IA-1271

VACUUM TECHNOLOGY

PART I.

A. ROTH

•/ ' I *^*1 v, I , << / l\« 1 lktfsi»?AiV8ffi/ri IA-127I Israel Atomic Energy Commission A. ROTH Vacuum Technology October 1972 582 p. This is the text of a Postgraduate Course given by the author at the Faculty of Engineer­ ing of the Tel-Aviv University, After an introduction dealing with the main applications and history of vacuum technology,

the course discusses relevant aspects of rarefied gas theory, and treats in detail molecular, viscous and intermediate flow through pipes of simple and complex geometry. Further chapters deal with relevant physico- chemical phenomena (evaporation-condensation, sorptlon-desorption, permeation), pumping and measuring techniques, and special techniques used for obtaining and maintaining high vacuum (sealing techniques, leak detection). (Parts I & II). VACUUM TECHNOLOGY

PART I

A. Roth

Israel Atomic Energy Commission October 1972 Head Vacuum Technology Dept. Soreq Nuclear Research Centre I

CONTENTS

Page 1. Introduction 1 1.1 The vacuum 1 1.11 Artificial vacuum 1 - Vacuum ranges 4 - Composition of the gas 4 1.12 Natural vacuum , 6 Vacuum on earth 6 Vacuum in space 6 1.2 Fields of application and importance 7 1.21 Applications of vacuum techniques 7 1.22 Importance of vacuum technology 13 1.3 Main stages in the history of vacuum techniques .... 14 1.4 Li terature sources 18

2. Rarefied gas theory for vacuum technology • 25 Commonly used symbols 25 2.1 Physical states of matter 27 2.2 Perfect and real gas laws 34 2. 21 Boyle' s law 34 - McLeod's gauge 35

2.22 Chales1 law 37 2.23 The general gas law 38 2.24 Molecular density 42 2.25 Equation of state of real gases 44 2.3 Motion of molecules in rarefied gases 46 2.31 Kinetic energy of molecules 46 2.32 Molecular velocities 49 2.33 Molecular incidence rate 51 2.4 Pressure and mean free path 53 2.41 Mean free path 53 2.42 Pressure units 57 II

Page 2.5 Transport phenomena in viscous state 61 2.51 Viscosity of a gas 61 2.52 Diffusion of gases 65 - Diffusion pump (principle) 66 2.6 Transport phenomena in molecular state 68 2.61 The viscous and molecular states 68 2.62 Molecular drag 70 - Time to form a monolayer 71 - Molecular pump (principle) 71 - Molecular gauge (principle) 72 2.7 Thermal diffusion and energy transport 73 2.71 Thermal transpiration 73 2.72' Thermal diffusion 74 2.73 Heat conductivity of rarefied gases 75 -. Heat conductivity in viscous state 75 - Heat conductivity in molecular state 77 - Thermal conductivity gauge (principle) 82 Appendix • • 83

3. Ga» flow at low pressures , 87 Connonly used symbols •. • < 67 3.1 Flow regimes, conductance and throughput B9 3.11 Flow regimes 89 - The Reynold number 90 -. The Knudsen number 91 3.12 Conductance 92 - Parallel and series connection 94 3.13 Throughput and pumping speed 95 3.2 Viscous and turbulent flow 99 3.21 Viscous flow-conductance of an aperture 99 3.22 ViBcoua flow-conductance of a cylindrical plpe-Polseuille'a law 103 3.23 Viscous flow-surface slip 107 3.24 Viscous flow-rectangular cross section 108 3.25 Viscous flow-annular cross section 110 3.26 Turbulent flow Ill Ill

Page 3.3 Molecular flow 112 3.31 Molecular flow-conductance of an aperture .... 112 3.32 Molecular flow-conductance of a diaphragm .... 113 3.33 Molecular flow-long tube of constant cross section , 115 - Circular cross section ,.....,,.,,.,...... 117 - Rectangular cross section 117 - Triangular cross section ,,...... ,.,,.,...... 118 - Annular cross section 118 3.34 Molecular flow-short tube of constant cross section 119 - Circular cross section 120 - Rectangular cross section 121 - Annular cross section 121 3.4 Conductance of combined shapes .*...-...„...... 122 3.41 Molecular flo^-tapered tubes 122 - Circular cross section 124 - Rectangular cross section 125 3*42 Molecular flow-elbows 125 3.43 Molecular flow-traps 126 3.44 Molecular flow-optical baffles 133 - Conductance of baffles with straight plates .. 134 - Conductance of baffles with concentric plates 135 3.45 Molecular flow-seal interface ..* 138 3.5 Analydico-Btatistlcal calculation of conductances... 142 - Transmission probability for elbows ...... 147 - Transmission probabiliry for annular pipes — 148 - Transmission probability for baffles 149 3.6 Intermediate flow — 154 3.61 Knudsen * s equation , 154 3.62 The minimus conductance — ...... 155 3.63 The transition pressure 157 3.64 Limits of the intermediate range 158 3.65 General equation of flow 159 3.66 The viscous-molecular intersection point ..... 160 3.67 Integrated equation of flow 164 IV

Page 3.7 Calculation of vacuum systems 16B 3.71 Sources of gas in vacuum systems 168 3.72 Pumpdown In the viscous range 170 3.73 Pumpdown in the molecular range 174 3.74 Steady state with distributed gas load 178 3.75 Nomographic calculations 181 4» Physico-chemical phenomena in vacuum techniques 187 4.1 Evaporation-condensation 187 4.11 Vapours in vacuum systemB 187 4.12 Vapour pressure and rate of evaporation 188 4.13 . Vapour pressure data 190 4.14 Cryopumping and vacuum coating 195 - Cryopumping 195 - Vacuum coating 200 4.2 Solubility and permeation 203 A.21 The permeation process 203 4.22 Permeation through vacuum envelopes 208 4.23 Consequences of permeation 211 4.3 Sorption 215 4.31 Sorption phenomena 215 4.32 Ad sorption energies 215 4.33 .Monolayer and sticking coefficients ...... 221 4.34 Adsorption isotherms 224 4.35 True surface 226 4.36 Sorption of gases by ahsorbants 229 - Sorption by activated charcoal 229 - Sorption by zeolites 231 - Sorption by silica gel 232 4.4 DeBorptlon-outgassing <•• 233 4.41 Desorptlon phenomena 233 4.42 Outgaseing 234 4.43 Outgaesing rates 238 V

Page 5. Production of low pressures 243 5.1 Vacuum pumpB 243 5.11 Principles of pumping 243 5.12 Parameters and classifications 244 5.2 Mechanical pumps 248 5.21 Liquid pumps 248 5.22 Piston pumps 250 5.23 Hater ring pumps . 252 5. 24 Rotating-vane pumps 253 - Gas ballast , 257 5.25 Sliding-vane pumps 261 5.26 Rotating-plunger pumps 264 5.27 Roots pumps 265 5. 28 Molecular pumps 267 5.3 Vapour pumps 269 5.31 Classification 269 5.32 Vapour ejector pumps 271 5.33 Diffusion pumps 274 - Pumping speed 274 - Ultimate pressure 276 - Roughing and backing 277 - Pump fluids 279 - Fractionating pumps , 282 - Back streaming and back-migration 283 - Characteristic curves 284 5.4 Ion pumps 286 5.41 Classification 286 5.42 Ion pumping 287 5.43 Kvapor-±on pumps ?PS - Small evapor-ion pumps 289 - Large evapor-ion pumps 290 - The Orbitron pump - 292 5.44 Sputter-ion pumps 294 VI

5.5 Sorption pumps 298 5. SI Mature of sorption pumping 296 5.52 The sorption pump 302 5.53 Multistage sorption pumping 303 5.6 Cryopumping 308 5.61 Cryopumping mechanism 308 5.62 Cryopumping arrays 316 5.63 Cryotrapping 320 5.64 Cryopumps 323 5.65 Liquid nitrogen traps 324 5.7 Gettering 328 5.71 Gettering principles 328 5-72 Flash getters , 331 5.73 Bulk and coating getters 334 5.74 Gettering capacity 336 5.8 Pumping by dilution 337 5.9 Measurement of pumping speed 333 5.91 Methods of measurement 338 5.92 Constant pressure methods 338 5.93 Constant volume methods 343 5.94 Measurement of the pumping speed of mechanical and diffusion pumps 344

6. Measurement of low pressures 347 6.1 Classification and selection of vacuum gauges 347 6.2 Mechanical gauges 349 6.21 Bourdon gauge 349 6.22 Diaphragm gauges 349 6.3 Gauges using liquids 354 6.31 U-tube manometers 354 6.32 Inclined manometers 355 6.33 Differential manometers 356 6.34 The Dubrovln gauge , 356 VII

Page 6.35 The McLeod gauge 359 - Sensitivity and limitations 359 - Raising systems 365 Forma of McLeod gauges 367 6.4 Viscosity (molecular) gauges 371 6.41 The decrement gauge 371 6.42 The rotating molecular gauge 373 6.43 The resonance type viscosity gauge 374 6.5 Radiometer (Knudsen) gauge 374 6.6 Thermal conductivity gauges 377 6.61 Thermal conductivity and heat losses 377 6.62 Pirani gauge 379 6 ,63 The thermocouple gauge 382 6.64 The thermistor gauge , 384 6.65 Combined McLeod-Pirani gauge 385 6.7 Ionization gauges 385 6.71 The discharge tube 385 6.72 Hot-cathode ionization gauges 386 - Principles 386 " Common ionization gauge 389 - Bayard-Alpert gauge 392 - Lafferty gauge 392 - Klopfer gauge , 395 6.73 Cold-cathode ionization gauges 396 - Penning gauge 396 - The inverted magnetron gauge 397 - Redhead magnetron gauge 398 6.74 Gauges with radioactive sources 399 6.8 Calibration of vacuum gauges 401 6.81 General 401 6.82 McLeod gauge method 401 6.83 Expansion method ,... 401 6.84 Plow method 402 6.85 Dynamical method - 403 VIII

6.9 Partial pressure measurement: ...... 404 6.91 General 404 6.92 Magnetic Reflection mass spectrometers 405 6.93 the trochoidal mass spectrometer 408 6.94 The omegatron 409 6.95 The Farvifcron 430 6.96 The quadrupola 412 6.97 Time-of-flight mass spectrometers ...... 413

7. Hifth, vacuum technoloj&y 415 7.1 Criteria for selection o£ materials ...... 415 7.11 General 415 7.12 Mechanical strength , 415 7.13 Permeability to gases 417 7.14 Vapour pressure and gas evolution ...... 417 7.15 Working conditions 417 7.16 Metal vessels and pipes 418 7.17 Glass vessels and pipes •. • 419 7.18 Elastomer and plastic pipes ....-...... •«• 420 7.2 Cleaning techniques 422 7.21 Cleaning of metals 422 7.22 Cleaning of glass 428 7.23 Cleaning of ceramics 429 7.24 Cleaning of rubber 430 7.25 Baking 430 7.3 Sealing techniques 430 7.31 General* classification » 430 7.32 Permanent sealB 4 31 - Welded seals 431 - Brazed seals 438 - Glass-glass seals ...... 446 - Glass-metal seals • 449 - Cerssaic-metal seals 459 IX

Page 7.33 Semipermanent and demountable seals 460 - Waxed seals 461 - Adheslves (Epony) 461 - Silver chloride 474 - Ground and lapped seals 475 - Liquid seals 479 7.34 Gasket seals 481 Sealing mechanism 481 - O-ring seals 493 - Assembly and maintenance of O-ring seals 502 - Shear seals 504 Knife edge seals 505 - Guard vacuum in the seals 506 7.35 Electrical lead-throughs 508 7.36 Motion transmission 512 7.37 Material transfer into vacuum 518 - Cut-offs 518 - Stopcocks 520 - Valves 521 - Controlled leaks 526 - Vacuum locks 526 7.4 Leak detection 531 7.41 Leak rate and detection 531 7.42 Leakage measurement 537 7.43 Leak location 543 7.44 Sealed unit testing 544 7.45 Sensitive leak detection methods 547 - Halogen leak detector 547 - Detectors using vacuum gauges 548 Principle of operation 548 Single gauge detection 551 Barrier leak detection 552 Differential" leak detection 554 - Mass spectrometer leak detectors 554 - Ion pump as leak detector 555

8. Vacuum systems 559 8.1 Basic criteria of design 559 8.2 Evaluation of the gas load 560 - Leakage 562 - Out gassing 566 X

Page - Permeation 569 - Pumping requirements 572 8.3 Vacuum chambers * 572 8.4 Pumping combinations 573 8.5 Rules for operating vacuum systems 576 - References (for Figs. 8.1 - 8.3) 579 - 1 -

1. INTRODUCTION 1.1. The vacuum Although the Latin word vacuum means "empty", the object of vacuum techniques is far from being spaces without matter. At the lowest pressures which can be obtained by modern pumping methods 3 there are still hundreds of molecules in each cm of evacuated apace. According to the definition of the American Vacuum Soviety, the term "vacuum" refers to a given space filled with gas at pressures below atmospheric, i.e. having a density of molecules less than about 19 3 2.5 x 10 molecules/cm . It can be concluded that the general 'term "vacuum" includes nowadays about 17 orders of magnitude of pressures (or densities) below that corresponding to the standard atmosphere. The lover limit of the range 1B continuously decreasing, as the vacuum technology improves its pumping and measuring techniques. 1.11. Artificial vacuum Here on the earth wcuum is achieved by pumping on a vessel, the degree of vacuum increasing as the pressure exerted by the residual gas decreases below atmospheric* Measuring a system's absolute pressure is the traditional way to classify the degree of vacuum. Thus we speak of low, medium, high and ultrahigh vacuum corresponding to regions of lower and lower pressures (Fig.1.1). At first approach the limits of these various ranges may look as arbitrary, since for each range there are specific kinds of pumps and measuring instruments. In fact, each of these various vacuum ranges correspond to a different physical situation. In order to describe these situations it is useful to utilize the concepts of molecular density, mean free path, and the time constant to form a monolayer, concepts which are related to the pressure, as veil as to the kind of gas and its temperature. - 2 - Molecular density, n(cm~*) L o EL 2L S_ I I I I I I I I I I ' 1 ' I I 1 S» S- SK S* S- "> Molecular incidence rate.Kcnr3***) 760 5? 5-

1 «V -o, -0 - I 61 I -1 q ,-, 2 6,- _ z r « -o ,» *•3•

*

Mean free path Mem) %• O d O O O Oi Oi

Tallies 1.1, and 1.2 list values of these terms.

TftBUi: 1.1.—Aiotecutar Incidrnrt Rair uiut Time To Farm a Mcnvlaytr lor Air el 25' C

Mi-nn fn-f- Molcrnlnr Timn to FrnjKiire, MotecitlurilL'nsity. ini'iilcMivp (nrml torr V i-in nilrvl\ mcnnlnypr, molrc-i]l['/i:m= s>t /.KM.

2.48 X 10'" (1.61) X W" 3.14 X 10" •tea x w* I 3.24 X ID" RJXI\ ID* 4.13 X 10" 3.00 X 10"* ID"* 3.24 X 10" JUSX1W 4.13 X 10" 2,00X10* in* 3.24 X 10'" 5JM X 10* 4.13 X 10" a.on x io* 3.24 X 10' ftflnxiw 4.13 X 10" 2.0ft X IP 10 " 32* X 10' .1rei y in* 4.13 v 10' MUX 10* IO-1" 354 X 10' js.mx i«" 4.13 X I03 2JOO y W

TAILE IZ—Moletuhr Incidrner Half anil Timr To Farm n Hondaytr for Same Common Hairs at :':i f; unit 10' Tarr

Mnlmilnr Mnlfciiliir Molrctihir Mwn frw Timn lo funn I CM weight, HI Ittlll, Kraut/moti- I.CDl monolayer iiiulmiWrin* jrr f.(re

Air 29 3.7JXIO' 1.13 X 1ft" IJMXlO" 2R 3.7fi r>.m 4.10 i.m a 32 3fi1 fl.41) SUB 222 iu 2 2.75 nai 1B.64 0.975 lb 4 2.IS M.7* 11. tl 21ft Hrf) ..„ 18 im 337 S31 1.01 CTO, 44 45Ti 8JM sax 1J» A 40 3.67 Ml 3 m 2,13 - 4 -

By analysing the ranges shown on Fig.1.1 and the values of the terms listed In Tables 1.1 and 1.2, It results that the physical situation* characterizing the various vacuum ranees are; Low (and Medium) vacuum - the number of molecules of the gas phase Is large compared to that covering the surfaces, thus in this range the pumping la directed towards rarefying the existing gas •* -2 phase. The range extends from atmospheric pressure to about 10 Torr.

High vacuum - the gas molecules in the system are located principally on surfaces, and the mean free path equals «? is greater than the pertinent dimensions of the enclosure. Therefore the punping consists In evacuating or capturing the molecules leaving the surfaces and Individually reaching (molecular flow) the pump. This is the range where particles can travel %n the vacuum enclosure without colliding to other particles. The range extends fron about 10~ to

10"7 Torr. ultra-high, vacuum - the time to form a nonalayer is eqttijfr or

longer than the usual time for laboratory measurments, thus "clean1' surfaces can be prepared and their properties can be determined before tha adsorbed gas layer is formed. -7 -14 This vacuum range extends from about 10 Torr to 10 Clover limit decreasing with progress of technology)* Composition of the gas. - HTillc the total pr ssure in a vacuum chamber decreases, the composition of the gas phase changes as well. In the low vacuum range the composition of the gas mainly reasemblea to that of the atmosphere (Table 1.3). In the high vacuum range the composition changes continuously, towards one which contains 80 - 90 percent water vapour. The water molecules come' from the surfaces. As pumping la continued and heating is applied,the carbon monoxide content Increases, in the ultra-high vacuum range hydrogen is the dominant component (Table 1.3), coming mostly from the bulk of the materials (permeation). Table l[.3. - Gas compositions

Atmosphere (^-) ultra-hinh vacuum Component: Percent by Partial pressure Partial pressure ' volume Torr (2) Torr (3)

2 11 N 78.OS 5.95 x 10 2 x ID' ' 2 -

°2 20.95 1.59 x 102 3 x 10"12 -

Ar 0.93 7.05 6 x 10"12 ; 1

1 11 1Z co2 0.033 2.5 x 10" 6.5 x I0" | 6 x 10"

Ne 1.8 x 10"? 1.4 x 10"2 5.2 xlO"11 | I

He 5.24 slO* 4 x Kf3 3.6 x 10"1 J

Kr 1.1 x 10"4 8.4,x 10"4 i

5 4 9 u H 5.0 x 10" 3.8 x 10" 1.79 x 10" ; 2 x io~ 2 i Xe • 8.7 x 10"6 6.6 x 10"5 ~" i V 1.57 1.19 x 101 1.25 x lO^"10 j 9 x 10"13 -11 2 x lO"4 1.5 x 10 "3 7.1 x 10 \, 3 x la"" "4

°3 7 x 10"6' 5.3.x 10 "5 -. '

N 0 5 4 2 5 x 10" 3.8 x 10" < -

CO - -' 1.4 x 10-10 9 x 10-12

(1) F.J. NOrton, IraoB. 2nd Internet. Vacuum Congress, Pergamon Pres

1.12. *atur«l vacuum •'Vacuum on urth

t Katun uses "low vacuus technique*" in some of the function* of Ufa of animals, but no natural high vacuum is known on earth. Some of these "applications" are very vital* aa our own respiration, others like the vacuum action of mosqultos, are rather bothering. Buman beings are pumping to about 740 Torr during their respiration, and may achieve pressures as low as 300 Torr by suction. The octopus la able to achieve pressures of about 100 Torr.

Vacuum in space As the pressure of 760 Torr at sea level is a result of the "atmospheric column", the pressure decreases with the altitude. Up to 100 km altitude (troposphere and stratosphere) the pressure decreases quite regularly by a factor of 10 for each increase in . —3 altitude of 15 km, which results in a pressure of 10 Torr at about 90 km altitude. At higher altitudes high vacuum exists. The ionosphere (100 - 400 km) contains a large number of ionized atoms, and its pressure decreases only by a factor of 10 every 100 - 200 km. This decrease results in a pressure of about 10~ Torr at an altitude of 100 km. According to Fig.1.2 above 400 km, ultra­ high vacuum conditions exist. Above this altitude the pressure decreases at an even slower rate, thus at 10000 km a pressure of about 10 Torr exists. Since the average spacecraft, travels at a velocity considerably In excess of that of the average gas molecule, the pressures measured on spacecrafts is actually determined by the spacecraft velocity and gas particle concentration, Thus the diagram (Fig.1.2) of the high altitude atmosphere is expressed in concentration (density) units. The gas molecule concentration (density) is estimated to fall in the shaded area of Tig.1.2, since the density varies with the time of day and the amount of solar activity. At an altitude below 200 km, the atmosphere Is essentially air. Between 200 - 1000 km the Particle Conwnlfofion, porHefes/em'

Figure 1.2. - Characteristics of high-altitude atmosphere surrounding the Earth, to an altitude at 105 kllonetres. gas is principally atomic nitrogen and oxygen, which may be largely ionized at periods oE solar maxima. There is same evidence of an appreciable amount of helium at shout 700 - 1000 km altitude. Above an altitude of 1500 km, the p,as consist a of neutral atomic hydrogen, protons and electrons.

•1,2, Fields of application and Importance 1.21. Applications of vacuum techniques The large variety ot applications of vacuum can be classified either according to the physical situation achieved by vacuum technology (labia 1.4) or according to the fields (Industries) where th« aaellcatioa beloaaa (Fig. 1.3).

labia 1.4. - appllcatloae of vacuus Technique

Ifcyaical Objective Applications

fcchieva pressure Boldins, Llftlm diffai Transport (pneuaatic, cleaners, filtering) Foralng

t—ove active Laapa (incandescent- , fluores­ itaospherlc cent, alactrlc tubas) Eonetituents Halting, sintering Packaging Low aolacular Encapsulation,' Leak'detection density taaowa occluded Drying, dehydration, concentra­ at disaolvad tion, Freeze drying, Lyophyllsation, Degassing Eaprsgnation

Thermal insulation mergy trans far Electrical insulation Vacuus alcrobalance Space sinulstion

Electron tubes, cathode ray, Large television, photocells, kvold pbotoaultlpliers. X-ray aaan free Accelerators, storage rings, collisions path •ass speetroaatera, Isotope ssparstors Electron adcroscopss Electron beam welding, nesting Coating (evaporation, sputtsring) Molecular distil1stion

Long ilean Friction, adhesion, ealsalon aooolayar foraation surfaces atudies^Hstariala testing for tiaa space. «pnic*TioHs or VACUUM TICHNIOJI

f EVUWItW J f fBM.*!"" ) (

1 I WO W5 n PHARMACEUTICALS * MEDICALS

J MECHAHKAL MDUSTR^

ALLOT FA»H - l*.i*IL-*I CP^WK ELECTRIC WSUSTW

(KHTI*ATOM1 hvtlSMIMJ |

(^)(^(±) r —I— -=—!— —T ' 1— i MGHAt+C»L : SPACE INCMEKK ^J •""-"•< ITCMWLW ;si".~ —~

11MMMCS CttMUCE 1 .„.., """"' - 10 -

Obviously •ach of tha applications of vacuus techosjlogy utilises on* or Mir* physical situations obtained by rarefying the gas. Some of them achieve products or facllltlss In which the vacuum agists during all their lif• (lamps, tubes, accelerators, ate), others -only use vacuum technology as a step la the production, the final product being u**d ia atmospheric conditions (vacuum coating, drying, taeteammtlon., etc.). According to the physical situations created by vacuum the various applications may be resumed (Table. 1,4) as follows: The pressure difference achieved by evacuating a vessel can realise forces on the walls up to 1 kg/cm . These forces ass us*d'--•*. y fox holdine, or lifting solids, for the transport of solids or liquids, emd for forming (shaping) objects. Plastic or rubber cups applied on surfaces so that tha air be eacdsded from the cup* can hold small objects. The same principle is used to fasten tools on work tables (chucks). Here the middle part of a larger rubber membrane forming the base of the Cool is mechanically pulled away, to form a vacuum enclosure with its periphery sitting on the table.

By using sniffers which are evacuated after being placed with.. „ their mouth on the object to be lifted, very small objects can be precisely lifted and trensfered (e.g. filaments in the mass production of leaps). Relatively large (flat) objectB can be lifted (platee, cars) if the mouth of tha lifting cup is large, 5-7 tons can be lifted with a mouth of 1 m . The vacuum cleaner is the simplest exanpLe of a widely used vacuum transport system. Vacuum cleaners are usually able to achieve 2 pressures of 600 Torr, thus to euck objects of tens of grans/cm , Vacuum ttansport systems foe grains and powders are based on features similar to vacuum cleaners. The pneumatic transport systems connecting post officii in Paris or London, are examples of very lsrge vacuum transport facilities. - 11 -

That of Pari* has a length of about 300 km, of double, 60 or 80 mm bore tubes; they are using a pressure of 450 Torr for the transport from post offices towards pumping stations, and an over pressure of C.B atn for transport in the opposite direction. The transport cylinders containing the letters move at speeds of 8 - 10 m/s. It is interesting to mention that pneumatic trains working on this principle were in function at Dublin (Ireland) and Saint-Germain. (France) in the 1840 - 1860 rears. Vacuum Is commonly used in laboratory and chemical industry to accelerate filtering speed. The pressure difference obtained by evacuation is used in the vacuum forming (molding) of plastics. Tha necessity of removing the chemically active constituents of the atmosphere (oxygen, water vapour) by vacuum pumping appeared together with the invention of the incandescent lamps. In order to avoid oxidation of the filament heated at very high tmperatures, it must be in an inert atmosphere. This atmosphere is constituted either

by a high vacuum (about 10- Torr), or by an in&rt gas filled into the lamp after its evacuation at a high vacuum. The possibility of evacuating large chambers at a high vacuum level is used in vacuum metallurgy to protect active metals from oxidation during melting, casting, sintering, etc. Vacuum packaging of food, or materials sensitive to reactions with atmospheric components is used at a large scale in modern industry, the level of evacuation being usually in the low vacuum rang*. Vacuum encapsulation of sensitive devices (translators, capacitors, etc) is oftan carried out at high vacuum levels. The leak testing techniques using high sensitivity detectors can control the tightness of tha encapsulation. Vacuum technology is uaad to remove humidity from food, chemicals, pharmaceutic products, concrete, etc., and occluded (dissolved) gas from oils, plastics, etc. The fabrication of fruit juice, and concentrated milk* ere examples of large scale productions based on vacuum concentration* This process does not require extensive heating in - 12 -

order to evaporate th* uttr or solvents contained in th* product*• Sy using tb« vacuoe drying proceaa In conjunction with cooling, the protects are first frozen, tht water being than removed by sublisetion. This £a the beale feature of fre*s* dry lag. In the, product! of freese drying th* final watar content la vary low, chesdcel cbangaa ar* slnlsal, volatile conatltuanta ara essentially kept In. th« product (e.g. Instant toffee), coagulation la avoided (blood plasm) and storage propsrtlas ar* excellent. Vacuus impregnation process conalat In removing tbe occludad humidity ox gaaee, and filling their plaea by another materiel. Although th* comsonly known Impregnation processes arc those used to improve tba dislsctrlc propartias of insulations (aotor windings, capacitors, cables), ¥acuum impregnation techniques ara also used to ineraasa strength, or dacraaaa combustibility of textiles, paper, wood, etc.

High vacuus is s thsrsal and alactrical insulant. Ibis property ia used in th* Dewer flaaka for tha atoraga of liquid air, nitrogen, helius, etc, as well aa in the "thtrsos flaaks11 used to keep cool drink or food. Both are double-walled flasks, the space between the walla being evacuated at high vacuus. The electrical insulation properties of high vacuum are used in vacuus switches, as well as in high voltage devices (accelerators, tubes). As th* snsrgy transfer in outer apace ia similar to that which occurs In ultrs-hlgh vacuumrSpscs lisulatlon became one of th* sophltlcatsd sppllcstlons of vacuus technology. Space simulator chasbers extend to volumes of sore than 1000 m ,and eose of than are evacuated to the lowest pressures which can be achieved today* Vacuus slcrobalanc* techniques us* high and ultra-high vacuus to avoid any "background" provening frost ens surrounding gas. Th* large seen free paths existing in high vacuus, is used to svold collisions between molecules, electrons. Ions in electron - 13 -

tubes, photocells, cathode ray tubes, X-ray tubes, accelerators, mass spectrometers, electron microscopes, etc. This same property is used in vacuum coating plants where the coating material evaporated from its source reaches the substrate being coated, by travelling in straight lines, without collisions, in this vay thin films are deposited for a large number of optical, research, or ornamental uses. Molecular distillation is another field where high vacuum is used in order co obtain very pure fractions by evaporating and condensing the molecules without any collisions to other gas molecules. Ultra-high vacuum permits to study the real properties of surfaces (friction, adhesion, emission, etc) since at theee low pressures the tines of formation of a monolayer are sufficiently long (hours, Fig.1.1).

