pump needed will probably be 150 L/s (the H2O pumping speed is about the same as N2 for ion pumps) because of conductance losses in the port connecting the pump to the device.  An ion pump this size would require a 6Ķ CF flange and a 4Ķ port about 4Ķ long.

 

Seff = 1/((1/150) +(1/(467 x 1.27))) = 120 L/s                                                            (from Eq. 5)

 

Considering the small size of the beam tube, the device may also be relatively small and may not be able to accommodate a 4Ķ port.  It would probably be a better idea to use an insulator made of another material, like ceramic. 


Appendix 2

 

Recommended Reference Material:

 

1)         Roth, A., Vacuum Technology, 3rd ed, 1990 (Elsevier Science B.V.)

 

2)         Lafferty, J.M., Foundations of Vacuum Science and Technology, 1998 (John Wiley & Sons, Inc.)

 

3)         Turner, S., et al., CERN Accelerator School Vacuum Technology, 1999 (CERN)

 

4)         Chao, A.W. & Tigner, M., Handbook of Accelerator Physics and Engineering, 1999 (World Scientific Publishing Co.)

 

5)         Drinkwine, M.J. & Lichtman, D., Partial Pressure Analyzers and Analysis (AVS)

 

6)         ManufacturerÕs Catalogs (Varian, Leybold, Alcatel, SAES, Granville Philips, and others)