1.22. Importance of vacuum technology The list of applications of vacuum technology include a large number of items which becaae symbols of the progress. From this point of view the importance of vacuum technology is evident. The size of the field can be shown by the number of persons (scientists, engineers, technicians and workers) involved in the world in the various aspects of vacuum technology. This number was in 1965 over one million, receiving a total of salaries of about 3 milliard dollara. At that time it was evaluated that more than 4 milliard lamps and 1 milliard electron tubes were produced per year. The number of persona active in the progress of vacuum science and technology can be evaluated to tens of thousands, according to the number of members of 1UVSTA. I.U.V.S.T.A. is the International Union for Vacuum Science, Techniques and Applications, which includes (in 1970) 18 National Vacuum Societies. The number of commercial firms producing general and specialised vacuum equipment ranges to about 100, (companies ranging from 100 to thousands of persons). - 14 -

1.3. Main stsE— in the history of vecoum techniques It can bs considered that ths history of vacuum techniques begins In 1643, whan Torrtcalll discovered tha vacuus which ii produced at tba top of a column of mercury Whan a long tuba aaalad AC ona and la filled with Bcrcury and inverted in a trough containing Bg. Tha pioneer period A vacuum techniques continue* up to tha invention of the electric leap. In this period important theoretical and experimental scientific progress is achieved In tha fundamental* of gas laws (Boyle-Maxlotte, Charles-Gar Lusac, Bernoulli;' Avogadro, Maxwell, Bolt*mann,etc). The first progress in the practical use of vacuus ms connected to the mechanical effect* which can be achieved by using the pressure difference between vacuum and atmosphere. Tha classic experiment of Guerieke (1654) showing that the two hemispheres of an 119 cm "evacuated" ball cannot be separated by pulling with 2x1 horses, demonstrated the atmospheric forces.

The application of this knowledge, to drive railway cars (Dublin) was used only e few years, but the pneumatic-vacuum transport systems begun in 1850 - 1860 in London and Patio are still in use {slighthy modernised!). The development of the incandescent lamp (Edison, 1879) was also a consequence of the pumping system Invented during previous yeers (loepler, Sprengel see Table 1.5). The He Leod gauge (1874) gave for the first time the possibility of measuring low presaures. The Incandescent lamp has shown the ussfulness of low molecular dsnsitlea (removel of the active etmoapherlc constituents), the cathode ray tubs of Crookes (1879) was the first Application, of the Increased mean free path, while the Dewar flask (1B93) constitutea the firat application of vacuum thermal insulation.

The invention of the vacuum diodes (1902) and trlode (1907), and of tha tungsten filament (1909), bsgln the development of the electron tubes, sad brought that of the Incandescent lamps to a maturity - 15 -

(Langmuir, 1315). The "quality" of the vacuus used in the production of the incandescent lamps revealed to be insufficient in the new field of electron tubes, which brought to research and development work cm pumping and measurement.

The Piranl Gauge (1906)t Gaede (1915) and Langmuir (1916); diffusion pumps, and the hoc cathode ionization gauge (1916), opened the possibilities of the high vacuus technology. The development of high vacuum technology continued up to the second world war, in the years 1935 - 1936 receiving three new items: the gas ballast pumps, tbo oil diffusion pump, and the Penning cold cathode ionization gauge, items which together with the Pirani gauge remained up toaow.tne- usual components of most vacuums systems. After 1940 vacuum technology had a very large development in the direction of equipment for nuclear research (cyclotron, isotope separation, etc,), vacuum metallurgy, vacuum coating, freeze? drying, etc.

Up to 1950 the usual vacuum range extended to 10 - 10 Torr, Perhaps lower pressures were obtained also before-, but no possibility existed for measuring lower pressures. The Bayard-Alpert gaug* Q&50) opened the way to measure lower pressures, in the range called later ultra-high vacuum. The ion-pumps produced after 1953, permitted to obtain very low pressures, and the so called " clean vacuum". In the last decade, the space research gave a new quantitative' jump to vacuum techniques, by tie numerous vacuum problems which bad to be solved for space missions. - 16 -

Table 1.5. - Stages in the history of vacuum techniques.

Tnr Author Work (Discovery)

1643 Evangelise* lorricelli Vacuus in the 760 am aarcur column

1650 Blaise Pascal Variation of Hg column with altitude

1654 Otto von Guerlck* Vacuum plato-pumpe; Magdeburg hemispheres

1662 lobart Boyle Pressure-volume law of idecl 1679 Mas Kariotte gasu

1775 A.t. Lavoisier Atmospheric air: * mlxtttrs of nitrogen and oxygen

1783 Danial Bernoulli Kinetic theory of gases

1802 J.A, Charles Volume temperature lav of J. -Gay-Lussac gaass

1810 Kedhurst Propoas first vacuus post lines

1811 Amedeo Avogadro Constant molecular density «f gases

1843 Clagg and Saauda First vacuusi railways (Dublin)

1850 Gelsslar and Toepler Mercury column vacuusi pump

1859 J.K. Maxwell Gas nolecul* velocity laws

1865 Sprengal Mercury drop vacuum pump

1874 H. HcLaod Compression, vacuum gsugs

1879 T.A. Edison 'Carbon filament, incandescent lamp

1879 W. Crookaa Cathode ray tube - 17 -

Table 1.5 (continued)

] 1881 J. Van der Waala Equation of state of reel gases

1893 Janes Dewar Vacuum insulated flask

1895 Wilhelm Roentgen X-raya

1902 A. Fleming Vacuum diode

1904 Arthur Wehnelt Oxyde-coated cathode

1905 Wolfgang Gaede Rotary vacuum pump

1906 Maixello FIrani Thermal conductivity vacuum gauge

1907 Lae de Forest Vacuum triode

1909 W.D. Coolidge Powder metallurgy of tungsten Tungsten filament lamp

1909 M. Knudsen Molecular £lo? of gaaes

1913 M. Gaede Molecular vacuum pump

L915 W.D. Coolidge x-ray tube

1915 W, Gaede Diffusion pump

1915 Irving Langmilr Gas filled incandescent lamp

1915 Saul Dushuan The kenotron

1916 Irving Langmuir Condensation pump

1916 O.E. Buckley Hat cathode ionisation gauge

1923 F. Holwack Molecular pump

1935 W. Gaede Gas-ballast pump

1936 Kenneth Hickman Oil diffusion pump

1937 P.M. Penning Cold cathode ionisation gauge

1950 R.T, Bayard and Ultra-high vacuum gauge D. Alpart

1953 H.J. Schwartz, Ion pumps R.G. Herb, etal - 18 -

1.4. friforaturfc sources.

Vacuum Technology is baaed today on a very extensive literature of books, journals and conference transaction dealing exclusively with the various aspects of the subject. The list vhich follows gives the names of the most inoortant literature sources currently- used in recent yean.

The list dees not include the books appeared between 1920-19+0, which have no aore practical interest, their content bain? republished in the nor«? recent works listed.

?or an historical interest, it must he mentioned that the first back published on vacuum was in latin :

Ottonis de Suericke: Experienta Nova Magdeturgica de Vacuo Spatio, J. Jansson, Amsterdam, 1672 which was republished in German, by VDI - Verlag, Dfisseldorf in 1968.

As regarding the journal miblications, besides the journals listed which are exclusively dedicated to vacuum urobleqis, a large number of papers on vacuum techniques were and are published also in: British Journal of Applied Physics, Experimentelle Technik der Physik, Japan

Journal of Applied Physicsr Journal of Applied Physics (USA)Journal of Scientific Instruments* Materials Evaluation, Nuclear Instruments and Methods, Review of Scientific Instruments, Research and Development*

SQL Review ant1 Solid State Technology.

Abstracts of rewires published on subjects of vacuum technology are currently listed in HASA - Scientific and Technical Aerospace Seporta, a miblicstion *Mch anpeara twice a month. - 19 -

Publications dedicated exclusively to Vacuum Techniques

1„ Jc Strong, Procedures in Experimental Physics, Prentice-Ball, New York, 1936, p.93-187,

2. CH.Bachman, Techniques in Experimental Electronics, J. Wiley, New York, 1948, p. 1-67, 89-140,

3. A. Guthrie and R.K.Wakerling, Vacuum Equipment and Techniques, Mc Graw-Hill, New York, 1949*

4. E. Jaekel, Kleinste Dracke, Ifare Measung und Erzeugung, Springer, Berlin, 1950, 302 p.

5. H. Auwartar, Ergebnisae der Hocbvakuumtechnik und der Phya$k dunner Schichten, Wisserschaftliche Verl, Stuttgart 1957, 282p„

6. fi. CimmpeiXi- ulementa de Technique du Vide, Dunod, Paria, 1958, 214p=

7. K. yorand, Traite Pratitme de Technique du Vide, Soc. G.B.P., Paris," 1958, 347p.

6. R.F. Buns hah, Vacuum Metallurgy, Eeinhold, New York, 1958, 472p. 9. 3.C. MBnch, Neues und Bewahrtes ana der HochvakuuiatechniJc, VEB Knapp, Halle, 1959-

lO.W.H.Kohl, Materiala and Techniques for Electron Tubes, Ksinhold, New York, I960, 638p.

ll.J.V. Cable, Vacuum Proceasing in Ifetalworking, teinhold, Hew Tork, I960, 202p. - 20 -

12. L. Holland, Vacuus Dtpoaition of Thin Pilao, Chapaan 4 Ball* London, I960, 542p.

13. H. Pirani and J. Tarvood, Principles of Vacuum Englneerinj?, Chapman A Hall, London, 1961, 578p.

14. J. Dalafoese and G. Bongodin, Les Calculs de la Technique du Vide, La Vide, 1961, 107p.

15* AJI. Turnbull, R.S. Barton and J.C. Riviere, An Introduction to Vacuua Technique, G. lennea, London, 1962, 19Pp.

16. S. Dosnaen, Scientific Foundations of Vacuum Technique, J. Wiley, HewTork 2nd £>d.,1962, 806p.

17. H.L Bscnbach, Piaktikum der Hochvakuumtechnik, Akadeaiache Verlag, Leipzig, 1962, 243p.

IB. 3. Eucb, Einfannmg in die Ailgeaeine Vakuuiitechnilc, Viasenschaftlicbe Verlag, Stuttgart, 1962, 207p.

19. G.A. Boutry, Physique Appliqnee aux Industries du Vide et de 1' Electronique, Kaason, Paris, 1962,388p.

20. A.B, Barrington, High Vacuus Engineering, Frantlce-Hall, Bnglewuod Cliffs, N.J.,1963.212p.

21. H.A. Steinherz, Handbook of HiA. Vacuum Engineering, Heinhold, Bav York, I965,358p.

22. K.tf. Hoberts and T.A. Vanierslice, Dltra-high Vacuum and its Applications, Prentice-Hall, Enfflewood Cliffs, H.J., 1963.

23. J2,A, Trendelenburg, Dltrahoenvakuum, Verl. Braun, KarlBruhe, 19<53»196p.

24. JJL* Belt, Vacuum Techniques in Metallurgy, Fergamon Press, Oxford, 1963,231u.

29. A. Guthrie, Vacuum Technology, J. Wiley, New York, 1963,532D. - 21

26. S.V. Spiuks, Vacuum Technology, Chapman 4 Hall, London, 1963.

27. J.H. Leek, Pressure Measurement in vacuum Systems, 2nd Ed, Chapman ft Hall, London, 1964.

28. V. Pupr>, Vakuumtachnit, I. Grundlagen(l962)lC9D, EE. Anirendungen (I964),3l2p. Verlag K. Thiemig, Wlnchen.

29- G. Lewin, fundamentals of Vacuum Science and Technology, He Graw-Hill, New York, 1965, 248p.

30. P. Roaebury, Handbook of Electron Tube and Vacuum Techniques, Addison-Wesley, Heading USA, 1965,597n.

31. V.F, Brunner and T.H. Batzer, Practical Vacuum Techniauea, fieinhold, New York, 1965,197D.

32. K. Wutz, Theoiie und Praxis der Vakuumtechnik, P. Vieweg 4 Sohn, Braunschweig, 1965.439D.

33. CM, Van Atta, Vacuum Science and Engineering. Kc Graw-Hill, Hew York, 1965.459D.

34. B.D. Power, High. Vacuum Punning Equipment, Chapman ft Hall, London, 1966,412p.

35. E. Diels and R.J. Jaeckel, Leybold Vacuum Handbook, Pergamon Press, Oxford, 1966,360p,

36. A. Both, Vacuum Sealing Techniques, Pergamon Press, Oxford, 1966,845n.

37. D.J. Santeler, et al, Vacuum Technology and Space Simulation, NASA, Washington, 1966,305p.

38. J* Yarwood, High Vacuum Technique, 4th Ed, Chapman 4 Hall, London, 1967.274p.

39. L. Ward and J.p. Bunn, Introduotion. to the Theory and Practice of High Vacuum Technology. Butterwortho, London, 1967,216p. - 22 -

40. V.B. Kohl, Handbook of Materials and Techniques for Vacuum Devices. Keinhold, leu Tork, 1967,623D.

41. A.E. Beck (Editor}, Handbook of Vacuum Physics. Tol. I. Bases and Vacua <1964.1966)419o. II. Physical Electronics (1965,1968) 598e. III. Technology (l964)270o.

42. 9.1.1!. Denote and T.A. Henpell, Vacuus Syatea Design. Chanson * Hall, London, 1968,223p.

43. PJL Bedhead, J.F. Bbbson and 8.V. Kornelsen, The Physical Baeis of Ultrahigh Vacuum, Chanman eb Ball, London, 1968,49BD.

44. I.W. Hobinson, The Physical Principles of Ultra-histo Vacuum Systems and Equipment Chanson * Hall, London, 1968,270n.

45. V. Green, The Design and Construction cf Small Vacuum Systems, Chapman A Ball, London, 1968,lSln.

46. V. Espe, materials of High Vacuum Technology, Vol. I. (totals and metalloids (l966),912p. II. Silicates (l968),660o. III. Auxiliar/ Materials (l96B),530p. Ferguson Press, Orford.

II. Journals

1. Vacuum (aigland), monthly, reached, volume 20 (1970)

2. Vakuum-Technik (Germany), monthly, reached vol.l9(l9,70).

3. Le Vide (France), Bimonthly, reached vol.25(l970).

4. Journal of Vacuum Science and Technology (U3a), Bimonthly* reached vol. 7(1970).

5. Vuoto Utalv), Trimestrial, reached vol.3(l970). - 23 -

III. Conferegoq Trspaactions

1. Vacuum Bymposxuoi Transactions) of the Conferences held hy the American Vacuum Society, Volumes Edited each yeex between 1954-1963, 10 volumes, Pergaacn Press, Oxford.

2. Advances in Vacuum Science and Technology, Proceeding? of the First International Congress on Vacuum Technology, held Samur (Belgium) 10-13 June 1958. 2 volumes, Pergsmon Preea, Oxford,824D.

3. Transactions of the Second International Congress on Vacuum Science and Technology, held Washington, 16-19 Oct. 196*1 # 2 volumes, Pergamon Press, Oxford, 1351n.

4-. Advances in Vacuum Science and Technology, Proceedings of the Third International Congress on Vacuum Techniques, held Stuttgart, 28 June-2 July 1965, 4 volumes, Pergaaon Press, Oiford, 929D.

5. Proceedings of the Fourth International Vacuum. Congress, held Hanchester, 17-20 Anr.1968, 2 volumes, Inst, of Physics, London, 827*.

6. Transactions of the Vacuum Metallurgy Conferences, Edited by the American Vacuum Society, each year between 1962-1968, 7 voluoes.

7. International SymD. on Hesidual Gases in Electron Tubes and Related Vacuum Systems, held Rome, 14-17 March 1967, Suopl. to Huovo Cimento, Bologna (Italy) 2 volumes, 619p,

8. Vacuum Microbalance Techniques, Proceedings of Conferences held each year between 1960-1966, 6 volumes, Plenum Press, NBW Tork.

9. DeurLeme Colloque International sur lea Amplications das Techniques du Vide a 1'Industrie des Seaiconducteura, Paris, 5-8 O^t. 1966., 151p.

10, Colloque International aur l'TJltra-Vide, Paris, 28-30 Jun, 1967il95p.

11. Congrea International sur lea Applications des Techniques du Vide a la Ketallurgie, Strasbourg, 13-17 HOT.,1967, 280p. - 24 -

12. I* ftelmologi* d» l'Bltra-Vide at das SasaM Praaaioas, YersalllaB, 20-23 May, 1969.307p.

15. Colloatte International Vide et Froid, Grenoble, 1-5 Dec. 19S9,226p.

14. Congrea International sur les Couches Mincea, Cannea, 5-10 Oct. 1970, 664p. - 25 -

2. RARFJIEU) GAS THEORT FOR VACUUM IECHHOLOCY

Comonly used symbols.- A •re* c specific beat at constant pressure p c specific heat at constant volume D dimeter of aperture or tube E energy e charge of the electron F force h height (of a column of liquid) k Boltrmann constant K heat conductivity L length a mass of molecule H molecular weight n number of molecules per unit volume N total ninber of molecules \ number of molecules per sole P pressure q gas flow (molecules per second) r radius R general gas constant R gas constant per mole 0 t time T absolute temperature V volume V velocity w specific mass (per sec, per cm )

(alpha) - accomodation coefficient (tou) - ratio C /C P v - 26 -

it (eta) - vlacoalty l (laabda) - Man f raa path A (lambda) - fnt solacular beat conductivity C (xi) - molecular diaaetar p (rfao) - density, maaa par unit voltaM T (tau) - (tlaw] period 4 Cpbi) - Molecular incidence rate. - 27 -

2.1. Physical states of matter

A collection of molecules can occur either In tbe solid, liquid or gaseous Btate, depending on the strength of the lntermolecular forces and the average kinetic energy pec iulecule (temperature).

The state in which molecules axe most independent fro each other Is called an ideal or perfect gas. This is a theoretical concept which corresponds to the assumptions that: a) the molecules are minute spheres; b) their volume la very small compared with that actually occupied by the gas; c) the molecules do not exert forces upon each other; d) they travel along rectilinear paths in a perfectly random cushion; e) the molecules make perfectly elastic collisions.

Some real gases, such as hydrogen, nitrogen, oxygen, argon, helium, krypton, neon, xenon, approximate closely at atmospheric pressures the behavior assumed for ideal gases. At lover pressures (vacuum) many more gases approach the ideal gases.

Real gaseB, unlike ideal ones, have internelecular forces. At pressures and temperatures where the molecules of the gas are brought close to each other they will begin to form new structures, which will have properties very different from those of the gas. When these new structures begin to form, the gas is aaid to be liquifying.

Figure 2.1 shows a plot of pressure versus volume for different temperatures of a raal gas (e.g. carbon dioxide). Curves A and B , fox which the temperatures are high, are hyperbolas conforming to Boyle's law, describing a behavior assumed for ideal gasea. At temperature T_ , curve C is no longer completely hyperbolic.

A small hump has formed at point P. At still lower temperatures* curves D and E show complete departure from the hyperbola of ideal gases; a flat plateau appears. When the system has the pressure and volume associated with points L or H , the material (CO.) is in the gaseous state. Along the plateau N-0 the temperature and pressure - 28 -

Tig,2,1. - Variation of pressure with volume in a real gas at various

temperatures. T, > T„ > T, > TA > Tc of tb« system axe both constant while the volume changes. At N the material ia in a gaseous atate, while at 0 it la a liquid. At K a fraction of the system is liquid. It is important to note that each curve (Fig,2.1) has only one plateau, that is, there is only one pressure for a given temperature, et which the gas will liquefy. At temperatures higher than that of curve C , there is no pressure at which the gas can be liquefied. Point P ou curve C la called the critical point. Table 2.1 lists these values for some gaaes.

From Fig-2.1 it is clear that at higher pressures the liquefaction process takes place at higher temperatures. The temperature at which a gas liqueflea ia called the boiling point and depends on the pressure of Che syscen. For example to boll at 20"C, water r*quire* tbmc the pressure of Che surrounding be 17.5A Torr, mercury requires 1.2 x 10 Torr, while C0„ requires 42959 Torr (56.5 At«).

Xable 2.1 -SOME PHYSICAL PROPERTIES OF SUBSTANCES AT LOW TEMPHRATURES

JW—IA^IW« C rt^K*. Ct

i'ubttancr r aal fudil °" ' »?• <~f- N.TT. •c 5=? 87c* © s wit. w*« lifi 0 01 10000 5M7 374 0 l^T 0-334 - CO, -2J 767 46-3 24) 454 0-538 llm 111 C.F.CI, -35 47* 35-1 21+1 33-7 0-J74 F(B»< II Ch-CI. — 111 23-B 433 19*0 433 0-5S4 CHI CI, -IM 89 57-9 17*5 51-0 0-323 CM-.CI, -94 5b 32 a 145 7 32-1 0-S82 Sulphur SO, -Hi -101 1-46 94V 137-2 77-7 0-524 2-91—7 Mclhyl clil.irida CH.CI -72 -13 a 100 101 143-1 63 9 0-J3J 2-301 t-CCDIl 11 CTdCI, -138 •Wl US 400 112 O 40 1 oja 1 31 Nil. -77-7 u u-tsis 327 1 112 4 • 111 O-2JS0 07714 W«n ..J CiQF. -JHD 514 S0-O 3I« 059 CHF.Cl -160 -408 MX 'WO 487 0 315 l-l7 C.1I, -117-7 -41 1 0 J8J 1018 V6-S 42 1 0-2254 2-001 Sf CO, -Sfi-fi - 78-4* 137-0 31-0 731 04419 Ficon 11 -III 338 381 0 31| Ircnn 2) 8£' -IM :!T 3* LllVlciic cV -103-D 96 30+ 0-210 12604 Xc -llll -108-1 J On 230 lb 6 »I 1103 3«97 o. -192 7 — If I** 1-46 73< -111 54-6 CXJ7 2-144 Cirbim

fluonJc CF« -IK -121-0 1-62 316 -490 374 0*4 Kr - isT-a -IS)-4 M-8 6.1* 542 owu J-T45 M"^!^ CH. -114 -1(11-S 0424 127-1 - Wl 4JJ 0 1615 0-71U O, -111! -IIWJ 1-140 SI-0 -1114 30-1 0-430 1-4291 A -1W-J -183-86 1-39 39-0 -122-4 480 MX* 1-TM CO -203-0 -191 i 0-719 304 -140-3 343 0-3010 1-2304 N, -210-0 -193-tt 0

t thai (It* condiilom nn ^jtkttlj lfa« pressure exerted by the molecules on Che surrounding atmosphere and liquid Is celled the vapour pressure. The vapour pressure depends oa the temperature of the material. Ilu boiling point of a liquid is that teaperature at which the Tapour pressure of the liquid is equal to the surrounding pressure.

If the vapour pressure is plotted vs. the temperaturek curves as that in Fig.2.2 are obtained. At the right side (below) of these

Fif. 2.2, ~ TIie vapour - pressure curve for water. curves vapour exists, while at the left side {above) the curve, liquid exists. Table 2.2 lists the vapour pressures of water and mercury ac various temperatures*

If (Fig.2.1) the liquid is conpressed below point Q , a second plateau appears (Fig.2.3); it is here chat the liquid undergoes, a change of phase, Into a solid. Table J.2 VAFtlUK PRISNUKfi I1F W*ll K (((1 ) »MI Ml KCVRY (lO(T>

IH3 1.4*10-" US: IO-3; / JO 31.82 2.8 v |0 - =

150 7.4-MO1"' 40 55.32 fi.1^10-3

MO 2.9-' 10->" Ml 92.51 1.27 10- =

-- 130 fc.«8 - 10-'J | 6(» 149.3 2.52*10--

: - 121) 1.1.1:- 10 J ; ™ 233.7 4.82 MO -

- no 1.25 *t0-s no 3iS.I B.Sif *.-(©- =

mo 1.1:10 -• 2.39 111- " w 525.7 1.58- 10" '

. w 7.45v 10" • , too 760 2.72 V 10-'

-BO 4.1 x 10 •» 2.38VIO-" 150 .i570.4 2.80

-70 1.98x10-* 1.6H:-IO-» ! 200 11 650 17.28

-60 8.1 xlO"1 9.89~-'[0-' I 250 29 8(7 74.17

-SO 2.9x10-2 4.94x10-' ' 500 64 432 146.8

-40 9.7xlO"! 2.51 X 10-' 400 - 1574 0 r--.18.9T) J -30 2.9 v 10-' 4.78 x 10-* I 500 - 7691 -20 7.8x10-' I.BlwIO-' 600 - 22.B a tin -10 1.95 6.06X10-1 700 - 52.5 aim 0 4.S8 1.85 v 10-' J 800 ~ 103.3 otiti 10 9.2 4.9vl0-« 900 " 180.9 aim 20 17.54 l.lxlfl-* 1000 ' 7W.S Pig.2.3. - Liquid-to-solld phase change represented by the plateau AB .

The temperature corresponding to the liquid-solid phase change it ataospuerle pressure Is called the freezing^ or melting point. Tin* solid-liquid transition point (freezing it various pressures)

varied according to curves an tliouc shown on Fig.2.4t and their slope is negative or positive, depending if the substanca expands on freezing (e.g. water) or contracts on freezing (*•$. Mercury)* The known experiment of the ice cut by the wire under load, shows that as Che pressure i« Increased, th* "freezing point" is lowered.

Fig.2.4. - Liquid-solid transition curves: e) for substances which expand upon freezing; b) for substances which contract upon freezing. At all points on the vapour pressure curves (Fig.2.2* ths liquid snd vspour coexist la equilibrium, and at all points on ths "frssiioj point" curves (Flg.2.4) the solid and liquid coexist. At the Intersection of these two curves (Fig.2.5) all three phases coexist. The point is called the triple point, of the substance. Values of the pressure and temperature corresponding to the triple point of various substances, are listed In Table 2.1.

Fig.2.5. - Dolling, freezing and sublimation curves a) for substances which expand upon freezing,and b) substances vhlch contract upon freezing. Points A and B refer to the respective triple points.

At pressures and temperatures below the triple point substances ar« changing fro* the solid to the vapour phase without passing through the liquid phase. This process la known as sublimation, and ths Una representing pressure - temperatures at which a solid and Its vapour coexist, la the sublimation curve (Fig.2.5).

Any equation of state which describes the changes In a thermodynamic system must be a function of three variables: pressure, volume snd temperature, and such an equation can be represented by a three - dimensional P - V - T surface (Fig.2.6). The equations of this surface will be discusssd In the following chapterers (2.2, 2.3, 2.4, etc). - 34

Fig.2.6. - Ths P - V - T surface for water end Its projections on the P - V and P - T planes.

2.2. Perfect and real tas laws 2.21. Boyle's law By an ideal or perfect gas we nean one which obeys Boyle's law at all teajpersturea. The relationships established by Boyle (1662) sod Harlots (1679) Is valid for esses over those ranges of pressures snd teaperaturas for which the forces between the «olecules ol eh* gas can be considered neflljlbla, Rsferlng to Table 2.1, at teaperatures higher than the critical point any gu bshaves as a perfect {**• The hyperbolas A and 1 on Flf.2.1 represent the Boyle's law;

P.V - const (2.1)

Considering two different point* on s hyperbola the relationship between then 1> written

p 1v1 • pV v2 (2.2) describing tin isothermal coaprcuuion.

If the apparatus shown In Fl£.2.7a, La considered, it can be seen that for any position of die aercury colum, the pressure P of the enclosed gas is equal to tne .itnospheric pressure P alaus the

Fig.2.7.- a) lioylo's law apparatus; b) KcLeod gauge. gauge pressure caused tiy the column of mercury of height h . The products of the pressure I* and volume V, renalns a constant. This principle was used by McLeod (1874) in his high vacuum gauge, which remained until now the reference gauge in calibrating other vacuum gauges. The essential elements of a HcLeod gauge are thoun in Fig,2,7 b, and consist of a glass bulh with a capillary tube extension on the top, a side arm connecting to the vacuum ayste*, and some means of raising and lowering the mercury level within the gauge. when the mercury level in the gauge t(t lowered below the branch point A. tlM bulb of volume V ia connected to tha ayatem. through alda arat B . Tha ga« in tha bulb ia than at tha same praaaura F aa that In tha system (vacuimO. Whan the mercury lavel ia ralaed, the bulb la cut off from tha aide am and the aample of (as coapreaaed into tha capillary C.. The capillary C, Is in parallel uith a section of the side an B and has the same bore as C .

Since the surface tension or capillary effect in C. and C, •re the law, the difference in level of the mercury is due to the pressure difference resulting froai compression of the gas sample from the large volume V into tha small volume of C above the mercury level. The pressure of the compressed gas in the closed capillary la proportional to P + (h.-b,)-

Aceordlng to Boyle'a law:

IP + (h2-hx)] x A(ho-h3) - PV <2.3) and A (by-fa, Hh-h.)

F • -. - A

A HcLeod gauge may conveniently be read by bringing the mercury lavel up to tha point where h- • h

By using a voliatt V - 1300 cm} , and capillar!** of 0.63 •• bora (A - 0.32 n ), for • difference ah • 1 tsa betveen the level* in the open and closed capillaciea, the pressure which can be determined la

In the second nethod with h. - a . h -h Is a constant, thus the pressure ie A(h -h ) * - v -°^ -H l %£ * -hi (fib), (2.6)

Thus the first method results In e pressure readirg proportional '3 the square of the reading (2.5) whereas the second nethod leads to ,2.5) In which the pressure Is proportional to the first power of the . aciing (linear scale).

Details on the McLeod gauge will be discussed In Chapter 6.

2.22. Charles' law Charles and Gay-Lussac (1802) observed that at constant voluie chu pressure o£ the gas increases linearly with its teaperature, and that at constant pressure, the same phenomena happens with the volume. Their experimental results were described by the relations:

F„ - P (1 + fit) C2.7)

V. - V (1 + 0t> (2.8J t and V preaaur* ttd voluat «t O'C. The extrapolation of these expranaleas to P - V - 0 , bu shown that thle would theoretically happen at t - -273"C. On this basis th* ebsolute tenoeracure icala was established, vbera the xaro point of the scale was set at t - -273.16*C exactly, ao that the teaperatur* T In *K (degrees Kelvin) la given by

X - t + 273

By Introducing thia last relation in cqa.2.7 and 2.8 It rcaults that:

PT-^rI (2.9)

<2.10)

2.23. Tha aenaral gas law Conaider an idaal gas which at a given instant la spacifiad

(Tig.2.6) by the thermodynamic quantitlee tQ, V . TQ . The change of these Initial quantities, to a sat of final coordinates P,, V,, T< can ba interpreted as first expending tha gas at constant pressure to coordinates P^, V^, X. and than expanding it at constant

teaperatute to Pj, V2, X£ (Fig.2.8), la the iecbaric expansion AB , Cberle'a law Is valid, where P, - P , while in the isothermal expanaion DC » Soyla'a lav can ba used, where T. - T, . By combining these two lava it results X, T^-V + (2.11) "o T~~ 'ol

iP--.Vi.T-!)

Fig.2.8. Isobaric expansion (AB) , followed by an laothenukl expansion (BC),

(2.12)

(2.13) which allows that for a given gas FV/T la a constant value. It was found that by relating FV/T to th« concept of *ole_, Ita value la the aaae for any perfect gaa. A »olc la the weight In Era— aqual numerically to the molecular weight off a aufaatance. One mole of oxygen (0,) ia 32 g, one mole of H, la 2.016 g, thua one mole of

H20 !• 18.016 g. Avogadro (1811) demonstrated that at the aaafe temperature and preaeure the maaa of a standard volume of gaa la proportional to lta •oleculer weight. Experiments have shown that under standard condition* of temperature (0*C or 273.16*K) and pressure (normal atmospheric pleasure defined as 760 Torr), one mole (or gram molecular weight) of soy fas occupies a voluac of 22415 ca (22,4 Hear). Values of

•olccular weights of same gascti ace listed in Table 2.3.

For one •ole of gas, Che expression

PV (2.14) was written, where R fit an universal constant

tulile J. 3 JloLKCii_m UiauilTa c

(;•» PunuuLi ^lolttculw weight, g/malB

He 4.003 Sfau Stt 20.18 An;«i Ar 39.044 Kr V-nun X.- 131.30

11, 2.01« ss O Teen 32.000 °i OiJurintf 70.t>l 1*1, A r 28.98 (moan)

J|y.ta>Ki-n chloride,.. H4I 36.4? H,S 34.fW 94.08 NO* 30.01 N»0 44.03 XH, 17.011

CVrboa fnanaaiili'— . CO 28.01 CO, 44.01

(H4 IB.Ov »'fH, 98.04 Wliytow Ijll, SH.0S

* Houreo: Hat\ilUink of iHirminirn and i'Uimim (Chemical Rubber PublMiiBg ,>., (%vf1«id. IMS), 44rli »l.

Th« nuBcricsl value of R depends upon the units nf pressure, voluae and temperature unci. IT the pressure is measured In Torr, the volww lo liters and the temperature In degrees Kelvin, than undsr standard conditions where P - 760 Torr, V - 22.415 litara and T - 273.1CX the value of T! Is:

The syaliol R refers to 11.V. KngnaulC (1610-1878) successor

of Gay-Lusaac. o T 273.16

By expressing pressure and volume In C.G.S. units, 1 ata - 1.0133 x 106 dynes/ca2, and

6 4 1.0133 » 10 x 2.24 x 10 fl ,.. ,ft7 .tw. . 8 31 x 10 RQ 273716 " ' * ergs/*K.«ole.

since 1 cal - 4.186 x 10 ergs (see Appendix), R % 2 cal/*K.Bole.

Table 2.4 lists numerical values of R for various systems of units.

-tuble 2.4 XUMEBICAL VALUES OF Rt GAS COSSTAKT FEB BIOLE FOX Vinous

I' r T *. ilyn^/cui* «m> JK 8.3U x 10' yg./*K IWWtOIw/ni* mJ -K 8,31* joules/*K torr cut3 CK 62,364 torr ea^l*K torr lite pa -K 62.36* ton- liurs/'K UHlt cm* -K 82.03T •tin cm*7'K pdi ft3 'K 1,5*3 Ibft/'R

* In engineering units, 1 lb rooki of gu occupie* 359 ft* *t 32*K •art *troo»- l>horio itruMiiru (14.67 pui). The Ran kino •baoluto temperature ac»M M bw«j upon ilu* Fahn-nhuit ucalu for which abnoluto WJTO temporal uro ia —431.61'F. u ThiwT-u r F i--ica.aojlu,tMT*K-T"C+ m-ie. f .Suiirot-ii: \V. E. Forty 11 m, SmWttoixian PAyjixJ 3*oUu (Smithsonian. IrMti- tution. Wudiinjiion. D.C.. 111.14. fitli rov. od.: T. Buunciner (od.). MvW Mtt/iaiticat EngittUT** Handbook (5leCn»w-Hill Book Company, Now York. 195UJ, Oth rd.

For a fi*i aa*pl« of TMBS VI , O( I g« having a Molecular weight M , the general gas lav is written;

(2.15) 2.2*. Molecular dMritr Awydro (Mil) concluded that equal volusas of all tun under thai saw.* condition* of temperature and urtiiuw contain am"i "'«*"• of nolecolss* Tha number of molecules in on* mole la defined as Avogadro's Msabanc. JL .. By X-ray techniques that accurately determine the interatomic spacing of solid crystals, the mass of the hydrogen atoa la known to be 1.67 x 10~ g. Tha molecular weight of H, (mole of hydrogen) being 2.016 g, it results that:

u 2.016 • 6.023 x 10 molec/mol*. 2 x 1.67 x 10r M The Avogadro number also remits from the precise measurement of the Faraday* F - 96,488 coulomb, defined as the electrical charge necessary to deposit a wile of s substance in electrolysis. Tha charge of sn electron being e • 1.602 :

H. - 7 _ 96488 A twitw

In equation 2.15 W/H denotes the number of moles, thus

« i— (2.16)

1* tht flyi^*** "* »ol.cul.« p«r unit VOIUM. From aquation! 2.15 and 2.16, it rMulti that

\ ? o

thus If P Is expressed In Torr, and S la Torr em /'It t the

nlB f (2.18) At normal pressure (P - 760 Torx) and temperature (T - 273.16't), ion 2.18 glv< Loschnidt number*, Prom the no: rasutls that the mass of a molecule, m (In grans), is

m - £- - 1.66035 x 10~24M (2.19) "A

The distances between molecules can be visualized by using a model In which all the molecules are steady and at same distances to each other. In this case* the distance L (cm) between molecules is given by using eq.2.18, and is

L - n~1/3 * 4.6 K 10"7 ||| (cm) (2.20)

which gives at T = 273°K

10 Torr.

It should he mentioned that these distances are (much) smaller than the mean free path (Bee Chapter 2.4), but are (very) large compared to the molecular diameters (see Chapters 2,4, and 2.5). Equation 2.17 can also be written

^•T - nkT (2.21) KA

She value of the Boltzunn constant is

'•A» »• *PL - 1.3805 x lO"16 «g/'K

* Sometimes the Avogadro number H. , is also rafat*d to as Loschmidt nunbar, since this latter calculated It In 1865. - 44 -

The molecular wight of gas mixtures, it established by using eq.2.15. The partial pressure of th» various guu being P., P^, P..

their una W' IL... H , and their molecular weights «1> M2-.- M , aquation 2.15 bacomaa

(2,22)

If the average molecular weight of the mixture is M , than

EW P.V • — R T - —^ R T (2.23) M ° M °

M - —f- (2.24)

2.25. Equation of atata of real Rases Tha ganaral gas lav (eq.2.14, 2.15) is valid for tha region above the critical point (Fig.2.6) where the ntattar ia In a atata of gaa. as we hava seen In Cfaaptar 2.1, tha P - V curva of real gases* ahows a flat plateau, corresponding to the liquid-gas transition. Near tha critical point tha bahsvior of real gaaaa can be described vary satisfactorily by a modified font of eq.2.14, deduced by Van Jar Waals (1M0)J

[p + ^] (V - b) - *oT (2.25)

In this aquation, tha tarsi A/v take* account of tha fact that the attractive f orcea between molecules will bring them closer together and will thus hava tha same effect ea an additional pressure governed by the constant 4 . This "pressure" snist be the stronger, the closer the 2 molecules axe together, hence A is devidad by V The correction b reduces the total volume, fa representing that part of it which is occupied by the molecules themselves. The volume which is excluded, wu» established to be for each molecule four time Lhat of the molecule itself, thus

b - 4.N,. ~ (2.26)

is the molecular dianeter.

Fig.2.9. - Isotherms corresponding to Van der Weals'equation of:»tate.

The plot of eq.2.25 appears in Fig.2,9. Exclusiv of the region inside the dashed curves (region of liquid-vapour equilibrium), Fig.2.9 agrees with Fig.2.1 (experimental data); point P correspond* to the critical point on Fig,2.1. The dashed portions of the curves which show the pressure and volume both decreasing simultaneously, such as RS, are physically untenable. However in the region where Vandex tfaals equation fails to agree with experimental results, the plataau can bs inserted so that the areas of the two lobe* I and tl - 46 -

are aqual (Hg.2.9). With this understanding, Van der Waala' equation can be need u • fair approximation of the behavior of real gatee. attempt* have bttn Bad* to explain the portion* ST and OR of th* curves, by assarting that they rater to tit* states of supercooled |uu cod superheated liquids. Th* values A and b in «q.2,25 van determined by writing that at the critical point th* thr«* rooti where the curve cuta the horizontal (Fig.2.9) are equal. Eqiutlon 2.25 can be written

V3- (*+-§-) V2+|v -^-o (2,27) whila th« cubic equation with three roots at the critical voluaa

c '

(V - V )3 - V3 - 3V V*+3V2V~VZ-Q (2.28) C C C C

Comparing the coefficients of eq.2.28 with those of eq.2.27t It raaulta that th* constants are

2 h - -|^£ and A - 2? b ?c (2.29) c

The values of the constants for various gates are Us tad in'tkBto%2J5.

2.3. Motion of molecules in rarefied Bases

2.31. Kinetic anrxfty of molecules Iha kinetic theory of gases rests anon the fundamental aaauuptioaa that the matter is made up of molecules, and that the mdleculee of a gas are in constant notion, Intimately related to the temperature of the gaa. During their motion the molecules suffer colllsiona between tbeswalvee, and alao impinge on the valla of the confining venial. Table 2.5 CHITICAI. CONSTANTS. VAN UEH WAALS' CONSTANTS. MOLECUIAA DIAHETEU, AND MEAN FHEH PATHS COMFUTEU mux THE CONSTANT t

A* Urn (E3n3/inute)' P" Itorr (lu FoimuU 1'! *c em*/mol) T=. O'C X 10-»

Ha -J07.9 0.03413 S3.M ZJi Kc -J3B.7 iS.9 0.2107 17.01 HI Ar -\Si. 4B.G 1.3IS 33.11 X.M Kr -6S.0 04.0 £318 39.71 3LM X. 18.1 4.194 MM 3.13 es.a H, -MM 0.S4S Mil *.»« N, - 147.1 !*.• 1.310 SMS 3.14 O, — 11S.1 33.5 1.3(0 31.13 3.93 C), 144.0 49.7 0.493 W.M 111 H« >I60D 70.1 UL093 iro S.3S >*oo HyUrojjoneJilorids. • HCI 3.0*7 40.11 9.11 S.494 HtO 374.0 217.7J 30.49 2.19 7.91 Hydrogun tulfido... H^S 100.4 U8.I 4.411 4S.B1 3.*4 0.07 SO, 1W.S 77.7 0.714 60.3B 3.5* 8.03 NO -W.0 65.0 1.340 37.89 3.31 a,w 3S.S 71.7 P.7I2 44.1 B N,U 3.27 SJ4J NH, 132.4 lll.S 4.170 37.07 3.W C«rlxjii manovid* .. CO -139.0 35.0 1.483 sto 3d« 1.J8 RMIKHI diuiidn .... CO, * 31.1 73.0 3.593 43.87 3.2S (.13 CK. >16S0 >aoo S.263 42.76 3.24 SI.4 3.44 C.M, 36.0 02.0 4.390 £7.14 3.56 e,ti. 9.1 HJ.9 4.471 73.9 3.04 4.11 CAAKUI duulBilo ... cs. 271.0 7B.0 11.01 • Source*: Amtrimn Innitvto of Fkyka Hindboat iMcGrw-Hill Book Company, N«w York, 1W3), Jndcd.i Hcnulboo* of I'hysic* ant Chtmutry {Ctionucal Rubber Publiil •ng Co„ ClmUnd, JWJ3J, MUted.

The noaentun transfer fron the molecules to the vails of the vessel results In the pressure P , which appears In previous chapters, thus the pressure can be related to the kinetic energy of the molecules. Consider the collision of just one particle (molecule) of mass *, traveling with the velocity v in the x direction of a box of

length L (in the x direction)t with a vails of area A perpendicular to the x .direction. The tine between successive collisions with the wall A is it - 2L/v . The change of momentum A(«v) of the particle in each collision is

A(nv) - nv - ni<~v ) - imv Kewton's second law defines the force F at the rate of change of momentum with reapect to time, thui 2 , ._... 2mv_ rav '-^-irfr-T- <"°>

The average pressure due to this particle Is 2 2 r*-i--nr'-$- <2-31> where V Is the volume of the box. The preiaure due to n molecules ie P - new2 (2.32) 2 2 where v Is the average value of v for n molecules. Similarly Py - nmv » and F - nmv . Since the motion la random there is no difference in the average motion in the various directions, so that ~ —S T Vx - V y - Vz According to the Flthagorean theorem

2 2 2 2 2 i?-v7+vx y -vz -3vx -3vy -3vz thun the pressure measured in any direction is

P - Px - Py - Pz - ^2_ (2,33)

Comparing the expression of pressure given In eq.2.33 with that given by eq.2.21, it results T P - ^- - ntT (2.34J

The average kinetic energy of a molecules being 2 mv 2 one concludes that

3kT (2.35) 2

i.e. the average kinetic energy of the molecules is the sane for all aeses, and ia proportional to the absolute temperature.

2.32. Molecular velocities The constant occurence of collisions produces a wide distribution of velocities. IF a collision of two aolecules with velocities v. and v. , the total kinetic energy is preserved, thus the quantity

m(v_ + v2 J

is the sane before and after the collision, even if v. and v* oust change. Maxwell and Boltzmann expressed the distribution of the velocities fay the law

f vft (2,36) n dV* v--J^T72 l2kfj where f is the fractional nunber of molecules in the velocity rang* between v and v + dv , per unit of velocity range. The value of f la zero for v » 0 , and v • ™ > and has V its iiaxiaum at «. value 1/2 ffl (2.37) OJ u 0.7 i >\ 0i !l\ 03 i • \ t. 04 OJ / & 02 / *" 01 *ii.i,.* . \>>-. . 0 02 M OS M 10 12 I* !£ 18 20 22 24 26 23 30

Fig.2.10. - Kaxuell-BoltzHann •olecular velocity distribution curve. given by differentiating f with respect to v and setting the result equal to zero:

3/2 3 2 dfv t> Ls_| |,„ S- ,. L"«» «kT . n ^m IzkfJ |2v ET v |e -°

Figure 2.10 shows equation 2.36, plotted versus v/v . The V have this velocity then any other value of the: velocity. The v value, is different from the arithaetlc averaRe value v , which results Cram

/ v«v°v T 1/Z v o 2 f2kT] , 128 v (2.38)

The ncan square velocity v is obtained from

v2U2—- 3^ (2.39) m n / Vv and it th« same as obtained in eq.2.34. The root-mean-aquare velocity is therefore

vr " M " I3 IT] " 1-225 vp <2-40)

Which of these velocities is of interest as representing the average behavior of a gas depnds upon the process under consideration. When the molecules directly influence the process by their velocity, (e.g. flov of gases), the arithmetic average is used, while when the kinetic energy of the molecule influences the process the root-mean- square should be used. Based on eqs.2.19 and 2.21

. R -'—• (2.41)

1 2 1.45 x 104 hi£ ' cm/s (2.42) it results that the average air molecule (M - 29) at T m 300 *K , A has a velocity of about 4.6 x 10 cm/sec.

2.33. Molecular incidence rate In a similar way Co eq.2.36 the distribution function fv of the velocities of molecules in the x direction was written as:

The number • of molecules striking an element of surface (perpendicular to the x direction), per unit time is given by

[ vx dnx (2.44) iy lMrodaclng dax fra .q.2.43 into 2.44 and integrating. It mult, that:

1/2 2 (2ttl •»l«c/eW.a (2.45) tmd by using aq.2.38, 2.42 and 2.17 it also results

• - i nv - 3.513 x 1022 £975- »l«e/eaZ.i (2.46) where ? is la Torr. Table 1.1 lists SOBS values of • . If a hoi* of ar«a A is cut in the thin wall of th« vassal beyoend which the gas density is cero, the rate at which Molecules of gas leave the vessel is

1 3 (T\UZ q * +A - j m A - 3.64 x 1ST jjj nA molec/sec (2.47)

The volae* of aaa at the pressure in the vessel escaping each second 1* obtained by dividing the flow q by the density n. thus

f-J. 3.64, 10' (I] Ac3/. (2.48)

•bleb for ill at 20*C would b.

dt (2.49) 2 «b.x* k If In a' Th. «aaa H of gM aacaplng, can ba found by cofd>lnlng aqa.2.46 I 2.19, tfcua

1/2 V - 5.83 x 1O-'P(M0 * P f ) g/a.c. (2.50) - 53 -

2.4. Pressure and MM free path

2.41. item free path During their notion the molecules suffer collisions between themselves. The distance traversed by a nolacule between successive collisions, is its free path. Since, the magnitude of this distance is * function of the velocities of the molecules, the conception of mean free path X is used. This is defined as the average distance traversed by all the molecules between successive collision to each other, or as the average of the distances traversed between successive collisions by the same molecule, in a Riven time.

A molecule having a diameter £ and a veLocity v moves a distance v6fc in the time fit . The molecule suffers a. collision with another molecule if anywhere its center is within the distance £ of the center of another molecule, therefore sweeps out without collision a cylinder of diameter 2£ • This cylinder has a voluae. 2 2 6V » *< » v6t (2.51)

3 Since there are n molecules/cm , the volume associated with one molecule is on the average 1/n cm . Wien the volume 6V is equal to 1/n , it must contain on the average one other molecule, thus a collision has occured. If T • St is the average time between collisions,

2 i - TF5 VT (2.52) n and the mean free path X is

X-vr -—^-j (2.53)

If we consider the more realistic case, in which not only the reference molecule is in motion but also the others, than eq.2.53 •c be mitten

(2.54) TO C *

•here t litti ebaolute velocity, While vf is the relative velocity

of tbs Molecule*. Cleuaiue eatabliahed thet v/v_ - 3/4 t thua:

(2.55)

Finally if the Maxwell-BoltEnann diatributiou of velocitlea ie alao considered, it reault thet

kT x i - _ . (2.56)

V2 wn K V2 IT € P

by using eq.2.21.

This glvee the relation

X - 2.33 x 10"20 -£- (e>) (2.57) 5 P where T ia in *K , & In cm ^j p £„ Torr. For air, at ambient temperature th* elmple fottuli

C2.58)

can be ueed, with P-Torr and X-cm. It can ba sen that at

f - 10 Torr X - 50 m f thus much larger than th« dlmeneicns of a vacuum encloaure, thua at auch preaaurea tha aoleculea collide only ritti the walla of tha vaaaal.

Figure 1.1 show* tha valuaa of X for air, whila Tabla 1.2 Hat aoao valueo for othar gasea. The value* of g are Hated In Table 2.5. - 55 -

Equation 2.53 shows no influence of the temperature on the ••an free path; it was established considering pure mechanical (elastic) collisions between molecules. Sutherland (1893) established experimentally that at constant

n p the mean free path is influenced by the temperature, and the dependence was described by the relation

* - ——h r (2-59) v2 TO r a + p where the constant c (Sutherland's constant), is a measure of the strength of the attractive forces between the molecules. From this equation It was deduced that

XT - -^ (2.60)

l+f

where X„ is the mean free path at the temperature T , \m is the

mean free path at very high temperatures (T - °°) t and c %% a constant, (see Table 2.6)

Table 2.6. - Values of Sutherland constant

G*» H H He A K 2 2 °2 Be r "2°

3 X„ x 10 at 1 Torr(cm) 10.56 6.1 6.87 16 11.2 7 5.96 9.5

C(*K> 76 112 132 79 56 169 142 600

Similarly to eq.2.56, th« mmmn fr.e path X _ of a «u artxtura of two gi.es 1 and 2, was written *<«1+ V V1 + !T>

where 6- , 5, are the aolecular diaaeters of Che gases 1 and 2 respectively, H. and M, their aolecular weight*, and

the partial pressure of gas 2, in ths Mixture. It can be seen that for L • L , and H_ - K* •this aquation laeds to

(2.62) 2 — 2 * V2 * 5* V2

which is identical with eq.2.56.

By introducing the value given by eq. 2,62, in eq.2.61, and-the

fact that ^t^ « C^/^)1'2 , it results thst the man free path of a gaa Mixture is given by ^"f^i/^lM,^]Tfhl 1'2 If the partial pressures of the two component* are P « P, » the nolecules of gas 1 will have much more collisions with thoaa of gas 2, than with those of gas 1. Thus in this esse X % X. _ . By -3 1 » adding to air (M, - 26.7 ; A • 5 x 10 cm) a very asall quantity

-3 of Ha CMX • 4 ; Aj - 14.5 x ID ) it results that

XHe * h 2 " z"^ ,- X2 * Z'1 *alr " X'2 1.59Z *TlS 2 air

Therefore the Mac free path of haiiisn •oLeculea la twice chat of the other Molecules, thus they diffusa fast In the mixture; leak detection - 57 -

takes advantage of these phenomena.

The mean free path of electrons X (very snail mass and diameter), according to eq.2.63 Is

X - 4/2 X (2.64) e

while for the mean free path of Ions X, the relationship

yyil (2.65)

was established.

2.42, Pressure units Pressure la the most widely quoted parameter in vacuum technology, and this brought to the use of a. larga number of pressure units, which are used in various texts.

Pressure in a gas, defined In terms of gas impingement on a surface (see Chapter 2.31), is the time rate of change of the normal component of momentum of the impinging gas molecules per unit area s£ surface. Thus, the pressure exerted by a gas on a real surface is defined as the force applied per unit area. The various pressure units belonging to the coherent unit system are based on this definition.

As pressu can be measured by the height of liquid colums, the various non-coherent units of the pressure are related to these columns.

In a coherent unit system., the unit of pressure [p] is expressed as

'ipj - aj"1 M ur2

if the unite of length [t]t mass [m] and time [t] are considered - 58 -

** th* basic ualta (CGS , IS , HTS )p or a*

[p] - tP) UfZ

if th* units of fore* [F] and length [£] are considered as basic units (technical systems). 2 ID tha CGS system the dyn/cm is tbe pressor* unit (Table 2.7). This unit is called microbar. Tb* microbar is also called '*barya" (in th* French litaratuxa). Tbe nime of "vac" wa* proposed for tha millibar. 2 Tha MES system uses the Newton par aquara meter (H/m ),- which Is called Pascal (Pa) in tbe French literature. The Gaede. Gd - 10"3 W.

lb* British system uses the pound par square inch (pslt or Ib/lm ), while the MIS (meter,ton, aacond) eystem has a pressure talc callad "Bless" dps - 103Fa). 2 Tbe technical atmosphere (at) is the name given to th* km (force)/cm .

Thee* various units are summarized in Table 2,7, their conversion to each other and to the various non-coherent units ie given in Table 2.8.

Th* non-coherent preaaura unite used are; tbe physical atmc*ph*r«, tb* millimeter and micron of mercury and the Torr, th* millimeter and centimeter of water, and the inch of mercury.

Tha physical or normal atmosphere (atm) was defined as th* pressure exerted by a mercury coluan of 760 mm when th* specific gravity of mercury is 13.595 g/cm (at 0*C).

1 atm - 76 cm x 13.595 g/cm3 x 980.665 cm/s2 -

- 1.013 x 106 dyn/cm2 - 1.013 x 105 K/cm2 - 59 -

Table 2.7. - Coherent: pressure units

Unit of Unit of Onit of Sy.cn area force pressure

2 CGS en dyne dyne/ca

1 dyn • 1 ber - 106 dyn/cm2 - 2 1 g.cm/s - 10"1 N/a2

2 IS(MKS) m Newton (N) N/a2

IN - 1 kg.m/s2 1 PasceKPo) - 1H/M2

2 2 Technical a Kgf Kgf/m

1 Kgf - 1 at - lKgf/ca • 9.81 N - 9.81 x 10* N

British 1ft2 1 lb 4 lb/ft2 Z 47.88 N/»2

lin2 % 4.448 N lb/in2 % 6894.7 H/a2

2 MIS in Sthene (sn) pieze (pz) 2 1 an« lc.m/s • 1 pz - 103N/m2

» 103N

The Torr (Torrlcelll) is defined aB the 760 part of the normal ataosphere, thus

1 Torr - 1.333 I 103dyn/cn2 - 133.32 H/a

Practically 1 Torr • 1 m ig, theoretically 1 BB Hg - 1.00000014 Torr. Tsblt 2*8 CONVERSION FACTORS (N) FOR Pustuitt UNm* (i.x - n.Y)

**•* i™tsH cm of water- tan in. of Hg j M/fai* (pi] • «i (Wis*) I

7JX10-'; 1x10- "H>-»! lxl0-»;i.0lxlO-» MxIQ-ilMSxIO-tjMSxIO1 1.01x11' f.lxl|-t

U5xlO-»|USxI»-*|l.»xlO-» 1K10-»J3JJXW-IJI*3«]0' USXIO- 1.33x10- Dixit-. !" lxlO-» 1,01 '2.«"IO"*ll.45 1x10- | 9.1. It- •

arkrm* 9.1x10- 1/10-1 T.JKlO-*;2.M>10-ijM2xlO-' ».»XlO- j 9.£> 10-1

I.OI 7.S-< IO-i i J.*S *• 10-1 j |.« x |0-< 1x10" I 9.lv 10-•

(UucrfcU) 910 730 7.3>:IO-i'2.»'flO-i] 1,42x16-'> txlO"11 ».lx!0-< i 9.6 10-

.orr 1.33 \ I0> 1x10* 1 lj.MxI0-l' 19JslO-*Ii.3S"IO-'; 1.13x10-: : i-»i 10 It-fHl 3.3 x 10" 2,94x10* 15.4 i 1 t 4.9*10-'! 3.4xl0"| 3.3x10-1 ! 3.3x10- (6 (in-(p.M.) 6.1 y 10* 5.17x10* i *• 10- Tcctuuaiaa- tpkti*

lir IXlO* TJXlO* 1x10* i 1,01X101 ; Ixltf" jl.DlxKP J Mull

»ph*r*[*in) IMl XIV i 7.*rl0« : 1JJI>I0« 1.03 xiC ;i.OlxlO» ; 1.03no*

* I Gd (Giede) - ID"1 Newtoa/m*. I Pascal •• 1 Newton/m* — lOmicrofear. 1 pz (pkzc) - 1000 NewtonM. The a Torr (•llli Toxr) is equal to the micron of Hg (u). The Inch of mercury (in Hg) - 3.386 x 103N/«2. 2 The xm H-0 % 1 kgf /• ; the ca of water was called Guericke (Ger).

2.5. Transport phenomena in viscous state

2.51. Viacoaity of a gas A gas streaming through a narrow-bore tube experiences a resistance to flow, so that the velocity in the direction of the flow decreases uniformly (parabolic distribution) from the axis until It reaches zero on the walla. In the same way the gas between two plates (Fig. 2,11) one at rest and the other pulled in the plane, has a drift velocity zero at the contact with the steady plate, and a —•gluw'H velocity at the contact with the noving plate. Each layer of gas paralled to the direction of flow exerts a tangential force on the adjacent layer, tending to decrease the velocity of the faster moving and to Increase that of the slower-moving layers. The property of the fluid by virtue of which it exhibits this phenomenon Is known as internal viscosity. Newton assumed that the internal viscous forces are directly proportional to the velocity gradient in Che fluid.

W*-/ y / x^^^V^V^VAVV-A^WAVO.i' W

Fig.2.11. - Drift velocity distribution due to Internal viscosity. - 62 -

Considering Che g.« between two parallel plate* (Fig.2.11) separated by a distance y , the upper plate being in Motion with a Telocity u- The gas will be steady at the level of the lower plate, whereas Its drift velocity will be u at the contact with the upper plate. The drift velocity of the gas u' at seme intermediate level y* will be :

u'-u^ (2.66) y

The coefficient of viscosity n is defined as the tangential force per unit area for unit rate of decrease of velocity with distance. Imagining the gas devided into layers parallel to the surface, each havimg a depth A , the mean free -path (layers in which the Molecule has no collisions), the tangential force between adjacent layers of area A is written

F - nA ^ C2.67) y where n is the coefficient of viscosity.

According to the kinetic theory, the tangent.4al force per unit area is measured by the rate at which momentum is transferred between adjacent layers. Molecules from a distance \ above move down into the layer u* with a momentum

+ y while thoie froM a distance A below move up with a nomentum

The number of molecules that cross unit area per unit time in any direction in a gae at rest is equal to -r nv . Hence the net rate of transference of momentum a cross the area A is equal to

P - £ nvav [+ - (mu')_] - | nvav miu/y (2.68) Fron eqs.2.67 and 2.6$, it results

where p - ran is the density of the gas. This equation is approximate only. When the distribution in randon velocities and the distribution in free paths are taken into account the result of the calculation (for rigid, elastic spherical molecules} gives.

n - 0.499 new \ (2.*0) av By using eq.2.56, and 2.38, It results that

0.«9 • v™ n OQQ L.M.ll/2 av 0.998 (2.71) ^2 2 m V2 IT C IT C From eqs.2.67, 2.70, 2.71 it can be seen that the dimeasloms of the coefficient of viscosity T\ are [M] [L] [T]" . In the CGS system the unit of viscosity is -1 -1 -2 1 poise « 1 g.cm 8 • 1 dyne s.ca Table 2.9 list values of n for various gases. Since ri is proportional to \ , the Sutherland equations (2.59), 2.60 also apply to the viscosity. Thus:

2 1/,£2 (0.998/* O(nkT/T0 {0 „. n 1 + c/T (2'72) where c is Sutherland's constant. According to this equation the viscosity of ftascs increase*

with temperaturet whereas in the case of liquids the viscosity is known to decrease as the temperature is increased. - 64 -

VlfOMRT Of OUSS AT 0*C AHD 7*0 TOM KHUCTm WITH COMTUIXD VlXUX* or llOLECvuji Duunu Ann MKAK PXEE PATHS

{ cm x!0-«

H. tM K. 313.4 Mi Ja 3.W Kr M4.3 4.S7 X. I1M 4.S7

H, H.7 J.»S 1H.I 3.78 o, lil.O 3,46 CI, 114.0 5.S1 171.J 3.7S

DM 4.53 B.S 117.5 4.73 SO, • 17 S.M NO 17».0 3.71 N.O 134.1 4.M MH, 8».t 4.47

00 1U.S 3.7» CO, U7.t 4.M 10U 4.1* «A UJ 4.H •>.f C.H, CM * jBouro*] BmnJkttk «/ Ohtmttlry and Pkytic* {Chemical Rubber PubHthini C*., Cbvriurf, MM), 44lh Hi. Thus predictions are valid in a given range of pressures. Ac both, vary high and very low pressures, the viscosity of a gas departs from this prediction. At very high pressures the average distance betveen the molecules is so saall that the latermolecular forces become Important and the momentum transfer is very different from that assumed here. At very low pressures, when the mean free path exceeds the distance between the wall*, collisions between Molecules almost do not occur. In this case the transfer of momentum la only between gas molecules and walls.

The mean free path determines the behavior of the gas, and whether the gas exhibits the property of viscous or molecular flow.

The theory of viscous and molecular flow will be treated in detail in Chapter 3.

2.52. Diffusion of gases Experience has shown that two gases placed in ..he same vessel, diffuse into each other until the relative concentrations are the sane everywhere in the vessel.

Meyer has established that the coefficient of interdiffusion of two gases is given by

X vavi n2 + X2 vav2 ^

D 2 73) ' 3(Pl + n2> < *

The coefficient of diffusion D Is expressed in [L] [T]-1 units. It is defined by

1- - (2.74) if the diffusion of molecules in the same gas (self diffusion) is considered.

By combining eqs.2.74 and 2.69 it results that

(2.75) nu - n/p la fact by Introducing various distribution factors* it was determined that

(2.76)

If concentration of one of the gases is vary small (traces) ,- n_ « n_ , aq.2.73 becomes

^ * vav 2 fk3 T311/2 2 2 m (2.77) 2. Table 2.10 gives the diffusion coefficients D 2 (cm /s) observed for several pairs of gases at 0"C and 1 atmosphere.

Table 2.10 Cveffitinu of Intcrdinrnstoa an* A>emge Muscular Dam 10">„ 10%, (u c from (cafc from

Cases Du (obs) i>„) n)

H-iir 0.661 3.23 3.23 HrO, 0.679 3.18 3.17 Or«ir 0.1775 3.69 36S

OrN, 0.174 3.74 3.70 ro-Hj 0.642 3.28 325 coo. 0.183 365 3.70

COrH, 0.533 3.S6 369 COj-aw 0.138 4.03 4.20 COrCO 0.136 4.09 4.22 N,OH, 0.535 3.57 3.69

N1fMT01 0.0953 4 53 4.66

The diffusion process found its application in the diffusion pumps. which are the most extensively used nystera for achieving high vacuum. In 1915 Caede published a description, of a high vacuum pump, which involves no mechanical motIOQ but depends for its operation on diffusion of residual gases through a slic or fine opening into a high velocity stream of mercury vapour traveling in front of the opening.

Gaede's apparatus (Fig.2.12) consists of a streaa of aercury vapour AB passing in front of the opening C of a tube connected to the volume E to be evacuated. At R the side tube is cooled by water with the result that any mercury vapour passing into the tube is condensed at D . The residual pressure of the vapour at I) is thus reduced to less than 10 Torr. The vapour stream ic the vertical tube entrains any molecules of gas that get into the stream, and consequently there Is a constant diffusion of gas from E towards C.

R Cn 0 r 0

Fig.2.12. - Principle of Gaede's diffusion pimp.

If u Is the velocity of the mercury vapour in the direction CD , and n the concentration of gas molecules at any point x along the length I (where x - 0 at D, and x *• -t at C), than* - 68 -

tlw rat* at wtilch gu paasea froa D to C ia givaa by

(P is the diffusion coefficient of the gas in mercury vapour), end the rate at which gee molecules are returned from C to D * is

In the stationary state these rates must be equal, thus

D^+nu-D^+Pu-0 (2.78) since XL is proportional to the pressure P . From 2.78 it results

dP u „ (2.79) and by integrating over £ ,

jr--f£--r (2.80) o where P and P_ denote the pressures at C and D , while D is the coefficient of diffusion. This equation shows that the gas flows from point D towards point C , since u , I and D are all positive.

The equations derived based on eq.2.80 for the design of diffusion pumps will be discussed in the chapter devoted to pumping.

2.6. Transport phenomena In molecular ststes 2.61. The viscous and molecular states

At low pressures, when the mean free path of the molecules of the gee becomes large compared with the dimensions of the enclosure, the energy transport from wall to wall does not include the collisions - 69 -

between molecules, thus it is not more a function of the viscosity.

In a vessel of volume V the number X of intermoleculsr collisions per unit time, is given according to eq.2.56 by

X - nV —^ - ^2 TT tZ n2 v V (2.81) A av

If the vessel has an internal surface A , the number" of molacules striking the vails, is given (eq.2.46) by

N - A* «invav A (2.82)

The ratio between the number of intermolecular collisions X and that of the collisions molecule-wall, is

| - 4 n i, n 5* I (2.83)

This ratio will show the limit between viscous state and Molecular state,and according to eq.2.83, this ratio is a function of n , thus of the pressure, of £ '(nature of the gas), and of the dimensions of the vessel V/A .

Considering the model of a cylindrical vessel having a diameter

D 9 and a length L large compared to D , the ratio V/A will be

2 V_ m n D L _D A * 4it D t - 4 thus

f - *f ir n 52 D (2.84) or for air

| - 6.2 x 10"19 n D , in IS units (see Table 2.7)

In table 2.11 it can be seen that at atmospheric pressure the number of molecule-molecule collisions is 15 million times that of - 70 -

molecule-vall collisions. The pressure mamt drop to 5 x 10 in order that their number be equal.

Table 2.11. - X/N as a function of F for D - 1 nt.

A X r ° 3 State (Tort) mol/a N

25 760 6.1 x 10"8 2.46 x 10 1.53 x 107

10 4.6 i 10"6 3.3 x 1023 2.05 x 105 Vifcoua'

ID'1 4.6 x 10-4 3.3 x 1021 2.05 x 103'

3 19 io- 4.6 i 10-2 3.3 x 10 20.5

io-* 4.6 3.3 x 1017 2.05 x 10"1

io-7 460 3.3 x 1015 2.05 x 10-3 Molecular

9 io- 4.6 x 104 3.3 x 1013 2.05 x M-5

The viscous state does not changes suddenly to molecular, between thee a intermediate state exists. Thia will-be analysed in Chapter 3, }A connection with the flow equations.

2.62. Molecular dray la the viscous state, all the collision* were assumed to be perfectly elastic, thus the molecules striking a surface would be reflected as elastic balls. At lev pressures, this image does not cover the experimental results. Experiments show phenomena, which can be explained by the Image that the molecule "condenses" on the surface, rests on it a given time, and then it is "r•evaporated" in a direction which is independent on that of incidence.

We aasume a surface which Is "free" of any adsorbed layer of molecules. In presence o£ a gas, tne molecules of the gas will "condense*" on the surface, and will "rest" on the surface a given time, before being reevaporated. The number of molecules striking the unit surface being (eq.-2.46)

% • -r n v and that necessary to form a monolayer being K - i/r the tine required to form this layer will be

1/2 (2.85) 2 n 1" C n \Vi jnJ in order to form the monolayer the*, molecules must rest on the surface at least this time, from eq.2.85, for nitrogen, at 20*C, we find that -6 1.72 x 10 (2.86)

where t Is in seconds, and F is in Torr. Thus for P * 10 Torr, -2 TN2 - 1.7 x 10 seconds a time which is sufficient to transfer energy to the molecule. If the surface is in motion it can transfer a velocity component to the molecule. This is the principle on which the molecular pump_s are designed. In such pumps the gas Is pumped by a groove (Fig.2.13) having a depth h , and a v.idth y , The groove is at rest, while the cover (bottom) is moving at a speed v » in the direction x (positif direction). In this case,, for steady state, (zero flow), the pressure will increase In the v direction. The result of the forces applied to a volume comprised between x , end x + dx . will be

df - — dx h (2.67) dx o

Fig.2.13. - Principle of molecular pump Tha dumber of molecules striking the unit surface being nv /4 (cq.2.46), the surface being ydx, and the momentum received by the "reevaporated" aolecule being (maximum) mv, the force applied on the gas la

dq • ~ n v .mv ydx (2.88)

From df » dq, we obtain:

4^h -rnmv .v (2.89) dx o 4 av from which by using eqs.2.21, and 2.38, it results

I1/2 .. !» f " ] (2.90) P [2* R TJ h_

and by integrating

4P p -<-!--yw]P« [Zir R Tl h^ (2.»D

which shows that the pressure ratio P/P , which can be achieved by molecular drag (zero flow), Is an exponential function of the length of the path x , and of the relative velocity v of the moving surfaces, and inverse to the distance h between the surfaces. Equation 2.91 also shows that the ratio P/P is greater whan K is greater (heavy gases).

Molecular gauges were constructed using the principle of molecular drag. These gauges use either the nathod of the "decrement" or that of the "torque". In the decrement type of gauges, a surface la set in oscillation and the rate of decrease of the amplitude of oscillation is taken as a measure of pressure. Physically, the damping nay be explained as due to the gradual equalization of energy between the moving surface aw* the molecules of gas striking it. - 73 -

In the torque type, a surface is set in continuous rotation, and the amount of twiBt imparted to an adjacent surface is used to measure tbe pressure. The molecules striking the moving surface acquire a momentum in the direction of notion which they tend in turn to iatpart to the other surface. If that surface is suspended on a filament, the filament will have a torsion.

2.7. Thermal diffusion and energy transport 2.71. Thermal transpiration The rate at which molecules leave a chamber through an opening in a thin wall, was shown (eq.2.47) as being

thus the mass of gas leaving the chamber (rate of efflux) is given by

W-mq-inmv A (2.92) and since na • p (specific gravity),

. fR T il/2

If we have two chambers A and B , separated by a porous plug, and the gas in the chambers Is at different temperatures T. and T_ , thermal transpiration will occur until an equilibrium state is established at which

and since p is proportional to P , and invera proportional to I (eq.2.18), it results

P. f-T.)l/2 -\ (2.95) Tills Is of Importance In vacuum system where low temperatures arc ussd (traps, cryogenic pumping), THUS if s chamber A lit part

of a system at liquid air temperature (IA - 90*K), and the pressure

Is aeasurad by swans of a gauge at room temperature (TR • 300 *K>| then the real value of

90 P. - 0.55 P_ (2.96)

When the two chambers are connected with a large bore tube or the pressure is higher, so that the mean free path is much smaller than the diameter, and collisions between molecules become predominant, the condition of equilibrium is P. - P_ (instead of eq.2,94), thus

PB TA

Details of flow at viscous conditions willbe discussed in chapter 3.

2.72. Thermal diffusion If a temperature gradient is applied to a mixture of two gasss of uniform concentration there la a tendency for the heavier and larger molecules (mass m. , diameter £-), to move to the cold aids, and for

the lighter and smaller molecules (m, t £,) to move to the hot side. The separating effect of thermal diffusion (coefficient B-) is ultimately balanced by the mixing effect of ordinary diffusion

(coefficient D12), so that finally a steady state Is reached and a concentration gradient is associated with the temperature gradient.

The coefficient of thermal separation is dsfined by

BT

It was established that if k_ is a constant then the amount of separation is given by af - ^ an ^- (2.99)

The practical device utilised for tbe separation of mixtures of different gases and of isotopes uses a long vertical tube with a hot wire along the axis. Because of thermal diffusion, the relative concentration of the heavier molecules is greater at the cold wall- Convection causes the gas at the hot surface to rise to the top, where It is deflected to the cold wall. As a result, the heavier component concentrates at the bottom, and the lighter at the top.

2.73. Heat conductivity of rarefied gases As in the case of viscosity, the process of heas transfer by gases, is different in the case of the viscous state and in that of molecular state. In the first case the totality of molecules is responsible for the heat transfer, while in the second case the individual molecules carry the heat from wall to wall.

Beat conductivity in viscous state

As in the case of viscosity (Fig. 2.11} we can consider layers of thickness A (mean free path), between two places whose temperatures are T. and T_, and distance apart y . The relative temperature drop between the layers is

2(1, - I,) A similarly to eq.2.68.

If c is the specific heat at constant volume, the heat transferred per unit area is

1 Tl " T2 • >;>v .2n c — X 0 av v y

1 Tl " T2 •TM-A.' ^T-* (MOW - 76 -

Therefore tlM beat conductivity X 1* tlvin by

K " T "v av *c v (2-101>

aM coaparlae, ea..2.101 with eq.2.69 It follow! that

K - n c (2.102)

As Is the case of the relation for n (viscosity), a sore careful consideration of the merhani— of energy transfer leads to the relation K - | OY - 5) n c^ (2.103)

C C ; where Y * D/ ** the ratio of the specific heat of the gas at

Table 2.12 - Beat conductivity of gases K 3t 10* at 0*C (cal cn~1s~1degK""2)

H Cat Air ». A K H CO co2 "2 *2 °2 . r t tl 10* 4.19 0.57 0.58 0.58 3.43 1.09 0.39 0.21 0.12 0.53 0.34

constant prassura to that at conatant voluaa. K is axprastad In cal.csT .a" *K~ , if c is axpressad in cal/gram. For •onoatonic gasea (A, Hg, ate) Y • 5/3 , for diatonic gaaaa

(02i H.. K.t ate) T " 7/5 . while for trlatoalc gaeee (a.g. CO-) Y - 4/3 . Valuas of K In labia 2.12, 2.14. Sinca tba viscosity la not a function of tha praaauta (aq.2,71),' it follows (eq.2.103) that tha thanaal conductivity of a taa Is indapandant of prassura. Ihia is valid as long as the pressure is -

* It will ba obsarvad that tha haat conductivity of H, and Ua are such graatar than those of heavier gases. - 77 -

1» higher than the range in which molecular state exists.

Heat conductivity in molecular state. When the gas pressure is so low that the molecular mean free path Is about equal to or greater than the distance between the vails of the containing vessel, the gas is no longer characterized by a viscosity. In that case the equation 2.103 is no longer valid, and the conductivity is then found to depend upon the preBaurc. The process of heat transfer under these conditions is called free Molecular conduction. In order to express the heat conductivity at lav pressures, Rnudsen introduced the concept of the accommodation coefficient.

The accomodation coefficient is defined as the ratio of the energy actually transferred between impinying gas molecules and a surface, and the energy which would be theoretically transferred if the impinging molecules reached complete thermal equilibrium with the surface.

When molecules originally at a temperature T strike a hot surface at temperature T (> T.), complete interchange does not occur at the first collisions, and it may require many collisions for this to occur. The molecules reemitted from the hot surface consequently possess a mean energy which corresponds to a temperature lower than T , which we shall designate as T . The accommodation coefficient a is defined by T - T • • ^r^ O.IM) B 1

If the molecules reach thermal equilibrium with the surface bsfore escaping, T - T , then o » 1 . On the other hand. If tht molecules are elastlcally reflected without undergoing any change in energy, T - T. and a - 0 . Table 2.13 lists some values of a . - 71 -

table 2.13. - V*lmae of tea accoeaedatio* coefficient a

IN-aaxfac* PC H heavily ordinary polished •lightly Cu >^ blackened blackened

0.2 0.36 0.358 0.712 *2 0.556

"» 0.57 0.89 -- -

A 0.85 0.89 '• - -

* 0.55 ' * - -

Air o.Jo - - -

°2 -- 0.835 0.927 0.956

co2 -- 0.868 0.945 0.975

According to eq*2*46 tbe nuifli.tr of eoleculea bavin* • velocity between v and v + dv » and vbicb atrlke an unit aurf ac« in tb* unit tie* i*

* " 4 aV vdv"i *<*»> (2.105) 1 2 Slnca tacb Molecule has a klnatic energy equal to 7 • v * tn* energy traneferred la

dE - ~m v3Wn) (2.106) thus v«» E-i-a J v3dn (2.107) v-o which solved by using the Kaxwel-Botsaann distribution, results ID

E-?"Invav^-}n,,,vav7T (2-108)

This Is the energy tranferred by all the molecules striking the unit surface in the unit time. Since their number Is (eq.2.46)

4 • -r n v 4 av

it follows that the average energy transferred per Molecule ts , n n v v2" E--f-T nT -|»v»-2ia (2.109) av 3 instead of E - — kT (eq.2.35) which is the average energy of the molecules in a volume.

For monoatomlc gases ac low pressures, Che energy transfer froi hot to cold surface, will be according to eqs.2.46 and 2.109:

E E n v 2t(T C2 ll0) o" * m"t l r" V *

and according to eq.2.21 and 2.104, it follows that

E - 7 IT1 C* " O - 5 T^ (2.111)

whare a is the accomodation coefficient (eq.2.104), P is the pressure of the gas, v. is the average velocity at temperature T, and T Is the temperature of the hot aurface. Thus the rate of energy transfsr at low pressures Is proportional to the preasura and the temperature difference.

For diatomic and polyatomic gases, the molecules striking the hot surface acquiTe not only Increased translations! energy but also Increased amounts of both rotational and vibrational energy. The amount of the vibrational energy possessed by molecules as compared with chat of translational an«r$y la aaaaurad by tha valua of Y - (aaa alao •q.2.103). k datallad calculation l«ada in thia caaa to

which for Y " 5/3 (caaa of Bonoatottic gaaaa) bacomaa idaatical with aq.2.111.

Snbatltutlng for v. (eq.2.19, 2.21 and 2.38) aa a function of T. and H , equation 2.112 bacoawa

Eo 2 T^l l2t H(273>J l~j CT. V?

" *o "[l2] (T.-T^P erge/Mc.ca2 (2.113)

;in which A la the faw molecular conductivity at 0*C, given by

o 2(v - 1) l2l H(273)J

"T/T Y~^T ««»/»«-=«2-*c- v»«

t For air (Y • 7/5, diatomic |au; M • 28.98):

A . l-*7 x 10-'.2.4 . 1M x 1()-2 Mtt/eM« .CiIorr( (28.98)". 0.4 ao that tha haat conductloa par unit area from a aurface at a - 81

and o - 0.7 , uill be

2 -2 EQ - 0.7 x 1.64 x 10~ IIJIJ (373 - 293H0 -

- 8.87 x 10~3 watt/cn2

Table 2.14 lists values for y , KCeq.2.103), and A

Table 2.14. - Values of heat conductivity

K Gas K y 2 1 1 H.m"1 'K_1 w.«- 4- p.-

H 2.016 1.41 0.173 6.072 2

He 4.003 1.67 0.143 2.935

18.016 1.30 H20 - 2.649

He 20.18 1.67 0.046 1.307

K 28.02 1.40 0.024 1.663 2

°2 32.00 1.40 0.024 1*557

A 39.94 1.67 0.016 0.929

co2 44.01 1.30 0.014 1.696

Bg 200.6 1.67 - 0.415 TkMTmMi. conductivity at law preaaurea is uaed for aeasurlsg the praaeure of gaeae hy uaing tha thwil conductivity launaa. Tbeae gauge* operate generally under condltlona in Htalch the energy Input for heating a fllanent la Maintained constant, and the preaaure la determined by the variation o£ the tanparature.

Tor coaxial cyllndera of radii r, and r2 » (r, > r2) the rate of energy tranefer from the inner cylinder or wire at tenperatura T., i.

lAaxe

ar ' 1— a- aKx^) a.u6) where a la the accomodation coefficient of tha aurfacaa. For a given gauge A , a , T. and E being kept conatanta, tha taapaxature I waaauree the preeeure P . Since A i» a function of tha nature of the gaa (aq.2.14), the gauges are to be calibrated for each gaa separately.

A detailed description of theut gauges will be given in the' appropriate Chapter. Appendix

Appendix 2.1

il lam uniti it ill bill u* rfUa

1 imm-Iorw - HOT dyuM,

.dyn. NT * (-11 cf kd

,.,„.- • 10"* 3.34 H 1.030 IAM xur» XIO"* vicr* 1 o.aws 7,233 1 [K^U-i - 4.4 IB 4.4*8 1 32.17 453.8 0.4434 X I0» I JH"ll>d|l — I.3b3 C13IJ3 3 108 14.10 1.410 X 10' X liT1 ' t sntm-lon* - 08Q.7 B.B07 3.205 7i- » 1

lib -31.173011*1!

Apj>eiulxx 2.2

i« Uit ficlar* in the ihidtd pi

E KO -« „u M 1* '- • „™. • aooi «.U3 UOi X 10-* X 10" xiir* xio"» 1 KlLWltlAM - •000 1 OKI B.Q54 3U7 xi-« xio* *"*' ItM- 9JJTJ V0T.3 «2.1A B4M XIOT* xio" rf 10' - 84 -

Appendix 2.3

BW/fcr rut/ it-ns k> cml/*M k* WATT! Mto

12JJ «i« MJ» TJ» MM> UN* , - XIP-* XIO * X«T« Mtt IJMJ &o» &M9 UM U» X 1*** XI*"1 x»o-» xi

Appendix 2*4 ELECTRIC CHARGE

***** COUJ. ^ *t«tM«l

1 9M0 3.TO0X HT* 1J)T» x lt"» 1 COULOMB - 1 1JJJ4 X »** LMX 10* MJI *.M2 X 101' 1 1 *IU*WltM»b (1 a JW x ID1* 1«) - ».MI X 10-*' fcJJS X 10'w iw x io-<* i

1 ritrtnqk cbvw - IJ02 X 10"M amtMib.

Appendix 2.5 TIME

" •M hr min DEC

MM a,TH X 10> MMX I0> S.I» X 1** 1 li I MO IM0X1I* 4.MXVT* 1 M> MO* I.M7 X WT* 1 00 •.HI x W* 1 SECOND- I.MT X 10~* 1 1.1W X W» i.m x «-•

t r**r - M&2131 *W» J>n I Ti T s'j = 2 y2. _ a 1";i I? Sx l\ a*

3 ?2 =4 -l - 5*= | !* 2 x 8 ; A i! 5 * %.- i !! :« i* 3> rib ° - !i i °* j; - 2 F- \ * 1*.a r _ i ;l :*^ -J '•' £

I If 1^1 »!

_ 1 5 : !* 1 : s .x sx S " * SI I !.I £ 1 si Sx i= —

oT 3 i - = *b

E l\- 3! ^ 2 5 "2

1 - 1^ 1 5^ Mil \ I as : ! • ; ! fiiii i s I I i i ; 11

- 87 -

3. GAS FLOW AT LOW PRESSURES

Commonly used symbols.- A area a,b sides of a rectangle B circumference C conductance c specific heat at constant pressure c specific heat at constant *olume D diameter f molecular sticking coefficient F force k Boltzmann constant L length HI mass of molecule M molecular weight N number of molecules P pressure P average pressure P probability factor q gas flow (molecules per second) Q throughput r radius B., gas constant per mole Re Reynolds number S pumping speed S pumping speed at the pump inlet t time T absolute temperature V volume v velocity 2 u specific mass (per sec, per cm ) W mass - M -

V (*••*) - "tlO C /Cy r\ («ta) - viscosity 1 (laabda) - Man fr«* path i (xi) - •olacular dljMttr 0 (rho) - density, seat par unit voluaw T (tau) - (tlaw) period 4 (phi) - molecular incidence rat* - 89 -

3.1. Flow regimes, conductance, and throughput

3.11. Flow regimes

From the previous chapters it results that the gas in a vacuus system can uc in a viscous state, In a Molecular state or in a state which is intermediate between these two. When a system Is brought from the atmospheric pressure to "high vacuus", the gas in the system goes through all these states. The scan free path of the gas molecules Is very small at atmospheric pressure (see Table 1.1)* so that the flow of the gas is limited by its viscosity (see Sec. 2.51). At low pressures where the mean free path of the molecules is similar to the dimensions of the vacuum enclosure, the flow of the gas is governed by viscosity as well by molecular phenomena; this Is the Intermediate flow. At very low pressures where the mean free path is much larger than the distensions of the vacuus enclosure, the flow Is molecular.

In the range where the state of the gas is viscous, the flow can be turbulent or laminar. When the velocity of the gas exceeds certain values, the flow is turbulent, the flowing gas layers are not parallel, their direction is influenced by any obstacle in the way. In the cavities formed between layers, spaces of lower pressures appear. At lower velocities the viscous flow is laminar, i.e. the layers are parallel, their velocity increasing from the walls towards the axis of the pips.

Thus the flow can be turbulent, laminar, intermediate and molecular (see Table 3.1). The limit between the turbulent and laminar flow is given by Reynold's number, .while those betveen laminar, intermediate and molecular flow are described by the value of the

Knuditn number. - 90 -

T«il« 1.1. - lion n$lau

-i^^ Stat* of Flow regis* Condition tit* gas

*• > 2100 turbulent Q > 200 D (air) VX8COU* K* < 1100 lasdnar Q < 100 D (air] ; D/x > 110

transition iutentediate 1 < D/X < 110

rarefied •oleculax D/X < 1

The Eeynold nunfcer is a dlaenslonlass quantity expressed by

where p la to* density of tba gee, T tb* velocity, n tb* viacoaity, and D tb* diaaatcr of tba tuba. It waa eetabllahed that for Reynold's niadiara^ larger than 2100. tha flow la anttralT turbulent. while for fca < 1100 It la entirely lataar. The exact values of n* for wbich to* flow changes froa turbulent to laatlnar depend upon tha roughness of tba surface of tba tuba and otbor expariaental factors, but for aoet cases tba •entlouad range la valid.

Tba expression of tha leynold number, can be related to tha throughput Q (aa* Sac. 3.13) which la defined aa the quantity of gaa flowing through a pipe, expraasad in pressure x voluae units par unit tins. Thus

C3.2) and tine* (according to eqs. 2.17, 2.19)

p - n.m -SMP_ (3.3) o the expression of the Reynold number (eq. 3.1), can be written

116 RT t, „ .t In D "•6) o nD P *• o '

For air at 20°C> n •= 1.829 x 10-4 poise, R - 62.364 Torr.liter/*K (table 2.4), and M = 28.98, so that according to eq. 3.4,

2 Qair - 9.06 x 10" Re.E (3.5) where Q . is, expressed in Torr.liter/sec, wiiile D is in centimeters. By using the limits Re » 2100 and Re - 1100 „ it results that the flow of air (at room temperature) will be turbulent if

Q > 200 D (3.6) and it will be laminar if

Q < 100 D (3.7)

The Knudsen number is the ratio D/X , between the dlaaeter of tht plpa and the mean free path. In terms of this number the ranges can ba defined (see Sec. 3.6) as

D/A > 110 Viscous flow (3.8)

1 < D/A < 110 Intarnediate flow (3.9)

D/A < 1 Molecular flow (3.10) If an error of about 10Z is admitted la calculating the conductances (u* Sac. 3.6).

By using cqs. 2.58, which gives A»P - 5.10 , fox air at root tsmoerature, it resvlts that the condition for viscous flow la

D.P > 5 x 10-1 cm.Torr <3.11) while that for molecular flow is

•rr (3.12) where D la the diameter of the pipe (cm), and F the* average pressure (Torr).

3.12. ^Conductance

The flow of a gas can he interpreted as the number of molecules N-, passing per unit tlae through a cross section of the pipe.

Considering two subsequent cross sections 1, and 2 of the Same pips, th« nuaber of molecules crossing- them v 11 be

SJ.DJ (3.13)

N2 " A2 v2 °2 " S2,n2 (3.W) where A Is the area of the cross sections, v is the flow velocity of the gas, n ie the number of molecules per unit volume (see eq. 2.18), while S - av la the rate of flow, or the pumping speed.

In a permanent flow, the Dumber of molecules crossing; the various cross sections is the same. Thus N, - H- » H , and

(3.15) which expresses Uoyle's law, t>lnce n the number of Molecule* per unit volute, is proportional to the pressure (eq. 2,17).

By writing that the drop in molecular density (or the pressure drop), is proportional to the number of molecules, it results chat

N - C(n. (3.16)

where the factor C can be a constant or a £unction of the molecular density (pressure). The factor t: is called the conductance of the pipe. From eqs. 3.16 and 3.15, we ha-re

(3.17)

aleo expressed in V/t units. Various such units used, are Listed in Table 3.2. The value of the conductance depends on the kind of flow and the geometry of the pipe; Sec.3.2 - 3.6 deal with their calculation.

i Pi.MPisi: SPUD UMIMI.X n.Vl

1 fix 10 J.6XI0-* I MylQ-» Iri.AT 1 6/10-:i3.5) 10 "f l.AT: 10"; m*

When two pipes are counected in parallel (Fig.3.1), the number of molecules N reaching the cross section 1, is divided in two parts. - 94

'Fig.3.1 - Coodvctancea in parallel

M *, flowing In pip* *. end N. In pipe b. If the •oleculax densities at 1 and 2 are o. and xu (Fig.3.1), then according to eq. 3.16,

Ma " Ca(nl " n2) (3*18)

"b-S^l"^ <3'19)

and sine*

"a + "b ' " " C(nl " "^ <3,20)

It raaulta that C - C + C.+ ... (3.21) a o vbara C la th« conductanc. of tha ayataa, and C , C. ara eh* conductancaa of tha plpaa connaetad In parallal.

Whan conductancaa ara connactad In aarlaa (Flf.3.2) and tha

•olacular danaltlaa at 1, 2, 3 aca n, , n- t *°d n, , It can ba wrlttan:

M - Ca(0l - n2) - (^(nj - n3) - C<0l - nj) (3.22)

i 1 - 2

Fig.3.2. - Conductances in series

where C , and C. are the individual conductances of part a, and b, while C is the conductance of the system. From eq.3.22 it results that

3.13. Throughput and pumping' speed

The various pumps used in a vacuus) system, remove (evacuate) gaa f roa the system. The rate at which the gas is removed is measured by the pumping speed S . The pumping specd_is defined as the volume of pas per unit of time dV/dt which the pumping devica removes from the system at the preesure existing at the Inlet to the pimmi. The pumping speed ts expressed In llter/iac, m /bx, etc (see Table 3.2). The throughput Q is defined as the product of the pumping speed and the inlet pressure, i.e.

The throughput is expressed in Torr.liter/sec, atm.cm /see, etc. (see Table 3.3). The throughput unit of u (microns of Hg) x liter/sec, received Che name of lmec (tu/atc). CtoNvnaoN FACTUM

1 n n *g»!\mmmH* pf£L j**™**'! »"- * *" ' i<*< I

SjUXlf-* J.3K1I-* 1.4 v 10' 1.11* 10"« J.74vK>-' J.lsie-i J.4>.H-t4.«Kli*t |.4VKli-H

( UxW Mftxt*-' 4,1x10-* •:.n»tt- t.»lxtO-» 1.2*10- )%?»>»-* niMi»-ti),j)xii*t IJ*H»'« 1 MHI-" ,4.72x10" ,J.7) UO-> 7.4x10" *.2*ID-* 4.7lJi18-*j7.t3Krt-i IHxH*1 ircRfthv t.«J 1 •*.u»io- 3.1 MOM LOT K 10- . 1.1*10-' MlxlO-tjt.llxlf** 4.ltxlt'« • 2.12 1.47 t : 7.».<10" xlO » IJ2 -I0"» I\ID-i |l.«txlt-< t,i2»i»-«

SJ4* 10* «*4 27 It.* 11.7 t Kl0-» , l.*H>. 10-1 1.27xlO-» j 2.14x10-1 7.T5x»'*

IT* It* JO** IM* HO MO 1 . 1.4. 10-' MxiO-f ,IMK W« I.M » «'•

uixir MM tt» I12U TtO LI* , 7.»KI0-I 11.2**10-1 I.MAII'* 4Jxl# 47M » SO I4T0 rw 1.5* 1.J2 1 |l.«xlO- 4.12 x II'*

StTQ 23xl*> | LUnlV L,2T>-MP aiso 9.12 7fl *,*T j ' IM •.!•-* ttxlt* t MJxll* >.4*A10» 2J»M0» I.M x I0« S3 1 214 l«l » 1 By multiplying eqs. 3.16, by W , it results

(3.25)

and with eq. 2.21, 3.15 and 3.24 we have

NkT - ^ - SP = Q = C(P. - P.) C3.26) n ^12

According to eq. 3.26, Q is the quantity of gas entering per unit of time Che pipe with conductance C , at the pressure P. . If DO additional gas leaks into or is removed from the pipe this same quantity of gas Q comes out the pipe at pressure P. . Thus if the system is isotherm!c (eq. 3.26), Q is the same all over the systen.

By analogy with the expression 3.24, the pumping speed at any point of the vacuum system is

S - £ (3.27)

where Q is the throughput in the system and P is the pressure at the point at which the pumping speed is defined. Substituting the

values of Pj^ - Q/S1 and P_ = Q/S, into eq. 3.26 we have

Sl S2 C which is identical to eq. 3.17. This equation shows that the pumping speed «t any point In the system can be obtained from the known pumping speed at some other paint and the conductance of the portion of the system (pips*, holes, valves, ate) In between.

The punping spaed S obtained in a chamber, connected by a conductance C , to • pump having a pumping speed S , Is given by

i-r + z (3-28) -M-

HM wy rim la Ms. 3.3, can to «Md to ««lv« fHlcUy ••,. 3.2*. n ••.. 3.»-

S m C, H 1— i h

*\<\

tif. 3.3. - Moaofm Cot cdculatlnt puaplut W—A at eoaductaac** in Mrtw. If «q. 3.28 U BpniHd In th« torn

W8 (3.29) - 99 -

til* decrease of tha pimping speed S/S t results ss e function of the retio C/S between Che conductance of the systcB end pusplng speed (of the pup). Thla relationship le repreMDted In Fig. 3.4.

C/s P

Pig. 3*4. - S/S as a function of C/S

It can be seen (Fig. 3,4) that when the value of the conductance Is equal to that of the pumping speed of the puap, 50Z of the ptaplnfe speed is used at the vacuum vessel. In order to use SOX of the pimping speed the ratio C/S nust be 4, while for s ratio C/S - 0.1 , only 10% of the pueplng speed of the pump is felt In the vacuum enclosure.

3.2. Viscous and turbulent flow

3.21. Viscous flow-conductance of an aperture

A large voluae where the gas is at « relatively high pressure 7. (e.g. atmospheric), is connected to a second voluee where the i t. , hy an., upartura of araa a, (Pit* 3.3). If tlW fMMM* f- is inch that tha naan lyii path »f tha aolaculas is cu—ara*' to tha aiatanalona of tha apartura, tha gaa will flow

valocitr i* *ha vicinity of tha apartura, no that aftar pasaing it. tha gas Jat haa a atalaua croaa aactlon (rig. 3.5). Aftar this

Vlg. 3.5. - Viscous flow through an apartura contraction* tha jat haa aoa* (about 10) auccaaalva axpanaiona and contractions, until finally It dlffusaa In tha aaia of gaa ?_. By kaaalng conatant P, , and decreasing P, , tha quantity of gaa and

its valoclty ara Increasing, up to tha atata whsrs tha ratio P2"i reaches a critical (nlnlnua) value, corresponding to a valocltr equal to that of.tha sound_

Basad on tha laws of tha adlabatic espanalon. It was found that tba throughput Q of gaa flowing through tha apartura* is given, byt

o-*-M5T".el^V'f-ri ss*T-»*i»-irfTlrf <••t3-»•> expreesed in C.G.S* unit a. In aq. 3.30, A la tha croaa aactlon of

tba apartura, y - C /Cv tha ratio of tha specific heat at conatant pressure to that at constant volume (see Table 2,Id), R the gas constant (aee Table 2.4), M molecular weight, and T temperature of the gas.

Since C • q/P. - P„ (eq. 3.26), It results that the conductance for viscous flow of an aperture is given by

f ill 11/2

2 where A (cm ) , C (liter/sec) , T (°K) , M (g). For atr at 20*C, Y - l.ft

, rF.-iO.712 r -P.,0.28^1/2

P P 1 and the throughput Q (eq, 3=30) is Q - 0 for 2' 1 " » is maximum for

Pj " l> + lj " Tc t3.33)

this is called the critical value.

For air at 20*C, thia value is y - 0.525, and

Qc - 20 APX (3.34)

2 where A (en ) . P- (Torr) and Q {Ton.Liter/sec), Therefore the p2 conductance* for "p~ ± 0.525 , ia given by

(3.35)

p2 while equation 3.32 glvea the conductance for the range 1 2. p- > 0*525, Ubaa == < 0.1 , (quatlo* 3.35 on b* »ritt«o

C * 20 * (1.36)

mi la till* tat* (»"1T) C cub! cooaldand lnd*r*ndnt: of th«

If th» aputu* i. cw«14«r«d by its paplng efface on roluaa 1 (lit. 3.5), *t» paaplag spMd la (Ivan by 8"^•^-[- • g- • C *. -»C |1-^^ | (3.37) Figim 3.6 ahova the Yalua* of C/A for air. u wall aa tha -vain*

120 _ Air 20° C

rioo - £ - 7» 80 h

< 60 Si 7 42 g 40F Q _S . < 1 AP«'A S 20 1 1

i . i I.I.N 0.2 04 0.6 08 UO

P2/P, fig. 3.6. - Conauctanca C and pwpinff apaad S of aparturaa (vlacova flow). A - araa of aaartuea. ? - -j&- , and show* that whale reaches s Maximal value at low N^'i **!«••» T haa values which arc always greater than ~ .

3.22. Viscous flow - Conductance of a cylindrical pipe Poisauillc'a law In a long tube of uniform circular cross section (Fig. 5.7),

lower pressure p. • T*1* 8as contained within « chlD-walled cylinder of radius r , a wall thickness dr , and within a differential length dx , experiences a force in the direction of Flow given fay the cross sectional area 2irrdr , and the pressure difference dP , so that

dF, • ^ 2*rdr.dx (3.38)

v = 0 2&

v*K2 0 i Pi p2

ttg.3.7. - Viscous flow In pipes

Tha Minus aign appear* since the pressure gradient la - dP/dx (the pressure decreases In the direction of the flow).

Due to tha viscosity, the velocity of the gas at the internal •urEace of the cylinder is greater then that en its external surface. The force due to the vlscoaity Is (eq. 2.67)

dv -iM-

ahaia n la cha coafflelant of vlacoaitr, and A tha aurfaca of tha cylladar A - 2itdx . Tbaratora ou tha internal autfaea of tha cjrlmdar, tola forca will ba:

F2 - - a 2T«X § - - 2„n [r g]d* (3.39) dv/dr being negative tola force is directed in the direction of tb* flow.

On Che outside surface of the element, the foree due to the viscosity is n - -h+TT *] -*•>*< [' £+fc (r &H °-«>

Which is directed opposite to the direction of the flow. Therefore the resulting (viscosity) force applied on the element Is1

2irr dr.dx + 2*n ~ |r ~]dr.dx - 0 (3.42)

Two subsequent Integrations will give dv/dr , and v t dv r_ dP •=1 dr * 2n IdxJ (3.45)

The constants K. and K_ can be determined by the boundary conditions. The velocity is a maximum foe r - 0 , thus for thin value dv/dr - 0 Ceq. 3.45), which result in K. - 0 . The velocity is zero near the wall, thus for r • a j v • 0 , which gives K_ - - •£— -j— . Finally equation 3.46 is written

which shows; -that the velocity of the gas is directed i.. the direction of the pressure drop, and -< that the velocity of the gas is a parabolic function of the radius, with a maximum velocity v - K» • - ~rr nr- otl tne axis (r • 0) ; and v - 0 at he wall (r - a ), es shown in Fig. 3,7.

The volume of gas flowing through the cross section of the tube per unit time is obtained by integrating eq. 3.47 across the cross section of the tube, i.'e.

a dv f. (3.48) dt j '

The throughput is given by

o - v£L. ™i* H dt 8n dx and by integrating for a length L , P.dP

au i r«» O.50)

In L 16nL

*1 r2 "1 '2'"1 and the average pressure P , is

r 2 equation 3.50, is written

" isSnl *>! " P2> (3-*» where D - 2* is the dieaeter of the tube. Thia equation le known as the PoieeullU lew. In its form expressed in eq. 3.51, ell factors ere in CGS unite.

The conductance is given

C-P7^T:-I3SL7 °-5"

A prsctlcsl for* of eq. 3.52 is

C - 3.27 x 10"2 2p P (3.53) vhere 7 (Torr), I,

for sir st 20*C. this equation is 4 cslr " 1B2 E" ?

In the derivation of eq, 3.51, it tras assumed that th> velocity of the gas 1B zero at the tube wall. Some gas molecules in striking the wall experience specular reflection and thus retain the same component of velocity in the direction of flow as before the impact. Other molecules strike irregularities on the wall and bounce several times, the molecule being adsorbed on the vail and then reemitted later with a random distribution in angle and velocity. These molecules represent a layer of gas which is at rest next to the wall, and provide the viscous drag. This effect is described by the coefficient e , which is given by

where f is the fraction of molecules which are adsorbed and reemitted and! 1 - f is the fraction which are specularly reflected.

Since the velocity of the gas is not aero at the wall, eq. 3.51 Is written

thus from eqs. 3.56 and 3.55, ie results that

P - P 3 2 Q -

i R 1\1/2 „ , c2 " 16 (2 M J f

— 4 when the pressure P is sufficiently high the term in D d< and che flow follows Poiaeuille's law. When the pressure P - 1M -

that taa tara* la 0 aa4 D ara aaual, ch« character of tba flow laparta fraa talaaallla'a lav* DM praaaara tor aalch thi« cooaltloa

oeeara la raffarafl to aa tba" traaaltlon araaaura Pt , which la

Tela alft be eoMUmd u the loveet limit of •oleeullle flow.

3.24. Tlecoee flog - lietmnlir crew aoctlon There am fwr indications la the literature, on the conductance (in Tlecoee) flow of pipse with noocirculer crou section. Cnthrio and WidrTl-f— (1949). give for ths conductance In viscous flew of recteaseler acti. toe expression

C - 3.5* x 10T* * *£• P <3.59) la CCS unit a, trim* T la a correction factor with values llated In Table 3.4. tarnation 3.S9 can be compared to eq. 3*52

Table 3.4* - Correction factor Y, « mm e function of the abaft a/b of the rectangle.

a/a .a 0.9 p.* 0.7 0.6 0.5 0.* 0.3 0.2 0.1

t l o.» 0,95 0.95 0.90 0.82 0.71 0.58 0.42 0.23

If «q. 3.52 for tha circular croaa •action la wrlttaa In tba fora It can be Men thmt the conductance of a duct vlll aquar. croie •action (Y - 1 , «q. 3.59) Is leea then that of a pipe with circular croaa aactloDp in tha ratio 3.54/4 - 0.88.

Equation 3.59 la written In practical unite aa

-2 A2- C - 4.71 IIO'TAJ-F (3.61) nL

n (poiee), C (llt/aac) aid 7 (Table 3.4).

For air at 20"C, this equation is

n _ ten TJ ~_ n (3.62)

in the saae units as 3.61.

Helnze gives the expression due to Boussinesque for the conductance of rectangular ducts in laminar flow as

in OGS units, where V is expressed by • - -!JB [*»»& + £ «•*• «*+...] (3.64)

and a/b la the ratio between che nail side a, and the large side side b of Che rectangle. Values of •> are plotted on fi*> 3*9.

At high values of a/b eqs. 3.63 and 3.59 give the s«aw values of C , but at low values of a/b their results ere different.

* W. Belie; Elnfuhrung in die Vakuis.tacholic, VEB Verl, Berlin, 1955, p.116. t:]l\ "i- vh

& «• t

Hg. 3'9> - Correction factor 4> for eq. 3.63.

3*25. Tlacoue flow - Annular crow aectlon

Tor a long duet having an annular croaa •action raaalted

betvean toe radlua of the tuba rtf , and that of a concentric cora

•4 , tfca conductance in viacoua flow la given by

(3.6S) In CGS units, or for air this will be

r> — oann ±_ L loio*-°g -^- J

where P* (Torr), L (cm), r (cm), o 3.26. Turbulent flow

Considering eq. 3.6, which indicates that turbulent flow occur* only for throughputs Q > 200 D (Torr.lit/sec; air), it can be shown that such situations are very rare in vacuus systems.

One of such cases occurs when air is admitted into a system which was previously evacuated to a low pressure. If the air is admitted through a pipe having a diameter D and length L , the condition for the existence of turbulent flow is

4P2-P2 X 2 Q - 182 2- 2 1 200 D (3.67)

Since F. - 760 Torr, this condition gives that turbulent flow exists if

r. < f7602 - 2.2 iJ (3.68) which shows that by admitting air through the'usual pipes, valves, etc, the flow ia turbulent practically until the pressur* in the vacuus vessel reaches 760 Torr.

Considering a large diffusion pump with S - 10000 lit/sec ttlfl'3 Torr, thus Q - 10000 x 10-3 - 10 Torr.lit/fee, the flow is turbulent if 10 ^ 200 » thus D <_ 10/200 • 0.05 cm, which is never tha caaa in vacua* aaatems. A- rotary pump can give Q - 60,000 Torr lit/sec. In this case tha flow is turbulent If 60000 >. 200 D thus 0 ^ 60000/200 - 300 ca , which la alwaya the cm in vacnua. >, ass-

3.3. Molecular flev 3.31. Molecular flow - Conductance of an aperture

A voluae where the presaure Is P, , la connected through an

aperture (area ft) to « aecoad voluaa where tha preaaure la P2 < P, . If the prasauca Pj la low anough for molecular flow (aq. 3.12), tha rata at vhicb tha gae paaaaa through tha apartura from P^ to P« la («q«. 2.48, and 3.24):

3 3 Qx - Px ^ - 3.64 x 10 f|] APj nbar.cs /eee (3.69) while tha gaa paaalng from P, to P.. , la

°-2 ' p2 at " 3,W " 1<)3 [M] "2 »»«'«"S/»»<' <3-70>

In molacular flow, where la no collision batuacn molecules, they pass through tha aperture In both dlractlona without anT Influence on aach othar. Tha throughput Is tha dlffaranca:

3 Q - qx - Q2 - 3.64 1 10 [Si A^ - P2) (3 71)

which la dlraetad from P1 towarda P£ , alaca Pj - P2 > 0. Thus tha conductance of an apartura of ersa A (In molecular flow) la :

)l/2 , ' 3.64 x 10 'P1, - P'2, ll/2 if A llter/eec (3.72) 1/2 (i) (3.74)

2 C , - 9.16 9 llter/Bec (3.75) air

From eq. 3.72 It can be seen that the conductance (Molecular flow) la Independent of the pressure.

The "pimping speed" of the aperture Is given (eq. 3.27) by:

C(F - P > , P ,

S-S 3, 2_.Cl--i (3.76) »r FI T \> and for air at 20*C

S - 11,6 A |l -=£•3| (3.77) 2 where A (cm ), and S (liter/sec); and for the usual case where

S - C - 11.6 A (3.78)

It can be aean that tha pumping speed is a function off F,/P. ^ up to a Maximal value of 11.6 A. Comparing this to eq. 3.37 and Figure 3.6, It result* that the maximum pumping spesd of an aperture at low prcMure (11.6 A) la smaller than that at high pressure (20 A).

3* 31- Molecular flow - Conductance of a diaphragm

Consider the diaphragm of aperture A as shown in Fig. 3.10. Here 1 and 3 are large volumes, connected by a pipe of cross section A • lb* pipe 2 is conaaeted to VOIUM 3 by a dlaphraap of aperture A , which la aaall ceapared Co 3, but of the eeaa order of aaiaituda - v The conductance of the ayetan la the direction 1-2-3 will be .

(3.79) when C la tha conductance of A In the direction 2-3.

Fig. 3.10. - Diaphragm of fact.

Tb« conductance of tho OH* ayetaa in the direction 3-2-1 will bo

+ <3.ao) c cA- ca tb« conductance* of the ayatea In both dlractlona mitt ha equal* If not,a flow ahould axlat avan If the preaeuraa In 1 and 3 are equal, which la lwpoealbla. Thua C % C2 Ce C* °a which leads to

1_ 1_ _1_ (3.82) «e " CA ~C A„ i.e.

(3.83) e "4 1 - A/A o and using eq. 3.73, for air at 20*C:

11.6 A (3.84) e 1 - A/A

For A « A , eo. 3.83 gives C ^ C, (aperture); for A - A C £ •" (no resistance to the flow); while e.g- for A - 0.5 A , C» 2 C. (dlaphrgam effect).

3.33. Molecular flow - Long tube of constant cross section

Knudsan (1909) derived the equations of the conductance of long tubes for low pressures (mclecular flow). In this flow the molecules move in random straight lines, between collisions with the wall. The number of molecules impinging on the unit surface per unit time is (eq. 2.46) a " *« and the number of molecules striking the wall each second la BL n v q . * BL 5—— O.S5) where S is tha periphery of tha cross section, and L chc length of tha tuba. The Molecules arrive on the surface having an *nergy corresponding to thair v velocity and the drift velocity v in tha direction of flow. They are atopped at the surface are reaaittad randomly with their velocity v . thus the aonentua transferred to tea Mil la asr . Th« eoaentua transferred by «11 the aolacolee to the mil, la that

q' - q.anr - SI n T mv/4 (3.86)

Tba ameer M of aoleeulea croeelng tha croaa aactlon A of the pipe par unit tie* la («q. 3.13):

a • A.».n (3.87)

eed ca« preaaure diffareoca oP achiavad correaponda to a force

a - A.AP - A.kX.en (3.88)

For eqeilibrlum condition q' - &F , thua

*AkT 4o - BL n v an (3.89) av Froa 3.87 and 3.89, wa hava

H__*A la_ o<90j An BL ...AW1 According to eq. 3.16 N/on * C , and uaina tha value of v 2_ feel)

(aq. 2.3B)( wa obtain

This equation contains tb* aaaueptioa that * unlfom drift velocity v Is superlaposed upon tha random Haxwell-Boltsaann distribution of cb« •olaculu. Knudaen bu shown that It should ha hattar to aeiuaa that tba auperlapoeed drift valocity of a molecule la proportional to Its raodoa velocity. On this modified assumption Knudtss fo :.d that tha numerical factor in aq. 3.91 must ba multiplied, by 8/3K , ao that the conductance will bat _5_ fell1'2 *L 3.U , 10* ftl1'2 A2 c --^^3,7 I •- J tr"• - Z ft Sr »•»»

In C6S units. The conductance of a tube of unifozm circular cross section is

fll1/2 D3

where D (en), L (cm) and c (liter/sec). For air at 20#C, (T/M)1/2 - 3.16 , thuB

where D (cm), t (cm) and G (liter'aec).

For a tube of rectangular crass section, with sides a and

A - a.b and B - 2(a + b),

thus eq. 3.92 becomes

C K C3 95> * 3 UTIHJ Ca + D)L 2/r (HJ (a + b)L *

In CGS unite, where K is an oacperiMntal correction factor taking Into acount the asysnetry of the cores aection.

Values of K are listed In Table 3.5.

For eir at 20*C, «q. 3.95 will be vritten

c.ir • 30-s TSTTETL * C3-96) where L , a , b , cm, and C (liter/sec). Tabl* 3.5. - Corraction factor J. ,

Wa 1 0.667 0.5 0.333 o.a 0.125 0.1

K 1.10B 1.126 1.151 1.198 1.297 1.40 1.444

For « triangular crow aactlon. with aida a (aqullataral i^3" 2 triangle). It ins found that X - 1.24 , and alnc. A - -^ a B . 3a ,

2mJ 1 (3.97)

1A 0GS unltu, and for air at 20*C*.

0.9S) yaare a, b, L (c), and C (llter/aac).

For an «.nvii»r eroaa aaetloo batvaan two eoncantrlc tubaa with dlaaatara P.I.U-fn'- D,Z) ; B - »(D + D.) , thus th* ' 0 1 4 0 1 o 1 conductaaca la . r MI tto2 - Di2>2 C-f [£) t5.+D*,L »o "•'»

In CCS units* Tti* factor KQ la ilvan in labia 3.6.

Tttla 3*6. - Correction factor K v». 0 0.259 0.5 0.707 0.866 0.966

*. 1 1.072 1.154 1.254 1.430 1.675 Por air at 20*C. eq. 3.99 becoawa

(D - D,>2(I] + D ) C,ir - "-1 ——Sr5—~ \ °-100> where L, D , D, (ai)i and C (liter/sec).

3.34. Molecular flow - Short tube of constant cross section

If the length of the tube Is decreased to zero, the conductance aniat decrease to that of an aperture. Thus the correct way of writing the equation of the conductance of a tube is (eq. 3.23}

£ + £-+?-

c - Trhrh - c, .,,L, O.ioi) ' ^ + ce • <% 1 + «L/ce)

Where C, Is the conductance of the tube (eq. 3.92) and C ia the conductance of the aperture (eq. 3.84).

Fran eon. 3.84 and 3.72, ve have

3 hi1'2 A 1 " V where A ia the cross section of the cube, and A, the croaa section of the upstreaa vessel.

By using the value of C_ glvta in eq. 3.92, It results that

c. 3.6a * io3./7 "• I V "• l V > 3.101 aaa ba vrlttaa

C • C^.V (3.10*)

•tan *' la ei»rf»'i factor, uhlch la azitaaaad by

-i ± (3.105) -tf Mr cltcalar ctoaa aactloaa:

2 »DJ A-I2_ = .-.D , A.-^,

••1 Claaaiaf'a factor la given by

r' - x O.10S) —H1-?)

For cim abate D < 0.2 DT (tuba diaaater D M>11 coaparad

to aaaaal illialtar Dy ), aq.S.lOt cu ba azpnaaad in It* aiaaliflad foxmf .. 1 (3.107) i*i.a»a tbufl tha coaaacraace of • abort tuba will ba:

c. ll}1/2£r iM -RTF-

lHJ L l + 1.3:

W 3.01 © P3 where L (ca) , D (ca) , and C (licar/aec).

This equation expreeeaa the fact that the "and effect" ean 6a takan into account by considering tha pip* as being longer by 1.33 diametera. Obviously for air at 20*C, tha conductance of a ahort pipe ia

C.lr - "•' L + f.33 D »-10"

For a pipe of rectangular croas aection (eqa. 3.95, 3.104, 3.105) the conductance will be given by

3 M1'2 -2 h2 3 a C - 9.71 x 10 (£] (.+b) Lt2.66.b * 0-"»>

In CGS unlta. This equation becomes for a slot in which a » b , and in which Che length I. of the slot In the direction of the flow ia not large compared with b

frl1'2 - K2

>vhere a (ca), b (ca), L (ca), c (liter/aec), while for a long narrow alot where a » b . L >=> b , the conductance will be:

1« . H* -rv K (3.112) vhar* a (en), b (ca) , L (ca) and C (llter/aec), and K ia the correction factor Hated in Table 3.5. For air at 20"C, this aquation reaulta in -.2 (3.113) a , b , L (ca) , C (Utar/aec). TIM CMfectuc* of • abort tuba of annular croaa aactloo., ia (lvaa (aaa. 3.», 3.14, 3.105} by

C3.«lfl]1/2(°.-V2<°.*V lM' L+1.33 (D - D ) \> W-U*J o l »,u (cs), c (UntfiK), which tot mil at Z0*C, becomes

CD - D.;2(D + B ) «.«C . - 12-«•1! ,.;,, A,., * .(3-1") alr " L + 1.33 CD0 - D1) *b snare D ) is the fftitM of the outer cylinder D, (cm) is the

diameter of lunar cylinder. !• (cm) f and C (liter/see).

3.4. Condnctsncs of combined shapes 3.41. Molecular floir - Tapered tubes equation 3.92, can be alao be written

1 fjkT\1/2 •lace (eq.2.38) V » — Fj-J . £ is the shape factor as shown In Table 3.5; 3.6. " •' If the conductance results from a series connection of conductancee of lettfht dl , it is written:

£-*^TJ7dl <3-117>

I " r. (3.118) \?-J dL Equation 3.118 1* a general formula which gives the conductances of pipes with constant cross section, as veil as those in which B and A are a continuously increasing or decreasing function of L. For constant cross sections, B and A are not functions of L , thus

2 1 _ 4 A* K 4 tr „ C"TvavB-L "T'avBL*

For a tapered (conical, pyramidal) pipe, having at the saall end the cross section with perimeter B. and area A. , and at the large end B. , and A, » at a distance x , these values will he

A + A k 2 119) „ - *i < 2 - V r • A[»i - c2 - "i> f] «- vHti a_ is one of the sides (or radius), and k_ is the constant ratio between the perimeter and this side (e.g. for a circle k_ " 2ira/a - 2T), while k. is the constant ratio between the area A 2 _ and the square of the aide (e.g. for a eircle k. - ita /a - if). From eq.3.U9 it results that

K MkA

\ [a, + <«2 - a,) fj thus (for eq.3.118);

1 f "» ,.. . *B j Jx r *B I. V 2 ,,,,,, I TT dL ~ r? J T—— 3T"^?T T (3m' epa the conductance (aq.S.lU) Kill be flven by

k2 n2 • a?. (3.122)

Vor « circular cro»» taction

B - 2«r

A-™2

k, 2, 2

and K - 1 . therefore the conductance of • tapered pipe of circular CUM aectlon mil be C«q.3.122):

2 2 „ ti *1 r2 C3.123) C-r (r. + r,iL '

la CCS tmltflj or for D * 2r,

c 7 62 (3.124) I - l«J Tsprsp; *ere tl , L , (ca> and C (liter/aec).

Comparing eq.3.124 with 3.93 It results that the equivalent dlanetcr for a taecred^aMbe-in

(3.125)

For a rectangular croaa section;

»-2C. + b) kj.lSiiSl.jn.ij

A-..h *K-*r-7 k2 2

K •• a function of b/a and has Che values listed in Table 3.5.

The conductance of a tapered pipe of rectangular cross section will be (eq-3.122)

C*._ V ^4&K (3.126 ' 3 av 1 + (b/a) 6^ + a2)L

The conductance of tapered pipes of triangular croaa Bectlon and annular cross aectlon can be calculated in the sane way.

3.42. Molecular flow - Elbows The molecules in a molecular flow through an elbow (Fig.3.11), can be divided in two categories J molecules (1) which collide with the wall in the region of the elbow, and molecules (2) which pass across the elbow. Molecules having the path CD fill see the opening of the tube as; an impedance, thus the conductance of the elbow will be given for path (1) by

1/2 3 »•« I

Molecules having the path (2 Fig.3.11) will pass the elbow without feeling its influenca. Thus in this case the conductance is given by

C - 3.81 3 7-TTT O-IMJ

According to eqs 3.127 and 3.128, an elbow can be represented as a tube with the diameter D , having an equivalent lenght L * which will be situated between < !•- < !• _ + 1-33 D (3.129) utterc L • L. + L_ i« the length as Maaured oa the axis of th* elbow. Li

! Q 1 ^

V1

Fig.3* 11. - Holeeular flow through an elbow.

For a. sure praclse evaluation it can be considered that all the •oleculcs will travel according to path (1 Fig..3.11), when the shape of the elbow Is that of a hairpin, thus the bend is at ISO*. Considering that th* nuaber of Molecules having path (1) is proportional to the angle 0 of the elbow. It results that the equivalent leneth las

, + 1.33ISS 1 C3.130)

3.43. Molecular flow - Traps. Laferty (Ehishaae 1962) considered the conductances of traps tuch as shown in Fig..3; 12, in which the diameter of the outer cylinder Is D • 2a, , and chat of the Inner cylinder la D. • 2a. . Fig.3.12. - Conductance of traps.

The conductance of such a trap Is that of the series connection of conductances C. of the Inner cylinder, and C. of the annular •pace between the two cylinders. According to eq»3.108 the conductance of the (abort) Inner cylinder is

(3.131) [a] L + 1.33 B and that of th. annular apace

C, 3 81 10 K (3.132) 2 ' * [HJ L + 1.33

B »y naglaeting tba corzaction C , and putting x - =ri'* •« —1 ; It ^a 2 I" —"!j-, It raaulta that T o C - 3.81 i 103 f|j | D* f d,K> -

- 3.*l x 103 [|] i 3 a| ffc.Y) (e» 3/a.e) i.e. ^•l.WiMl'IJ f(x,I) (3.133) *2

x|

Flfur* 3.12 plots tba valua

%- 36.7 £

X a u 4n " l'*2 ' * « **1««" of T - A/«2 (4, 10. 20..*.) aa a paraMtax. Tba dottad Una ahova, cba valuaa of X for urimi conductanc* at different valusa of T * FOE large values of T . tbo 2 conductance corresponds to X + X - 1 j that la X - 0.618.

For a rigorous calculation tha valua of K (Tabla 3.6) aust be takac Into account, and tha conductancaa of tha extension of tha Innar tuba and that of tba aid* connection Bust also ba conaidarad. - 129 -

Besides all this the real calculation must also take into account that the trap la cooled and the temperature Is not equal in all the carta.

Figure 3.13 shot's a trap where these various details are alao considered. The trap Is immersed in liquid nitrogen on an effective

Fig.3.13. - Liquid nitrogen trap. depth L. , considered from the level of the liquid nitrogen to the outlet of the inner (Inlet) tube. In such a case it nay be sssuated chat the tenperature of the inlet (Inner) tube, decreases linearly

fro« Tj at the level of the liquid nitrogen, to TQ at the botcoa of the inner tube, so that at a height h , the tenperature is V * *, - * f ft, f ., -.i**-Jl L . T + M. (3.136) 130 -

the outer vail of the trap mxf be considered at T. above of the liquid nitrogen, and at T below thla level.

The trap shown In Fig.3.13 Is constituted by the various parts listed In Table 3.7, which are connected In series (see eq.3.151)

Table 3.7. - farts of trap In Fig.3.7

p«rt> Description Disaster Length Temperature

inlet-elbow I^ + Lj, r & 2ri i

2 B straight pipe L I (eq.3.136) h 3

C dlaphragji 2r2 - *o

2 2r D annular pipe T V l V o

E » 2 2 V 'i l4"L3 Tl

7 aperture 2r T l - l

G exit tube 2'1 S"*i *!

Part A Uaa a conductance expressed (eq.3,'93) by

ft,-,1/S 8r; (3.137) where L la the equivalent length for an elbow (eq.3.130) of 90*:

90 C 3 ai A • - rJ H+L2'i.33ri «•"«

Part B has a temperature distribution according to eq,3.136, thus a- length dl at a distance L from the bottom, will have a tenparature T , and the conductance of this portion will be

dcs •3-81 i dir •3-81 Ir] ^ -STE— C3*uo)

The conductance of the whole length L, , will be given by

(3.141) dC*"3.8l(3f2 *2 1 «.•' ftoa which it results that:

fT,vl/2 8r* ^T7+ /r~ 1 11 ^1 1 o (3.142) L3 ^

Part C Is a diaphragm, which has to be considered since the nplacules coming from the Inner tube will collide with the round botfon. and iron here they have to pass by the annular diaphragm.

According to eq.3.83, the conductance of this diaphrag* is given by <.-«Br*•T \l/2 M

,1,1/2 [rg - r*) r\ [WJ * T (3.H3) - 132 -

Tart • SM am '•—ilae eraaa aactloa., havlac tba outalda mil at IMiimn T , aa4 tha law "all at taaaaratara I , nMeh urlaa acearalaa; ta aa..3.13t. It ca* ba aaaaaad that tha ralatlw aaafcar of a»lacalaa haviaa, avaraga mlocltiaa corraapoBdlnf to tha varlooo taaaarataraa la proportional ta tba ratio o£ auxfaeaa havlag tbaaa laatmlaiaa. Thaa tka avaraga tanparatora 1^ , can ba coaaldarad

I rlt + r2ro rl(T. + bt>+r2To * *l**t cl + t2

T *!' '1 ' O.IW r, + r,

ma ia tha avarafa taaaaratnra of tha outar and laoar wall at a nlafaaca L fma tha bottom. Iha eonaoetanea of a portion dL at aiaraaca L «U? ha il 11/2 » (r, - r.)2 Cr, + r_) *S - 3-« fcr] —*—he——- h <3-U5) thaa tha eoaaactanca of tha nbala laaxth Lj U (Ivan by

Fat 7TTT72- 5 )*£ *•»»

FTaai aa..3,14a« It xaaalta that tha coaaoctanca of part D, la axpreaaad

fT.il/2 tt (r, - r,)* Cr, + t,)1'2 1 % - 3.11 U ••*-« 1—1— 2L3 "T (3.1*7)

<*1 Tl + H V^ U* -. [ITVo <*r1. *, .r2> , J »lW Vart_K, of tb* annular apaca anMrglnf above cha laval of cba liquid nitrogen, hia a conductance

H - »-81 [-B] ° 2 * 1—L-

Pert T. is the exit aperture, and has the conductance (eq.3.74);

rTii1/2 2

Part G, is the outlet tube, with a conductance

C3.150) ft" '2 The conductance o£ the trap (?ig.3.7) Is given by

C CK CB CC °D °E °F CG

For a trap having r, • 1 cm ; r. - 3 ce ; L. - 3 en ; L, - 7 ca ; L- - 8 cm ; L. - 10 cm ; L_ - 6 cm , T. - 293*K ; and T - 77'K „ j 4 5 1 o a conductance of C • 3 llter/aec results, for air assuming chat none of the components of the air condenses on the walls.

3.44. Molecular flow - Optical bafflea Bafflea are systems of cooled valla, or plates placed near the inlet of vapour pump* to condense back-streaming vapour, and return the liquid to the puap. In order to Increase chair efficiency for

condensing- the vapour, the baffles are constructed in such a way that no molecule can traverse them without colliding to tha wall. Ihey are called optical bafflas, since they are opaques for light ray* (transmitted in straight line In any direction. Bafflea are constructed with straight parallel plates (Fig.3.14), or with concentric platea (Fig.3.15). - 134 -

ri».3.H. - laffla with straight plataa

Coadoctaaca of bafflaa with atralaht plataa. Coaaldar a baffla havlss straight » ahapad plataa (chavroa), lncllaad at an anjla i to tha vartlcal (Flt.I.M) ami apacad froa aach othat at a dlataaca p Ik* plataa croaa tha clrcls of tbtlx aocloaura, thus thalr lanjtha dlffar with thalr position. In ordar to calculate tha coaductanca of tha Va

L > L'/eoa Y (J.1S2)

bap. eoa Y (3.153) Tha aid* hn , will be of a length

(3.154) iftMra n la the aeriml nuaber of tbe plate 1, 2, 3... CFis.3.14)

H t tfaa tocal nuaber of equidistant platea, and D the dleaeter of tha baffle.

Tha conductance of an opening between adjacent platea, vlll be (aq.3.110)

2 Til/2 a? b K. (3.155) (h + b) L + 2.66 h *b »• tihera b (eq.3.154) b (eg.3.153) and K , the correction factor Q a K (Table 3.5). for b/a • b/h . L Is the equivalent length of the elbow, given as In eq.3.127 and 3.130, by

18^0 *-nrK «.i56) where L Is the axial length (eq.3.152).

By combining cqs.3.152 - 3.156, the conductance C of a baffle as shown In Figure 3.15, is calculated aa

N/2

C - 2 I Cn (3.157)

The application of eq.3.157 to,a baffle of D - 25 CM „

N - 10 plates, bent at an angle of 90*, (y - 45*) and l1 • 5 ci, lsada to a conductance far air at rooat tesperature of 2400 liter/sec.

Conductance of baffles with concentric platea

Consider a baffle constituted of concentric plates aa shown in Fig•3.15. The conductance of each annular opening left between the concentric plates, can be calculated by using eq.3.118. IX-

Tlg.i.li. - UllU vita eccentric pl«ti»

HM uaulu* aumfcar a, (froa, CIM cantar), naa m lunar radluai

r, - np + t t» T (3.131) aad aa aatar radlua

r - (a + I) f + ( t| T ».«»> thua tin pariaatax la

Bn - 1* [p(2a + 1) + it t( Y] (3.160) and the cross section

1 An - 2* 2 ° p - np[p(2n + 1) + 21.tg Y] (3.161)

The conductance Is given by (3.118)

(3.162)

Di ri The factor K (Table 3.6) depends on the value of rr~ » — . on * D0 r The value r,/r , varies from 0 to 1/2 for the first annulus, trem 1/2 to 2/3 for the second, from 2/3 to 3/4 for the third, etc. According to Table 3,6, K varies in such a way that for each annulus, it can be assumed as varying linearly, i.e.

(3.163) &nL

Thus the quantity in the integral can be expressed as:

2ir[p(2n + 1) + 2*.tg Y!

KonAn K + 6n § T* P*(*<2n + l) + 21't« Y)

2 p (an + Bn £) [p(2n + 1) + 2t.tg vj

It results that:

2 JK A " pC2n + 1) + 2.1 '« r 2 TST t p [2 o tg -j^pCn + D] + n Y «n «.r

(3.1«) ttfc.

4 2 B^.tg T-l - Sn p<2n + 1) L ^['4^> (3.166)

In order Co account for the ell)OV,lnstemd of .1 , the effecclTe longth Z. Is to be considered

2.(6 K (180 - 2 T)P (3.1ST) Tir

• r - r. * p * r*.o + •rt 'l ° 4

The total conductance of the baffle 1» the aua of tha tUMauctaneas Si- J.45, Holacular flow - Saal Interface

Katfi hae given a eodel for the sachenlea of tha aeellng proceae between t*#o eurfacea which are coapxawad oo each other. In thia •odel tha rooghneae of the ••alios eurfacee la conaidexed being cenaELtuted of flat equlleteral pyramid* which penetrate Into the oppoaloa curfaee or axe flattened and leave hetwaen then leakage path*, aa abom In fig.3.16.

The groove between two adjacent prreeide la celled unit groove. The total conductance of the aeal la regarded aa the xeenlt of the aerlea-parallel connection of all the unit groovea.

the eroea aeetion of the unit groove varlaa elong the groove aa ahown In Fig.3.17. Ihc unit groove conalata of two parta connected In

A. loth and A. Aellanl -'Trene. 3rd. Internet. Vacuum Coagxeaa, Vergaaon freae, 1966, p.181 - 188. :,—,

Fig.3.16. - Inturfacn scaling processes, a) by tnterptnetratlon; h) by flattening.

series. The first part (between fronts 1 and 3 ) has a profile changing with the distance L . The second part of the unit groove (between fronts 3 and 4, Fig.3.17) has a constant cross section. If C- la the conductance of the first part of the groove, nnd C„ la that of the second part, the conductance of the unit groove of length I , will be given by

*T'MJ (3.16B) #inc« Che length £ of the unit groove, is formed by the eerles connection of two parts C_ end of two parts C« •

Tot part 1 of the groove the perimeter B. , and the cross section area A- are given by;

1 tg a x (cosa J

i A2

*{* + «&" O.170) • and the conductance ie

—s* dL 4 *• casaJ J4 .'Eor.part 2 of the groove the perimeter is

2A B7 "T^T f1 +^Srl '03.172) 2 C£ a ( CC-SCt/ and -titer ^JJMS ^secttaa Area.:

(3.173) hence • the xonductauoav >'€•- of this p»rt Is given'by J

•* "2 4 ^h^^-r) .3Jie,cood»«ait»ge of £he txtit ..groove will b* £eq,3.166}:

A C. -^ v... **A , ^-^ '(3.175, "t T "av !*(l^^b) !-0.36^]

'It waa found that the total conductance of a contact aea'l iia

" ^TTT Cl (3.176) «n s

••aling Interface* HMrtfon: k 3 A2 f-2E» C-±T -i! i-Si i-r- (3.177) 3 "tag] •!"£> i-o.*£] A- The penetration j- was found to be a function of the tightening pressor* P , sod the sealing factor R of the gasket material:

j± - e~r/* (3.178)

Equations 3.178, and 3.177, permitted to derive the basic equation of the sealing process:

1/2 ,. .* -3P/a e-i.».»'lg„4 (V) gT^rnf^;,,^ «.179,

wierc A la the peak to valley height of the initial surface roughness.

3.5. Aaslytlco-etatietical calculation of conductances.

Davis, Levenaon and MiHeron , used the Konte-Carlo calculational method, for determining the conductance of simple and complex shapes. To sake such calculations, individual historiee of moleculee entering the model are generated from • act of random numbers. When enough histories have been generated to satisfy the accuracy required^ the calculation is terminated•

The entering molecule is followed over its probable path. At each collision with the wall the molecule is assumed to be stopped end promptly reemitted. The molecule is then assigned random numbers to spsclfy the velocity and direction after leaving the vail. The

* D.H. Davis, J. Appl. Phys., 31, 1169 (I960) L.L. Levsnson, H. Mllleron end D.H. Davis, Vacuum Symp. Trans. (I960), Pergsmon Press, 1961 p.372, and Le Tide, IS, 42 (1963). - 143 -

•election of direction is based upon Lambert's lay of emission, i.e. the molecules leaving a unit area of the vail are distributed according to:

Ifl/In - uose (3.180)

where IQ is the tumber of molecules leaving per second in a direction at an angle 9 with respect to the normal to the surface, and I is the total number of molecules leaving the sarfac?' per second, the history of the molecule is followed until it either leave the geometry or return to the entrance opening.

Davis. Levenson and Milleron, use the conductance C of the o aperture to Che geometrical configuration being investigated, as their reference. The computed and measured.conductance C , is related to

C t hy the probability factor P

C/C - P (3.181) .o r

Ihe assumptions mad* in the calculations and the experimental conditions provided ace: a) Steady molecular flow exists., b) Molecules enter she inlet aperture uniformly distributed over its surface. c) The geometries unter study connect effectively large volumes. d) The probability of the-molecules entering a solid angle is proportional to the cosine of the angle to the normal to the surface of the opening. e) The walls are microscopically rough, so that molecules are difusely reflected according to the cosine law.

Consider a tube with two openings of areas A. and A. » through which particles are diffusing froa Infinite volumes at a net rate of N particles per second under: steady-stats conditions. If f, and 4_ are the numbers of molecules striking unit area per second at A. - 144 -

m€ *~ thee the nuaber of aoltculta par aecoad entering th« tuba at arlflca 1 ee-4 orifice 2 an •^ «nd 4^ . Lac P^ ba the fcaaaallity chat a particle entering, orifice 1 will leave through 2,

aW Pr2 'ha probability thac a particla entering orifice 2 will laava car—g> 1. Tba equation for the aac flow of partlclaa ia then

?r Pr M *1 *1 l - *2 *2 2 " (3,182)

HMB •. IS canal to •- and Che system la Isothermsl, K - 0 , thus

^ PC2 - ^ Pr2 (3.183)

Efutiot 3.1S3 abows chat tb* probability Pr for a aolecel* txaaaertieion la dependent not only on the geometry of the pipe, btfC

also on the area of the orifice with which Pr ia seaodated. W&en the orifices at each end of a pipe have equal areaa (Independent, of their shape),

Pri - Prz - Pr (3.184)

If eq.2.46 la applied, aq.3.182 can be written

7 *a»l "l *1 rn - X »«v2 H h. *'i " « <3-I85>

Slnc« j i A Is tha volua* flow rat* of taa * " 4r (aq.2.A8), through an orlflca of aiaa A , aquation 3.189 la wrlcttn

p n s p °1 *1 'l " 2 2 t2 " * (3.186)

Uhan n, la equal to n, , than H • 0 , and

s p (3.187) l 'ti • »2 r2 Froa eqa.3.186 and 3,187, It results that

sipn * S2F'2 " s " H^-q " c "-"«

The quantity C is the conductance of the systea. Ik* Monte

Carl? aethod uses eq.3.182 by assuring $2 - 0 . Let +.A, • L the nuaber of aarticles that enter Che geometry per unit else through

opening 1£ then this Method counts K_ and H to find Fr. • H/lL -

The first geom&try investigated, is one vhi.cn can also be determined by simple calculations. It is that of a tube of circular cross section, where the conductance can be calculated by the equation 3.108. The

value of C is in this case given by eq.3.74. The value of Pr , is in this case

C 3.61 DJ 1 C 2.86 L + 1.33 D 0 „2

4 S 1 (3.189) 3 i + *» Wl

The Monta-Carlo computation, and experimental results shown in Fig.3.18, are in good agreeaent with eq.3.189. Tha second geometry investigated by Davis, Lavenson and Hillaroa, la a 90*-elbow (rig.3.19). The Monta-Carlo computation and the experimental results show that the conductance of the elbow do not differ significantly froa

-. bsose»o£< *'»ta#ight tube-{•«*'««tr3^13&>. Results Tor'thi crtrtllaii«T?TS'p>ob'ibilrCy"of' a cylih'driCAl'InnUlUS are shown in Fig. 3.20, Equation 3.72 bacoaas for an annulua: - 144 -

flf.3nlf< - trmml*mXom probability for tu^es with circular cross •SCtloft*

U.JW)

(3.W1)

• . iL • 341 - J i.1 j i' to <3.»M) r r'it XB V * l.H W2 - »|1 • - 1»7 -

rig.3.19. - Tr«naal«alon probability for a 90* albou.

Valval of rT coaputtd by the Moot* Carlo naC.iod, an ahovn la Fig.3.20. Ttia raaulta for two noaatrlai coaaonly mad la optical bafflaa in ahvon In rig.3.21 an) 3.22. It can ba atari that tha traaaalaalon probability Pr , doaa not lncraaaa aignlglcaotly for gaoaatrlaa havtmf ratio* A/B graatar than S. Furthanora, tha 60* angla la both thaaa caaaa la ahown to hava battar conductanca propartlaa thaa tha 43* and 30*.anglaa. Tha chavton gaoaatry haa approxuacaly half tha valaa of Pr , for eorraapoodini valuat of angla and A/1 • -ltt-

;t .«

Flf.3.20. - IraaaalMloai probability for aa armular *lpa.

Ttaa»»laalo» arobabllltT rc , haa a •MUMM wlua of 0.28 (71|.3.22) for an apartara Out io eoaplataly covaraa with tho bait chavron arraagoaaat. la a practical caaa cooling tubaa for liquid nltrogan or •char rafrlfaraat will ba attachad to tha bafflaa with tha r«iult that tka affactlva arta will ba aoaawhat raducad. Coealdarlng thaaa factor", a raallatle ralwa sf tT - 0.1 , raaulta for a carafulljr daalfaad dwrroa baffla. •-'—• /Al L

Fig.3.21. - Pr Ear straight baffla plate*.

Flj.3.22. - Pr for baffle plataa with albsw (diavrtm). - ISO -

Da mdu fax aaetaer |oaa*ary at* shown 1m Fig.1.23. This aaaaacry mftma cmHinUi lyl la traaaalasloa probability fr , mt tan HIM ektelaaele fram tba ckeraa type tsoaetry.

Vlg.3.23. - f for straight cylinder with tvo restricted ends and circular blocking place.

rig.3.24. - Ft as la Ma.3.23, In diffusion puap s?stea. - 151 -

Furthermore, the blocking place between the two openings provide* a possibility of using this corfiguration In » baffle-valve combination. Ihla allows to avoid t le additional impedance that would lie brought Into the system "by addln*. a valve la series. The valve action can I>e obtained by moving the blocking plate to either end of the tube and sealing it over an opening.

Figure 3.24 shows tie values of P for various arrangements of this geometry ID a diffusion pump system. Case A(Fig.3.24) is the same geometry as in Fig.3.23 and case B shows the probability ? for the particles to pass through the annulus betueen the jet-cap cover of the diffusion pump and the edge of the crap opening. Case C (Fig.3.24) simulates the use of this baffle geometry on an oil diffusion pump.

Figure 3.25 shows results for another geometry which offers possibilities as a baffle-valve. The effect of using this geometry In a diffusion pump la shown In Fig 3.26. Tha value of P reaches a maximum for a value H/L - 0.26.

71g.3.23. - * - cn small and of straight ay Under with one restricted . I and circular blocking plate. - 152 -

0.4 . '( >=tj ' ' ' - J>X i n >\ > \ : i °* -/ -tdu ^ : "1 P-HTFI •SMHJHSMiS.. - W ssa~- : o U-• A • A ' i ' 1

Mg.3.2t. - OMMCiy fro» Fig.3.25, «>

TIM nault* for another gooMtrgr ttut my b« v«4 M * k»ffl«- yilm coablaatlou «• ohoim In rij.3.27.

tKN-MIM• .<*M «.» «. U wmn ctM* K*H

G..Si Q i •.I*

«VtH> |1>W WW CM¥—I »»t

rif.3.27. - Pr for bulged albow g*oMtrl>». - 153 -

lb* circular blocking plate has the asm* diameter as the orifices. Ihls plat*, in principle, can be swung to cover either of the orifice*. The experimental results for this geometry show that high valoes of

Pc can be obtained for values of W/D > 1.3. The variation of Pr with the arrangement of this geometry in a diffusion pump is shown in Fig.3.27. A bulged elbow containing a chevron array la also shown in fit.3.27.

From Fig.3.28 It can be seen that the bulged end cubic elbows

hav* approximately the same values of Pr * for similar values of V/D.

cuaie tiao* •VD V [WOMfNTM LOII© t-0t> 0.*» CUilC CLSOa WITH JCT CAP 0.1«

I.IO

OWC HIM M MffUKM NW Gi :::: D.I*

Fig.3.28. - Pr for cubic elbow geometries.

Figure 3.29 ahowa the average number of collisions In 10** geometries that require a Minimum of one collision (optically opaque-*, as It result* from the Mont* Carlo calculation*. It can be teen that this number '.* at least 4. - 134 -

ftg.3.29. - Average number of.collision* for transmitted *oleculsa in so*e geometries.

3.6. Intcracd1,ate flow 3.61. Knudscn'a equation The conductance of a tube in viscous reglae (high pressures)* la directly proportional (eq,3.52) to the average pressure p" , and therefore at ? - 0 the conductance should fall to zero. Equation 3.57 shows that as the pressure decreases, an additional term C_D should bo added• Ifaus the conductance is given bv the general equation:

3 c,D* P + C,D

c . -A k- (3.193)

Knudsen (1909) gave to this equation the fom

„ _ „* , tin * T,X'2 l + Is ij r, .3 1 + 1-"fe] r , ul/I 1 + 1.10 X lO"* (f) **• t,-J81. 103 £ T •,- 2 <3-l«> 1 + 1.36 x 10 [&] K

At small values o£ the preaaure P , the tans in the friction in cq.3.196 become amall compared with unity, ao that

J c2 - 3.81 i 10 gj C3.197)

If this relation is compared with that resulting fro* eq.3.57,

/2 c2-Te(fW ¥-—oMr¥ '»•«

it results hat £ • 0.74. Ibis means that at low pressures 741 of the molecules are adsorbed and reemltted randomly, and 26X are specularly reflected.

At high pressures eq.3.196, gives

(3.199)

Coopering eq.3.199 with 3.57. it results that f - 0.85, Xhudssn'a results * jply that the fraction of molecules adsorbed and retmitted, changes slowly in the intermediate region. Equation 3.194 gives the conductance in any regime, assuming that over the whole length of the tube L . the regime is the same (molecular* intermediate or viscous).

3.62. The minimum conductance Equation 3.194 can be written in the form

(3.200) -1» -

'-* n a - 3.27 x 10'* »• 3-« (!)

= - 0.147 (f)1" d - 0.181 |3|

C (Utn/He); F (Torr); D (ca); L (c»); n (polao) M (j) , t <*K>.

Uff«r*ntl«tlnc aq.3.200 with raspact to x and aattlai th« rasultaot aquation «qual to aero tbo valua of x «t which y tui a adnlwoai valpaa la determined; this ia

thus

and

fll1/2 ?mSa D - 5.47 gj nTBrr.Oi (3.203)

Froa eqa.2.70, 2.IB, 2.19 ud 2.42 it reaulta that

H - 0.117 P ^J. * (3.204)

and froa eqa.3.204 and 3.203( we hava

-r-B--0.63 (3.205)

Slnea for air at rooa teaperatura (aq.2.58)

3 x . ?»v>~ *alr F

It raaalta that for air J P-ln D - 0.63 x 5 JC 10 * - 3.15 x 10 Torr.cm (3.206)

According to eq.3.205 Che ainiaua conductance occurs when the ••in free path of the Molecule* la 1.57 tlxas to* disaster of cbe tube. For X larger then this value, the conductance Increases

asymptotically towards that given for •olr_iflow (eq.3.93), and for A. values leas than A , the conductance increase* with ineree- mln sing pressure and tends to becoae proportional to the pressure in the viscous .flow range (eq.3.53). This evolution is shown in Fig.3.30.

3*63. The transition pressure In equation 3.58, the transition pressure P is defined as — k the value of the pressure for which the viscous ten C.P D in 3 eq.3.193, Is equal to the non viscous ten CJ> . In the notation of eq.3.200 this swans

ax-bf^ (3.207) 1 + dx from which 5M (be - a] ± f(bc - a)2 + 4 ahdl > (3.:

By using the positive sign (the negative is Meaningless) and substituting the values for x , a , b , c , d as specified for aq.3.200, it results that the transition pressure is given by

Pt D - 95.7 gj n Torr.c* (3.209)

By using eq.3.204, we have

7- - 11.1 (3.210) *t and for air at roam, temperature (eq.2.58), we have

PL.D - 11.1 x 5 x 10~3 - 5.55 x 10"2 Tbrr.ds (3.2U) Thla meama that for a tub* of D - 1 cm the transition praaaura la

2 Ft - S.S x 10" Tbrr.

according to tba condition expressed in *q.3*267 the transition preeeure Indndas a mlxtura of viacoue and non viscous flow* where both ara sigplfleant. As it can b« aaan In Fig,3.30, tbia point la situated somehow in tha middle of the range of intermediate flow.

3.64. limits of tba intermediate reuse The limits of tb* intaraadiata range can be considered as being thoaa where the contribution of one of tha flow condiclane predominatec, e.g. where the contribution of one of then ia an order of magnitude more important than that of the other.

Therefore the upper limit of the intermediate ranee, i.e. that above which the flow can be considered viscous, is given by

ax- «>bf-rg 0.212) thus

2 2ad AlO be - a) ± J (10 be - a) + 40 abd] ? (3.213)

using again the positive sign, and the values of x, a, b, c, d (Cq.3.200), It results chat the upper limit of the Intermediate rente la given by

IT}1/2

Pa D - M2 ffl ntorr.cm (3.214) and (aq.3.204)

f- • 111 (3.115) *u and for air at room tempereture (eq.2.5B):

? D • 111 x 5 x 10"3 - 5.5 Thia Mann that for plpea of D « 2 ca the flow can ba cooaldetad Tlacous, tor praaaura P £•""*-?-• 2.8 x 10 Torr. (aae Table 3.1).

Tha lower llalt of tha inter—diatc ranaa l.a. that below which th* flow caa ba conaldared wolecular, la given by

1 + dx which gives (elailarlv to eq.3.213):

P11/2 P, D - 10 \~\(H) nlorr.nTm-c c (3.218)

f- - 1.1 (3.219) and for air at room teaperature

1,1'lliii 10"3 - 5.5 x 10"3 Torr.cai (3.220)

From 3.220, 3.211, and 3.216, it results that

P - 10 r and Pt • 0.1 t . thus the interaediate range extends on two ordexa of aaanituda of the praaaure. A coaparlaon ia ehon in Fig. 3.30.

3.65. Canaral aquation of flow

Equation 3.194 can be written

C - C^.J (3.221) where C ia the conductance >r aolecular flow (eq.3.92):

1/2 , r2» R T, -3 160 -

i» CM .It. _d jjjjl/2 £

J - 7s* •-•:—Li/2 - <3-223> where C Is the conductance for viscous flow (eg..3.52)

c,-iih;f7 <3-22«> In CGS unit*.

For air at 20*C. the value of J becomes?

\2 0.225) 1 + 316 DP

Figure 3.30 shows the values of J for air at 20*C la * log- log diagram. On the aame diagram the various PD values, and D/X Tallies are also plotted. In this diagram it can be seen that J-l for vary low pressures, drops slowly to J - 0.96 for P_. D , increases to J - 1.7 at P D, and further Increases to J - 9 at PD, The diagram also shows chat from a constant value J-l for the eolecular range where the conductance is independent of the pressure, the value of J tends at high pressures to be proportional to the pressure (viscous flow).

3.66. The molecular-rlscous. Intersection point As it can be seen In Pig.3*30 the line representing the viscous flow intersects that of molecular flow at a point 1 . loth has shown that the position of this point Is specific for the kind of gas and its temperature. The molecular-viscous intersection point 1 » corresponds to c, " c„ » where C is the conductance for viscous flow (eq.3.224) and C is that for molecular flow (eq.3.222). Therefore O* *• T-* * Fig. 3.30. - J as a function of FD C_ 128 tl |2» RTJ

t T-,1/2 ft,*!"*r FtD « 17 n l-J- In CGS unit. (3.226)

1/2 (SP Torr.ia (3.227)

I by Ming eq.3.204:

7- - 13.5 W««V

For ME («q.2.58) at 20*Cs

?1° Figure 3.30 shows point 1 at thia value. Roth expre?a*«/ th.«. whole range of molecular - intermediate - viseoua flow, in term* oif the P.D value, with the aid of the ratio

6 -^p~ (3.230)

(PD)t By Including eq.3.230, In sq.3.223 the factor J become*

equation which 1* valid for any fas at any temperature.

Figure 3.31 ehowe the plot of J aa a function of FD for.air and helium (according to eq.3.223), aa well aa a acale of 5 . It can be seenjtbet the same & acale la valid fox all the gases, thua for 6 - . • 1 j J - 1.82 both for air and helium. Sine* for SSSS S S S 8 <"*

Flg.3.31. -Hi ill (uoctloo of t -164-

_ -2 air, t\D • (.7 i 10 , t • 2 Hill correspond to

Mine at vklcfc J • 2.U.

X For halleei ?±D - 1.9 x 10" (.q.3.227), ehu« for 4-2 » • 2 x 1.9 x 10"1 - 3.8 x 10"1 Torr.oa. At this velue (rig.3.31) the J for the is alio J • 2.81 .

3.C7. Istaarstss asaation of flow Eeuationa 3.194, 3.223. 3.225 and 3.231, are velid only In eases where the flow is of the eaae kind (molecular, intaraediate, viscous) over the satire length of th« tab*. If tin cut is not tus, tee ••nations can be uaed fot each portion of tha tube where the flow regfant Is the seat. Considering a short portion of th« tuba where the variation of the arassure is dP, aqs. 3.221, 3.230, and 3.231, give the through pat aa

Q.dX - Cm U dP - Cm 7± IS + * * |* °\ d« (3.232>

17 latsgrstlni eq.3.232 it results:

2 •p^r - 4" + S « + 9 s 10"3 In C 1 + 21«) (3.233) lsstaad of the nonlntsgraced Talus which results from as.,3.231, vhieh is .

sanations 3.233 and 3.234 are plotted on Fig.3.32, on which It can be sees that while the •olaeslar viscous lntaraseetlon point corrasponds to o - 1 uslag nonintsgrated valoae, this point appaara at o • 2 if integrated values srs considered. Since esaaurad Taluai of 163

Hf-3.32. - Q/C "Pj u a function of 6 throughput alvsya reflect Integrated Taluaa» eaparlagntallT obtained

data ehould b« raanttJ only with «o.3.233. I.e. «t - 2 should be uaad. - 166 -

Basse OD aq.3.230, wo obtain F P

?t - f- - -^ (3.235)

where V. Is the average pressure at point 1 (nonintegrated), P. is the Inlet pressure (nonicitegratcd; outlet pressure negligible),

and P.1 is the effective inlet pressure (integrated).

Froa eq.3.226 and 2.71, It results that

PD - 3.1 x l

•there P-, is the inlet pressure (the outlet pressure being negligible). Fro* eq.3.237, and 3.222 it results that the position of the molecular^: viscous intersection point i is gtrea by

LQ, - -f- (3.238) P li

(3.239)

Figure 3.33 shows the values of the intersection constant I , Cor various gases as a function of their temperature.

By plotting LQ (Sttpr>lit,«r ^ ( M , function of the effective Inlet pressure P. and representing tha lines of molecular flow (slope 1), those of viscous flow (slope 2) and the lines on which the molecular-viscous Intersection-point should be (eq,3.238), a graph as that shown in Fig.3.34 is obtained. In ths example shown on Fig.3.34 for a capillary of diameter D - 10 cm, the intersection point for air lilt t, which corresponds to P . - 260 Torr. For helium the same pipe has the intersection point at b, (about 900 Torr). I IIMI JIIIM llll IIU^IU-

(jiXH-uo-ji^sj-) I

(."" "iiTK>r)» ' s - p s Fig. 3.34. - 14 w * function of t^ - 168 -

For hydrogen the Interaction point is at e. . The relative positions of the cumi illustrate tht ratios which exiat between the conductances of a duct for flow of various gasas (Table 3.8). It cm be seen tbmt while in Molecular flow the throughput (conductance) for helium is higher that that for air, in viscous flow the opposite la true. For COz. the conductance ratio C/gasC ai..Ir s < 1 'in mole- cular flow, and > 1 in viscous flow.

Table 3.8. - Conductance for various gases

C Gas V *r Molecular Viscous

hvdrogen 3.78 2.1

! Heliu: 2.67 0.93

Watt-r vapSur 1.26 1.9

Argon 0.85 0.82

C02 0.81 1.30

Mercury vapour 0.38 -

3.7. Calculation, of vacuum systems 2.71. Sources of gas in vacuum systems A vacuum system is the assembly of the components used to obtain to measure and to maintain the vacuum in a chamber» or device. Any vacuum system ±m made up of a pump (or pumps), gauges and pipes connecting them together. The system contains also valves* temps, motion seals, electric lead-throughs, stc. Figure 3*35 shows a typical vacuum system. - 169 -

In order to express the behaviour of a vacuus system, the various •ourcM of gas existing in it mat be considered* as being at any moment in equilibrium with the pumping action of the pumps on the «ystem.

Fig.3.35. - Vacuum system. 1) rotary (backing) pimp; 2) moisture trap with window; 3) air admittance valve; 4) throttling valve; 5) backing line; 6) roughing valve; 7) roughing line; 6) Pirani gauge; 9) backing valva; 10) diffusion pump; 11) baffle valve; 12) vacuum chamber; 13) electric lead-through; 14) ahaft seal; 15) Penning gauge; 16) window. It can be considered that the source of gas in a vacuum avatam are: a) The gas molecules of the Initial atmosphere ^enclosed in the •yataau b) The gas which paoetratee Into th* systss. ma a result of

laekag* (QL) c) Tb« CM provening fro* th* outgasaing of th* p*t«ri*la In th*

aystta (QD) d) lb* g«* (or vapour*) resulting from the vapour pr**sur* of the •atsrlals (Q^, •) Th* gaa entering the systea by p*neation through walls,

window* (Qp).

Th* quantities of gaa resulting ftftl source* b to c *r* functions of th* construction of the systea. For th* present discussion v* only consider, the totality of th* gas resulting fro* thai* sources (Qg):

% • % * % * %+ qi (3-JW>

as being constant for tht tiae interval which we consider.

3.72. rusjpdpwn in tb* viscous rang* We aasuae that the pumping speed of the pumps S ia * constant in the range considered. According to cq;.3.2S» the puaping apead obtained through the conductance C of th* pipes connecting th*. piajp to th* chamber la

Sn C p If the pressure le enough high, for viscous flow (Teble 3.1), th« conductance of the pipe Is given by equations of the for. of •q.3.52: n* ? + P c-isr^rH •"-«-*-* <3-?w

where £ • ?2fl ; P is th* pressure in the vessel and P fa th* preeaure at the Inlet to the puap. ' substituting eq.3.242 In 3.241, ve have F +P S E • - 2 (3.243)

and the throughput (eq.3.2A):

• ? + PP s E ~ HP q . p.s . p -B 3- - - V» 7T (3.244) P + - ?P dt s + E p

Since ? is also * function of P we hare to write P

p S v 3 245 " - , P - - f <- > buid on Che fact that the throughput is the same in the chamber and *t the pump. Fron eg.3.245, it-results that

(3.246)

By introducing this value in eq.3.244, we obtain fens'*?£••'-•-,2 ,2 By putting A - |i B " [fj and solving eq.3.247 we have

2 2 1/2 dP . - A ± CA + 4BP ) ., 243) dt - 2B «••»<>) Sine* the praMura deereaaea In tlat only the loluclaa rj~ < 0 p dc In rani, thus dP 21 dc (.3.11,9) A - (A2 + 4B?2)1'2

- 2B A + <*' + 4f2>V2 IP - dt - 4BP2 By integrating eq.3.249 we obtain + «-!?** m. - tn s r +K!(3. 250)

For t • 0 , P - P (Initial pressure), and

K - <5

— . T _ "S - — n I + V E(P - P4) 1 P

C3.25J) Sp LP + K^ + P2!1'2

This equation i* plotted for air in Pig.3.36, by using 'as # parameter the value r- , E and considering P. - 10* dyne/c*2 (760 Tore), and P - 102 dyne/CM2 -2 (7*6 x 10 Torr), I.e. the pressure range in which usually the flow la viscous.

Fig.3.36. - Tine required to decrease the pressure from 760 Torr to -2 7.6 x 10 Torr| In a volume* V , connected by a pipe of diameter D (en) and length L (en) to a punp of pusping speed S

It is Interesting to mention that if the pump 1B connected directly to the vessel, L - 0 » thus E - - , equation 3.252 becomes, p equation which also appears M th* puaetowB tlaa In aol«cular flow where the conductance Is not « function of th* frunm.

3*73. ruapdown in th* aoleculer rana* The puapdown in th* Molecular rase* *» llalt*d by the equlll- brluai between the gas load end puaping ap*«4 in the puap itself, aa well *» by this equilibrium in the vaevn chaaber.

The gas load Q of the puap itaalf la constituted by Che leakage into the puap, and the baefcstresadne; of th* eassilng fluid. If the theoretical puaping speed of the puaer - is S , the throughput will be

« " *t PP " % ' St % [l " S^r] ».253) where P is the pressure at the Inlet of the puap. the lowest pressure of the puap F • will be obtained when Q - 0 , thus (eq..3.253)

(3.254)

la given by

<3 255 vt-^-ftj-^-V1 L BP' - >

At the vacuus chamber, the pumping speed li S« & + c * *ni* the gaa load In the chaabcr li Q . D

The throughput at the chamber la

S G - 1?S -

Vrp-s"sVc* »"" fro* 3.255 «nd 3.256, in obtain

'« + <^-Tfcff-r.)st a.2»)

S3 * f «.Z59,

froa which th« tia* rfetuixtd fro* lowering th* pr*ssux* in th* chaab.tr

f rat P£ Co P , !• .a.4 (3.260) and the piuiun ruchad aftax Cla* t ia: v r i V\ -&+ r> a* l ^ ? " |?i - Po "{ 1+ C> Sj * <+ Po+ <+1 ? S* C3'261) If the conductance C • in *arr laraa. and toe ultlnate pressure due to the gas load 1.. P' • Qg/S^ • than. eee.3.260, and 3.261 can be written:

p. - r - r t • {- *• .* .T . ..» C3.2«2) -17*-

Ihaa • tufccfli C tannacea tha puaa to tha ehaabar, tha ultlaata •namxa Js tha chaaaar dua ta tha |aa load la

.i._3S.iC , i^iSP_Sa. (3.264) - - - F - (^Vcj P

Istroaudat aq.3.264 In 3.260, tia ban

•ih^"^^ (3.265) F-Po-(i-^)ro

ami for a system where the final pressure of the puap PQ Is such less then P ,

P„ - P r(1 + §|toF^ra (3-266) t v ' u sed Independent of P » It results that t-%^*1]ln^T^ <3-26'>

f2_C P " ffi " V • CSP + ^ V "* PM <3.26fiJ

Ibis equation shows that after a Varr long pumping tin*, the pressure tends towards the ultimate pressure P , determined (sq. 3.264) by the gss load- Thus eq.3.265 describes the transient

P • Pj a (3.269) aa wall tm tha ataadj^.aexta (3.270)

•If 3-"- - Fwpdwa and ataaay atata lUa latar em nault in a «onatant altlaata praaaaxa (Fig.3.37) l£ Qg - const

x - V/S (3.271)

Tl* "lulf Hf." or th* tiat to mnfc m naif of the lalttal •reuan. Is (lv«t by

Tl/2 " °Mi t (3.272)

iAll« th* tia* raqolrW to none tfao ptosaurc by m doeado la

T1/W - 2.3 | (3.273)

3 74. StwAr stats with distributed ass load

The steady state pressure In a vacuum chaaber 1* given by tha siapla relation eq.3.270. If tha gas load la distributed along tha pipe.a caee vhich appears due to the outsassing of the surface, than tha steady state i* characterized by « pressure gradient along the pipe,

Coaalder that tha pusp (Fig.3-38 a) evacuates a pipe of coBdnetsace C , closed at tha end. Let tha specific outgassiag rate 2 be q (e.g. Tozr.liter/sec.c* ). The gss land due to an aleaentaxy length dx (Fig. 3.38) will be

- dQ - q . B dx (3.274)

v era B Is the pjriawrsr of the tube cro*s section, and the alno* •vgo shots thst the gas flows towards - x.

The rhrougltput through the length dx Is

Q - C ^ dP (3.275)

thus 2 dQ - CL *•=£ dx (3.276) Px « Po Pump a)

CT

b)

M*. 3.38. - Distributed ga» load

By the aquallty betuaen 3.276 and 3.27* in hart

£±. .sa (3.277) dx2 'CL thus

(3.278) dx CLX**1 Since at tha cloaad, and of tha pipe, x " *-|dxl . 0, It raaulta that K • ^ •and therefore - ISO -

JC..UItU (3.2W

...£,>*U, + I

It dw lal«t of the 9«Mp* •**» x * 0 t thm p«Mu» Is

Finally «a flat that tba praaaura «t a cutanea n aloas th* ply* ia

2 • x f,(i e JO tj (3.2S2) .-p vhieb ahawa that tha dlatrlbatlon la parabolic aalnc aazlaoai at tha claaad and, !••• ft"«,l(jr*fe] (3.283) Ttaa praaaara drop la:

*«-*."'»l(5-^ (1.284)

a . P .SLA (3.285) I o 2C

^--^t^-h'*^ (3.2M) in shlch the constants ace determinated by

XG ' -r-««lt- + ffl «-28» l -«(H From 3.286, 3.287 and 3.288, we have

1 o, (2C S J C (3.289) r and the pressure distribution in the pipe will be qBL + H. + % ^ [2c spJ c J

Thla shows that even ualng a pump with a very lar^e pimping speed

From eqs.3.287

3.75. nomographic calculation of conductances and pumpdown Many nomograms vera built for the calculation of the conductances, pumping apaad and punpdown times. We will show here just the most typical one*, built for the evaluation of the system In viscous flow, molecular How and the intermediate Tangs respectively.

For the evaluation of the pumping speed in viscous flow, the Harries* nomogram shown in Fig,3.39 is the most known one,

V, Harries, Chen. Ing. Techn. 21, 139 (1949). - 182 -

P(lmi *4 ^xM W»» W

Fig.3.39. - Noaogra* for evaluation of pumping system In viscous flow, (Air, 20»C). This graph Is based on equation 3.54, and 3.29. It shows how eh* pipe (with a die** tar 6, and length ,1) must be dlneneiooed for a puep with a pueping speed 5 at Inlet pressure P » so that the puaplng speed at the vacuum chaaber be 0.7 S. If three of the factor* s, F, 1, d are known, the fourth can be found froa the noaogra* (Fig.3.39). The exanple shows that the line joining 1 • 9 * and d • 5 cm * Intersects the A seals at a given point. For an Inlet pressure of P • 0.1 Torr * a second straight line extended from 0.1 Tort, through the point found by the first Una on seele At shows that the •ajdaxai adnlsalble pumping speed S should be 50 m%T. - 183 -

For tb* evaluation of the conductance of (short) pipes In the Molecular rang*, Delafoase and Mongodin (1961) published the nomo­ gram ahown In Pig.3.40. The aouogxaa Is baaed on eqs.3.94 and 3.104, and Rives tha correlation between the length X. of the pipe, its diaawter £, Its conductance C , and the corresponding correction K for the short pipe* The example shown on the noaograai (Fig.3,40)

Pig. 3.40. - ZtaaograM for determining the conductance of pipaa - aolacular flow, air* 20*0. - 184 -

evaluate* to* required diameter d , of i pipe i » l.S B long, im order to have a conductance of about C • 1000 liter/sec, in the molecular flow range. The line through I - l.S a , and C - 1000 liter/sec cuts scale d , at d - 24 cm. The conductance at the entrance is included by the correction factor K . It is a function of the ratio d/t , and in the example It is given at the intersection of the line t-d . with scale K , where it shows K - 0.83. Thus the real conductance of the pipe considered is

C - 1000 x 0.83 -- 830 liter/sec flhen d is- incsaasad to d - 25L rat , the value 61 fifce eonducftance (dotted line) will be

C - 1200 x 0.82 - 985 liter/sec.

Dclafosse and Mbngotilh $>) also present a nomogram 'fte? die, evaluation of the' conductance• in the intermediate range . This nomogram is shwon in Fig.3.41, and is based on eqs.3.221; 3.94 and 3.225. The examples shown on Vig.3.4l refer to a pipe of length 1 - 2 a, and diameter d - 10 en . The straight line joining these two points, Intersects scale C at C - 60 liter/sec. If the average pressure is p • 4 x 10- 2 Torr, from this point and d - 10 the correction factor J(eq.3.225) results on scale J , J « 6.8. Thus the conductance will be

C - 60 x 6.8 - 408 liter/aec.

The pumpdown time, t , can be evaluated from the nomogram in Fig.3.42. This nomogram is basftd on eq.3.269. The first example - 185 -

Fig.3.41. - Nonogran Cor determining the conductance of pipes of circular cross section, In the whole pressure range (sir, 20*C).

(lover lints) shows that for a valine V - 5000 liter and a pimping 3 V speed of S - 120 m /hr, the tin* constant T - g" - 140 s. If the final pressure to be reached Is P - 10- Torr (Initial pressure 760 Torr), than it results that the required puspdown tine Is t - 0.37 hr . The second exasple Copper lines), show chat If the volume V - 5000 liter has to be evacuated by a puap with a puaping speed I - 700 ar/hr , tkaa tao tlaa conataot which ruulta Is about 2S «c la orriar to teeraaaa tho araaaura from 10~Horr to 10~ Tbrr, thus *,/!, - 100 , It rasalts that tha suapdora tlaa la t • 120 aae.

Hg.3.«2. - Noaograa for datcralnlng tha puapdown tlu (Molecular Clow, alt, 20*C). - 187 -

4. PHISICO-CHEMICAL PHENOKEHA IN VACUUM TECHNIQUES

4.1. Evaporation - condensation

4.11. Vapours In vacuum systems In addition to gas-a, vacuus systems also eon tain vapours. The, name vapour refers to a real gas, when it Is below its critical temperature (see Sec. 2.1).

When a substance is present* some of the molecules near its surface have sufficient kinetic energy to escape into the atmosphere and exist as a gas. Raising the temperature facilitates this process (see Fig. 2.2). If the liquid is in the open, the vapour nolecules rapidly diffuse away from the liquid* and in general produce! what is known as an unsaturated vapour. If the substance is in an enclosed space, the pressure of the vapour will reach a maximum, which depends only upon 'he nature of the substance and the temperature. The vapour is then saturated and its pressure Is the saturated vapour pressure. In this case, a dynamic equilibrium is established, between the number of molecules escaping from the surface (evaporation), and the number of molecules recaptured on the surface (condensation), in which the net number of free molecules in the gaseous state is constant.

Vacuum systems contain saturated as well as unsaturated vapours. All these vapours are maintaining their physical state or changing it according to the pressure - volume - tenperature conditions (see Fig.2.6) existing In the system.

According to these conditions* - Any liquid surface inside the vacuum system is a source of vapour, and as long as any liquid remains in the system, the minimum pressure attainable is the vapour pressure of that liquid at the existing temperature At room temperature the presence of water limits the pressure to about 17 Torr (Table 2.2), while the presence of mercury to about 1 x 10 Torr. - 1M -

- XI the vapour axUting la th« vacuum system la cDapraiaad aa a

r?*ult ef aaaplng or haaallag operationst its preeaura will increase sely tan tka vapour ynuiiri. Further coaorasaIon came vspowr to eaaWtaajM. la this way vsaouro which. In the system art unsaturated mi ef law praesere» will condense. In ,tho puapa or K*fw whtn they at* compressed, la order to avoid the condensation of water vapour la rotary pumps, the gas, .ballast system la used, In which a controlled amount of atmospheric air la admitted in the pump at a gives* stag* of the compression, »o that the pressure of the vapour la not increased above its saturation.

lac compression occuring. on the HeLeod gauge (Fig. 2*76), ceaeeaaaa the vapours, and therefore this instruaent does not aeasuxe accvxately the contribution to the total pressure of any vapour* In the

- A reduction in the temperature of any part of the vacuum system, rosacea the vapour pressure of sny vapours present. This la the principle on which the use of cold traps* refrigerated baffles, and cryogenic puaps is based.

*.12. Vapour pressure and, rate of evaporation

The vapour pressure Pv of a substance is derived froa the Clsuslus - Oapeyron equation:

1 - f <"c - V 3^ «•» where L Is the latent heat of evaporation, J la the Mechanical

7 aqaivalsat of heat thus aquation 4.1 bacoaaa

fa. £ fa (4.3) ' J P„ dl BM ratio It /J • !• aqoal to

|o . 8.314 x 10? «nTfc-U . 1M firfrfcj|0te 4.185 x 10" erg/cal

The latent heat of evaporation can be expressed by

h - L - I.I (4.4) vfaere I. la the latent heat of evaporation at T • 0 „ and I la a constant.

Froa 4.3 and 4.4 it results that

dP (4.5) tfaua In daciaal logarithaa

lo» fT - A - | - C log T (4.6)

A- A72.302

1.987Js x_ 2.30 2 4.575 I Uiually -«•-

lha valaaa aC tkaaa immw an aacaaajaai far • lacfa wtmm at Hmili. taala 4.1 gtoaa (hair Mail Jar tha varlaoa aaoala.

•haa aaaimrtoa adata titwwi tin aalU or lffoU ml pMW l>«ii, tka Tift 1H tTIMWli* * • *» *t**l to tha rata of aaaaaaaatlaa, ta areiaraiaa, to oa..2.M

2 • - 5.M x 10" PT fej • <4.7>

2 aiata V la tfc* rata of maaoratlcn (i/a.ca ), *T la tha WIIBJ •raaaara Clarr), K •alarelar aalfjit: a. la tka atlekUg naafflaloat iiftaai aa tha araaaWlty that aa

tf aalac aaa.4.7 aai 4.6, It raaolta that tha arapsratiaa rata CM ha aaanaaat aa

le« H - A" - f - C'lag I (4.»)

A" - A + las 0-<»»3 + O.S la» H C • 0.5 + C 1 - 1./4.S75.

4.13. Taaaax araaaaia of tha ntloaa aatarlala aaa oataatal k* • lacaa • •# aathira. lhaaa aiaanraaanta aara haaaf. altkar aa tha aixatt i at tha aachailcal affact of tha ajajaaonj, aaactai hf tha i mat', at hr aaaairlaa, tka aiaaatllaa. raU (Mtgkt law ac taarnal) aa* aalaalatlac tka aaaocr ftaaaura fraa aaa.4.7. nana 4.1*4.3 (l*a tha aa-to-iaea oaaaar araaaaro aata, mlrtlai to laals aai Kraatr*.

1.1. kaalg mi D.4. KCaaar, 1CA Urtmt, Joaa, 1*M ao. 215-303. T«blu 4.1.

l.i 10.99 8.07 II 15 8.6.1 Na 10 72 SJ9 II.1H 95J k. 10.2a 4 -4K Kb 10.11 4.08 I:.-H 30.00 C» •>.yi .VKO It.M .to.« 12.40 4U.6B fu 11.96 I4> MH 11 5'J 2 J. 31 *E 11.85 14.27 Au 11.89 17 JK 12.14 13.74 12.44 19.97 BC 12.01 16.47 12.70 21.11 «S ll.t>4 7.65 12.75 20.96 CJ 11.22 8.94 I J.SO 33.80 Sr 10.71 7.K3 12.94 27.72 Bil 10.70 8.76 11.78 19.71 13.59 37.00 Zn 11.63 6.S6 13.07 31.23 Cd 11.56 5.72 I2J3 27JS

B 13.07 29.62 • Dushman, 1st edition or this book. Al 11.79 15.94 t Values or A for the pressure in microns. Sc» 11.94 18.57 Y1 12.43 21.97 U 11.60 20.85 Ce- 13.74 20.10

Ga 11.41 13.84 In 11.23 12.48 Tl 11.07 8.96 C 15.73 40.03 Si 12.72 21.30 Tl 12.50 23.23 Zr 12.3} 30.26 TV 12.32 28.44

Cc 11.71 18.03 Sn 10.88 14.87 M> 10.77 9.71

V 13.07 25.72 Cb' 14.37 40.40 la I3.04 40.21 MEwTOMS/WCtCK* % i a "fr i - & $ » V ff t *e *»

W*OI nttt»>K M MyOH^EKS

HK1. •» 1H.1V."Mi f'^M

Fig.4.1. - Vapour rrc.isurt* of «lcnciitti. - 193

*b \ "a "a ?e \ -s 's "a

Fig.4.2. - Vapour pressure of elements. NCwTODS/METEft'

"g *fi "S T2 % B « *9 fc "s Te

WW 'WSSUftE M »rU0W«Efl£5 "e ^s "e *o •%

5-1

Pig.4.3. - Vapour pressure of elements. The rate of evaporation of various elements is given in Table 4.2.

The vapour pressure curves of various common gases is given in Fig.4.4.

Figure 4.5. shows the vapour pressures of some oils used in diffusion pumps, while Fig.4.6 gives the vapour pressures of aowe cleaning liquids.

4.14, Cryopurapinfl and vacuum, coating Evaporation and condensation phenomena are often complicating the pumpdovm process in vacuum systems, but they are also the basis of vacuum technology applications. Although techniques like freeze- drying and molecular distillation are also based on evaporation, phenomena, we intend to illustrate here the field of applications by techniques representing the use of extreme temperatures, i.e. the cryopmapinfl and the vacuum coating.

Cryopumping Cryogenic pumping is based on the fact that if a surface within a vacuum system is cooled, vapours (gases) will tend to condense upon it, thus reducing the pressure. The ultimate pressure of such a puwp

for a given gas is determined by the vapour pressure Pv at the

temperature Tv of the condenser surface. Since the quantity of gas evaporated from the surfaces of the system at temperature T is

equal to that condensed on the surface at Tv , from eq.4.7 it results

that the ultimate pressure Pu for the particular gas (M) considered is

\ i ii1/2 u s V llj where s and s are the sticking coefficients at temperature T and T i respectively. Since both s and a are cloae to unity, it results that for T - 300"K , and T - 4.2*K (liquid heliuaj), P /P - 8.4. In this case, according to Fig.4.4, for aoat of the gases, T«1.1t 4.2.

DM flm Ret u4 •* »» If ro io-* 10 • 1 10 100 1000

Li 4 4S9-10M i: HI 399 460 53* 62?. 737 W: 6.17-10" 5.93 • 10 ' 5.68 • 10 « S.4I - 10 ' 5.13-ID- 4.84 > 10 * N» 4 2M-*2f /: 131 l»i 238 290 353 437 If: 135-10-' 1.29 • 1U • 1.24 • 10 » 1.18-10 « 1.12-10 » 1X35- 10-* K 4 i: «l III 162 20* 2bfi 341 !«>-»» If: 1.91 • l0-» 1J3 • 10 * l."S • 10 •" I.**- 10 •* 1.57-10* M7-I0* PU> 4 r: «4 133 17b 228 300 W: 194 10' 2.11 «• 10 • 24H IU" 2.5J - 10 • 241 -10 * 2.22-10' C* 4 i: 4* 75 no 152 206 277 If: 3.77 • 10-' J.«l • 10 • 3.44-»0-» J.26-IQ-* 3.07-10 » 2J7-10 * Gi 4 969-1606 i: 942 1032 1142 1272 1427 1622 1ft 1J3 • 10-* 1.29 • 10 » 1.24 • 10 * 1.18-10' 1.1 J - 10 » U7 • 10 » Afi 4 721-JWO /: 757 tn 9-2 1032 1167 1337 If: 1.W-IQ-* 1.112-10" 1.75 • 10" 1.61-10 ' I.DO-IO* 1.51 • 10-• An 4 727~*7 v. 987 10B2 ll'J7 1332 1507 1707 W: 2.31 • 10-' 2J4-I0* 2.14 10 » 2.03- 10' 1.94-10"* 1.84 10 • Be 4 MM-1279 t: W> W ICM2 1212 1367 1567 w: s.it -IQ-» 4.93 • 10* 4.74 • 10 ' 4.55- 10 > 4.33 - KM 4.08- 10 * Me 4 736-1D20 i: 3X7 3)0 82 442 517 612 W: IJ2 • 10* 1.17-10 • J.12 10-» 1.08 -10 * 1.02- ID » 0.97-10-" Ca 4 527-**7 fi - 402 452 .17 592 687 817 IF: 1.42-10' 1.37 -10 • 1.31 • 10 "' 1.2(1-10-' 1.19-10* 1.12 10 » 5r 4 i: 342 3W 45f> 531 623 742 If: 2.20 • 10 ' 2.11 • 10 -• 2.02 • 10.* 1.93 • 10 ' 1.82-10 » I.T1 • ID"* 4 10e0-ll3t r. 417 467 537 617 727 867 *> If: 2.60 -10 * 2.51 - 10" 2.40-10 ' 2.28-10* 2.16-10-» 2.03 • 10~* Z* 4 235-377 r; 2M 246 290 342 405 4(5 If; 2.15 • 10'* 2X17-10" 1.99 - 10-' 1.90- 10-' 1.BI • 10* 1.71 - Ifr* Cd 4 aoo-wo V. 149 182 221 267 321 392 If: 3.01-KM 2.9B-IO* 2.71-KH 266- 10" Z54I0> 2.44 -10>* "t 4 V. -IS -8 16 45 HI 125 W: 5.21 • 10' 5^» • 10 • 4.86-KM 4.63-10* 4.39 I0-1 4.14 10* 4 /: 1M7 1827 1977 2157 2377 2657 • II': 4.33-10-' 4.19 - IO-» 4.03 -10-* 3.89 • 10-* 3.73 10* 3.55 10* Al 4 1137-1 IK /: H2 972 ioe: 1207 1347 1547 w. ».«• io-« 8.39 • 10-' B.2J 10-* 7.18- 10 ' 7,53 • 10 ' 7.10 -10-* Sc sir #: 1051 1161 12X2 1423 (595 HUM V' St>* /: 1249 1362 14'M 1MQ 1833 20J6 u 4 1327-1627 I: 1262 1377 1537 1697 1897 2147 If; ».»I0» 1.69 • 10 • I.MI0' IJ5-10* 1.48' 10 • 1.40 • 10' c* SD* C 1004 1091 1190 1303 1419 1599 N4 4 I9SM279 i: «7 1062 1192 1342 1517 1777 If: 2.00 • 10-' 1.92 • 10 • 1J3 • 10 ' 1.74- I0"« 1.65 • 10 • 1.35 -10 * tia 4 «M»5 v. 737 MI 937 1057 1197 I37Z M': 1.52-10' 1.4* • 10 • 1.40- 10" 1,34- 10" 1.27-10-* 1.22-I0* In 4 727-1075 r: 6TO 747 837 947 1077 1242 If! 2.04 -10 » l.w - \a • I.1W-I0-' 1.79-10-' 1.70 • 10-* f.fl • 10 • Tl 4 /: 412 it* 535 415 713 837 If: 3.19 10' 3.06- 10" 193 • 10 • 2.HO-I0-* 2.A6 • 10 * 2.30 -10-* C 4 2084-2397 i: 1977 2107 2247 2427 2627 2867 If: 4.27 • IO-' 4.14-10' 4.01 • 10 ' 3.89-10-' 3.76-10 * 3.61* 10» SI 4 /: 1177 12H2 1357 1547 1717 1927 H': tU2-10» 7.K4-10' 7.54 • 10 « 7.24-10' 6.93-10' 6.39 10* Ti MoniWHl till 1-123 i: 1321 14JI (558 1703 1177 20*3 W: |.fl| -10 ' 0.98 • 10 ' 0.94 • 10 » 0.90 • 10 • 0.86 • 10 * 0.82-10* Zr 4 i<>7ft iw r: IK.I7 2002 2187 2197 2647 2977 •f: 1.21 • 10 » 1.17 -(0 • 1.12-10 » 1.IW- 10 * 1.03 -10 * O.Qf-10* 1h sn» t: IM6 1831 1999 2I'»6 2431 2713 II: 201 10' 1.94 • 10" 1.86- ID" 1.79- Id * 1.71 • I0-" 1.63 • 10 * Oc 4 1237-1612 /: 1037 1142 1262 1407 15X2 1797 If: 1.37-I01 1.32- 10-' 1.27-10-' 1.21 - 10 < 1.15 10* 1.09 10"*

v («.- Table 4.2. - (coi.t.j - 197 -

Tcmt. F = Ran„e CC) 10' 10 ' ' 10 100 8B2 •ill HM2 rii l)i- 1.87 • 10 ' 1.81) 10 ' 1.72-It ' 1.6- • 10 ' 1.51 10 » 487 551 627 719 8) 3.05 • 10 » 2.9J • 10 ' 2.WJ 10 > 2 n7 10 • K.l - 0 • 1432 LSI US7 1*47 :o.n 1.01 • 10 T am • io • O.M 10 * J.90 10 * 0 87. 0-' 21M :-5s 2?.W 1.16-10 ' 1.08 • 10-' 1.06 iO * . 2397 21K7 2S07 JUt>- tV2 3737 1.52 W 1.47 • JO"' 1.41 - 10 • J.J&-JO * 1 0 • 10 ' 1JH • 10 • 107 no 157 187 2 2 3.3 • 10-' 3.24 10 ' 3.13- 10 • iot to-1 2 2 -10 » 382 427 477 542 67 2.52-10 ' 2.43 10 • 2.35- 10 ' 2.26- 10-* 216-10 " 450 508 57W Ml 72 892 3.14 • 10-* 3.02 10 • 2.H9 • 10 < -.76- 10 • 262-10-» 47 • 10 « 1002 II-.2 1267 1392 15 7 1737 1.15- 10"' 1.H10-" 1.07 10 » 1.03 10 « O^S- 10 •* 0.94 • ID"' 1987 21*7 2377 21.27 2927 3297 1.20- 10-* 1.16- I0-* Lit • 10 » l.« • 10* 1 01 -10-* 0.95 • 10-' 2547 2757 3007 297 3647 1.49* 10" * 1.44 • '0-' 38 • 10"* I.."*-10 ' 126- -Q-* 1*12 IWJ 1737 1927 2137 2447 711- 10* 109 • ifl-» 201 • H>-» l.D>. io-« 1 Ml • 10 * 1.73 -10-'

2.54 - 10-1 261 4.03 • 10

130 • IO"» 767 94 1067 ' 19 • JO 1.34 - 10 1.29 10 124- 0 1.18-10- 2J67 2157 27X7 3057 3397 1.35- 10- 1.50- 10 1.44 • 10 1.38 -10 I I • 10- 1092-1246 1107 1207 1322 1461 1637 1847 1.17- 10" 1.13-10' .09 -10-' 1.0* • 10- 099- 10 ' ass - to-' 1090-1249 1162 1262 1377 1311 1697 •907 1 18-10 1.14 • 10-' 110-10-' 1.06-I0-* 13)1 - 10-1 0.96 tO* 1034-1310 1142 1247 I3S7 1497 1667 1*77 1.19 • 10- 1.15 10' 1.11 10-' 1.06 • I0-1 .01 * 10-' 0.96 • 10-' 1913 2058 *2J0 2431 2666 2946 1.26 IO-» 1.22- 10 ' 1.1*-HH l.» 10 I.M • JO' 1.04 10 * 1587 1T07 1857 2077 2247 2527 1.37 • "0-1 133-10' 1 28 • 10'" 1.23 - 10 1.18- 10 1.12-10-' 1157 1263 1387 1547 1727 1967 1.59 It-' 154-10-' 1.4* -10-' I 41 • 10- US-10' 127- 0-» 2101 2264 24H 2667 2920 7221 1.65 • ID-' 1.50-10' 1.54-10 • 1.48 10 1.42- 10-' 1.36 10 * 1797 1947 2107 2307 2527 2827 1.78 • 10-' 1.72 • 10 « 1,6*-10" 1.60 > 10 LSI • 10-' ' 4* 10-» )«02 1742 1907 2077 :m 2587 1.88 • 10-' i.82-10' 1.73 • tO « 1.68-10 1.52 • I0-» 1.60 10 *

* SD -. Oinbrmn. 1st edition of Ihii book. Fit;. 4,4 - Vapour pressure of conmm gftsea

s mumilBlilJ

i * *sji i «(.4.5 - Vapour prcMuro of oll». g Vapour pressure (torr)

?ig.4.6, - Vapour pressure of solvents. Ik* Mllmm >it»m tka anuiir at aalsealaa rnalaailag ana laavlng tka alt aarfaca iru aaek aacana. la txm aa..2.tt

« - a» (J. nkl) "* - «..%«• u«Y> "* (4.10)

It fnUaaa fm aa..2.15 an* 2.21 that to* throughput is Q - M.kT,

nhcrc A la tkc eras of condaanar aurface. *1*-*«rfm a P by aa..4.9, tha puaplng apand ia obtaiaad aa

S - 3.64 a *fe] (l - j*] litar/aac (4.12)

•aaad on tfala principle, tba cryoaaalc puapa attain ralatlvaly high r—ptag apaaaa, (10* - 106 lltar/aae).

"Ill"I I I'll III Vacnan coating la baaad on tha evaporation of the required ' notarial, and ita aubaeauaot eoadattaatiea oa tba aiibatxata to ka coatad. Tba proceaa la done In a high vacuoa, ao that tba paxtlclaa do aoe collide «lth gee aolaculaa in thalr way bantam enporatloe

figure 4.7 enow tfca basic faaturaa of a vacuum coating plant. Tha aatarlal to aa, evaporated (natal or non-natal) la placed In toe evaporator (a eplral or boat of tungataa, uolybdenua, or tantalum), vklck la heated (In vacua*), up to a teaperature where tba vapour preeeure of the natarlal to ba evaporated la enough high. To obtain nanlaalbl* evaporation ratal (at.4.7), vapour preeaurea -3 -2 of 10 - 10 Ton ara uaaally required, thue tba natarlala hnra to ba heated up to tba tenpereturee corresponding to tbeie vapour praaaarai (Fig. 4.1 - 4*3). Eq. gold have to ba titrdtod CO About lttO'C, to obtain P - 10 "2 Tore.

EWOftA'OR

.^-L-5tcriON NEonEME 6ASOT 1

ItHrOODCTKAP

TOSWH 1 *«•« tOTMT PUMP PUMP I

Fig.4*7. - Laboratory plant For vacuus coating by evaporation. Tha avaporatad •attrlal travala in straight lines la all tb« directions,, coating that work (aubatrat*) aa wall aa tha bell jar. If the utarlal to ba evaporated Is trough concentrated (a small fllaacnt or basket) to ba considered aa a point source, the thickness of the deposit t (at) In th« middle of the eubstrste (work) opposlta o tha evaporator, will bo

(4.13) *i Oh - 202 -

•here W la CIM evaporated aua \R). P la tlw apaclflc gravity (•/cm ). a U tan distance evaporator - work (en). If the notarial Is evaporated froai a boat, than tha tklcknaaa la the alsela of tha wort will b*

a p h

To* talekneea t at any point of tha aubatrate at a distance froa the alddle la given In Hg.4.8.

' - . i =3 O O-l 1« j. r» to

rit.4.1. - Distribution of deposit on a plana futfaca (or a) evaporation

•acmes coating tachnlquea are tha eubjoct of a vary let*" nuaoar of suolleaeloa*. Tor a coapralianelve treataont of the subject in refer to f^UTi (raf.l* Hooka, Chapter 1.4). 4.21. The, permeation PH^III Guu h*va the poaelbility Co psas through solid*» even if tha openings present era not large enough to permit a ragulax flow. Tha passage of a gaa Into, through and out of a solid barrlar hairing aa holaa larger enough to paradt •ore than a small fraction of tha am* to pass through any one hola la known as par—atlon. The steady atata rata of flow In these condition* la the permeability coefficient or eimply the permeability, this la usually expressed la cm of gas at 2 SIP flowing per second through a est of cross section, per at of vail thickness and 1 Torr (10 Torr, 1 at) of pressure drop across the barrier.

The process of permeation la described by Horton as shorn In Fig.4.9. It inTolves first the adsorption of the gas on the surface where the gaa pressure Is higher. After being dissolved In the outside s*jrface layer the gas slides down the concentration gradient and diffuses to the vacuum side where it is desorbed.

Generally, cases dissolve ia solids . •» a concentration c :

c - b.PX/i (4.15)

where F la tha gaa pressure, J la the dissociation constant of the gaa, and b is tha solubility of the gas in the aolid.

The dissociation constant J , ia J - 2 for diatomic gaaaa In •stale, and J • 1 for all gasea In nonaatals.

Tha concentration e la tha amount of gaa (In Torr.cm , or

The solubility b Is the quantity of gaa (in cm ) at STP (293"E and 1 Atm) that la dissolved ia 1 cm3 of the substance at a pressure of 1 Atm. It is dimeasloolees for J - 1, but baa the dimensions of AtmX/2 for j - 2.

F.J. Norton, 1961, Vacuum Syap. Trans, p.8; 1962, - 204 -

O—;

DtSORPTKW

?V2£C7 MtHBRUt L,nH IMtCKMCSS

rtg.4.9. - Hi* aaraaatloa procasa.

™» iiffytXSt. ef tfca §as Into tni throvth tba Mild otejra IllK>-lML£UtflWla «hleh ara:

Tick's tint 11. la cha staady stats, whan eka fas coaeaa- craelsa la laisanuaat of clat, fas

(•.«)

«*•*• ° 1» tha

•(-!« where H Is the activation energy of absorption, D la * constant fox a given gas and material, and R Is the gas constant. flck's second law. In most cases, equilibria* Is reached only after a long time or not at all, sine* D la small. In this tramalma period, when the concentration varies with time, Pick's second las states that

dx For the case of the steady state, the concentrations at the two surfaces with pressures P. and P- , are c. • b ?. J and 1/1 L •£ X l e, =• b.P., J. Prom eq.4.16 It follows that 2 -f dc (4-19)

1/3 „ 1/J I 12 .20) where d Is the thickness of the natarlaJ. The product .Q.b. between the diffusion coefficient and the solubility, is called the per—tion constant K . It is coaeonly 3 2 expressed as tha anount of gas (cat Sl¥), permeating through a I cm croas »ectlon of a slab of 1 cm thickness for a pressure difference of 1 Atm. Figure 4*10 and 4.11 show a review of value* of the permeation constant K. - 2fM> -

•iliai* •abioaOinll • 2l*I2MMlMCnM» r. A. M»h«d. 1.1. IMM, aaj K. V. K.. •.•••, -1*. illlli'M MKITM J*!*-, nl. 17,p-Xt],M?;N_.l.l3,u4 U tea V. O.

* W»t7%ll—iM nnwh 7. K»-Yjr«tr

11.1 U.tU-StmtVm M.IU-frrM7«U ^.i*/* S^—

n».4.io. 1 .»'l

IVnnCftiiu* cun-tnn'n Tor vatinu» ilbluiuM Kiw-mt-Ul c«ttt''inn'i>Mu> »• K funrtion of tmi[«-rti I urr. Ilait* Ktc m1 KIIUI/H-T. ((fiittnlilx t4 Rn« in «ul>ic rrnUmclrra (KTI*J puriHR |«-» urund tKtim^ti • «*M of I-WB' uri «wl l-riii lliirkftftv, *lwn n |irrtnir* tlHfcrtiwi eC I »tn» mUlu urnwi thr *all| (N'UHIIICM I U» * from r. A. I(»ll-f».l, J. l». lli>>»»X. sn.l K V. Komcbru, ,HJr. K(tttt*n. tihttm* I'hg*., l 8. Srhultm. IVMIW. ml i:t, p. M:t. HK3.1

1. llrIM .'. UrNI 7 HrKc

3. IIrM«. H. llrOi «. NrFr H. fft-.1MlftYir«*(Bfein«»ftvl

5. NrMu ID. ||r4mM>rin»t

Pig.4.11. ~ -208-

According to Norton , certain criteria Mist be fulfilled to distinguish, mmambiguoely, true permeation from IMM flowing through an actual hole, and gases derived from tha walla of the orfrelope (ovxmaeaiag). After « thorough degassing by a good vacuo* applied on each aid* of tha wall, tba effect of * hole can ha dlatingulahad froai true permeation In two way*. Vary rapid rise of tha particular gaa on the low aid* after application of pressure to the high side say indicate 1/2 a hole. Variation of the rate with (T/M) la shown by tasting with gases of differing molecular weight. Variation following this lav shows that a ho}e exists. The diffusion coefficient D is nost conveniently Measured by the tine lag netbod. In this, the affective time lag t. (sec), to attain steady state permeation through a membrane of thickness d. (en), 1* related to the diffualon coefficient: *~& <*-2l>

4.22. t.jmaatlon through vacuum envelopes The metallic, glass or rubber vails of vacuus vesseld or pipes are more or lass permeable to tie** tha permeation mechaniaa can he atomic or ayjflacujlar,. Hydrogen permeation through metal* increases with the samara root of the pressure; this fact 1* explained by the dlasociatiOB of tha hvdroeen to atom* and their paasage as such through tha metal. Xacomblnatloo occurs on desorption and on tha lev pressure side molecular hydrogen appears. In glasses and alascomer* tha gas permeation is proportional to the pressure. Sere the permeation Itself occurs in molecular font. The permeation of etnoapharic gases through metal walla does not include the raae gaaaa (He, A, Me, Fr, Xa) since Iff T1TT lit iitfWff throuah, metal* at any temperature under purely thermal activation* There can be penetration of rare gas ions under a potential gradient, or rare ganes can be formed in situ In tha metal interior by nuclear desintegration processes.

7.J. Norton, 1961, Vacuus) Symp. Iran*, p.8, 1962. The permeability of aluminium ror hydrogen La very small {Fig. 4.1-"), It 1* negligible In all caici except in ultra-high vacuum chambers, or with chambers heated at high tuttpcracure.s and havinj; tliln walls. Copper is a metal with low permeability for all the gases. Including hydrogen (Fig.4.12). Nickel lias a higher permeability for hydrogen.

Fig.k* 12. • Permeation of Hydrop.cn through various materials. - 210 -

Therefore-for; weto r^cooled chambers where th« danger of hydrogen perms attorn la greater, copper Is to be preferred to nickel* jron ncm containers have high permeability (Fig. 4.12) for hydrogen, especially If tha hydrogen Is In atoartc form oo tbt high pressure side do* to chaadcal or alactrolytlc offacta. Thua tha cooling of Iron vacuum coatalnaxa abould ba made with liquids which do not contain hydrogen ions, or air cooling should be used* The permeation of hydrogen through steels, increases with Increasing carbon content; low carbon steels ace thus preferred as vacuum containers.

The permeability of glasses la Important only for systems in the ultra-high vacuum range (P < 10~ Torr). The permeation la Influenced by the kind of glass and the gas involved. As a general rule the denser the structure of the glass and the greater the molecule of tha gaa the less the permeation. This is the reason why gases permeate easier through silica (S10_) than through technical glasses (Ftg.4.13). In technical glasses the open meshes of silica or other glass formers are occupied by network modifiers as Na»'K, Ba.

The permeation of helium through vitreous silica (or vycor) Is 10 times greater than through crystalline quarts. Vitreous silica has a considerable permeability also for other gases, like hydrogen, nitrogen, oxygen and argon.

Organic polymers (rubbers, plastics) are permeated by all the gasas including the rare ones (Me, He, A, Kr, Xe). There are vide variations In permeability. That of CO, through natural rubber is -5 3 2 high (about 10 cm STP.mm/cm .sec.atm.)» that of air Is lower (10~ cm STF mm/cm .sec.atm.). Saran, polyethylene and Kal - F —7 3 '2 have generally low permeability (about 3 x 10 cm STP.mm/cm .sac.atm. at 25*'Z). rhe^main featuraa of the permeation processes are summarised In Table 4.3. Table 4.3 Gal rmmalv, n,IO<,oh,

He, II., 1>„ Ne, A.O. Mo riiro jtu Iff mill IMIimujtl All gaM-a IMTI urate DteiMirulile tl»nmj(li IhrniiKl" uny UruiiUHi. til pulynicni. MO. nirUl. U» periiirttUui Ni*, A out mew W»tw r*(» apt to lllddt, l-KIH'cilllly umblu. be high. 1M. Vitrcuuti Hilica >1 IKTUlWttOt A|(. Many apecifoiUM. (fustL-at) Ill through Kuby ciirru.iiuii, clci--

Alt rntri vary infiM viiry HN If, rale vxnra M All r*tra vary w il:«itly ao iimgi ((irwaufi- > (|inwnT)t pirn iburv. 4

— — •-- »* - '!&' - c5>' e 10* #7 a •5 10'

• "' "^ 7 t S^yf A ft '0' / 1* «o "'* **7 -i3 #5>- S Jj •f"e Ki* «P-^ ^7 ' ''I ;* - - 1 7I-I\ •a" W — -' .#, - B /f- »o *.- ... .1 . —., -_

1 n 3t XI « W K OKwI0 D•0 0 *, Teirewolwft. *C Fig.4.13. - Permeation of various eases, turouj-h various materials.

4.23. jCpnsociuerices^f Eenneatjloii The consequence of permeation, la obviously the transfer .of SSB froa tile nich pressure to the low pressure side. This process Is Ujdtinftjhe fAuaJLirossure to which a vessel ran he evacuated, but also permits to li»t»duco.jno.ajpjMX JIUJI^^^ Into the evacuated systems. labia 4.4.

Ow»m OP FLUW OP ATMtKHIWIC GAXS INIO SlO, 01U AT 25 ^C (mH I MM THICK, I CM1 ARM)

.. u~: u Pcmwalion r (tin* Aunosphcncabim- ; inflow Order or Alomt/Kc fcmV«ej inflow

59.S 2>. 10-" 1.2xl0-*{ 15.9 l»IO-» I.6XI0-" 0.705 2>:lo-» I.4X10-" I.JXI0-" 2XI0-"

WxlO-* 5XI0-" 2.0x10-" 500.000 H, 3.»xl0-» 2.8x10-" 1.0XI0-" 25

In order to show the Importance of the Inflow of atmospheric taies throurh tin wnlls of a vacuum vessel, Morton gives none Interes­ ting examples. Re considers a bulb of vitreous silica at 25*C with 2 3 walla 1 mat thick, surface area 100 cm and volume 330 c« , and assumes that tha walla have been completely degassed, that the Initial pressure Is 10~ Torr and that the steadyatate flow Is established at 25*C. From the abundance (partial pressure) of the various fames In the atmosphere, and from the permeation extrapolated to 25*C (Table 4.4) the order of Inflow of the gases (Table 4./ , and the gas accumulation (Fig.4.14 and 4.IS) la established. - 213 -

It can be seen that the gases of low abundance show the highest inflow and the order of accumulation In the silica bulb (Fig.4.14) is (1) helium, (2) neon, and (3) hydrogen. A big difference separates the succeeding gasea oxygen, nitrogen and argon. For the vicreoua silica bulb (Fig. 4.14) the gases and pressures at the and of one year in air at 25"C, would be 10 Torr helium, 10~ Torr neon, and 10 Torr hydrogen. Only a few molecules of oxygen would have pemeated even after a hundred years.

r—r1 1 !— Atmosphere qo5 occuinukition

totj tune, ' sec

Fig.4.14. - Atmospheric g.is accumulation at 25"C, In « silica bulb 3 2 330 cm , 101) cm wall area, 1 no wall thickness (Norton)* The increase in pressure in bulbs of various galases is shown in Fig.4,15, for helium permeating from the atmosphere. To reach a helium pressure of 10~ Torr, requires, 3 days for silica, a month for Fyrax and vary long times for soda-lime glass or other glasses. From this,^it is evident that if we are concerned with sealed off - 214 -

VACUUS container* with pnuum In th* range of 10 Tbrr, 1c la HCMsary to make thalr envelope of a (lass of low pemeablllty, or surround It by a aubaidlary evacuated chaaber.

—}—I 1 1 Hetwn octwrxJotmn from the (rifflMptttr*,25*C

log lime, stc

Fig.4.15. - RelluM accumulation from the atmosphere In bulbs of various glasses, at 25*C.

The perscatlon Id uied when ,3pe_cif ic ftaaca are to be Introduced in highly evacuated ayatcas. Calibrated leaks of helium are extenti- vely used in leak detection. These are glass bulbe filled with heliua, and sealed with a graded aeal end a silica tube having thin walli. Such calibrated letka uy give leak rates as low aa 10 Ata.ca /sec, and are very constant for uny years. - 215 -

For the Introduction of pure hydrogen, palladium or nickel cubes are used, while silver Is the best material for the diffusion of oxygen. In order to Increase and control the flow rate (permeability), Pd , Hi, or Ag tubes with thin walls are heated by colls, or direct electric current.

4.3. Sorption

4.31. Sorption phenomena la the kinetic theory (Chapter 2) and the flow (Chapter 3) of gases. It was assumed that the interactions between gas molecules and the walls of the containing vessel are aainly elastic collisions In fact other types of interaction occur which have: a profound effect upon the degree of vacuum obtained and upon the processes used to achieve the vacuum. The group of interactions in which the gas Is retained by the solid (or liquid) received the name of sorption. This includes two mechanisas; The adsorption and the absorption.

The term adsorption refers to the process whereby molecules are attracted to and become attached to the surface of a solid, the * resulting layer of adsorbed gas being one (or a few) molecule (s) thick. The attracting forces of the solid may be physical - physisorptlon, or chemical - chenisorptlon.

The term absorption refers to gas which enters into the solid in much the same manner as gas dissolving in a liquid*

The solid which takes up the gas is known as adsorbent or absorbent; the gaa removed ds known as adsorbate or abeorbate. Terms as adatom or admolecule are also used to refer to the specific particles involved In the process.

4.32. Adsorption energies Any surface of a solid or liquid exhibit! forces of attraction normal to the surface,hence gas molecules impinging on the surface are - 216 -

eaeorbed. When the preaeure in thi- syatea la low enough (high ncuua) tin eoleculea adaorbef at the wall exceed thoee In the malum, thu« the vuaping 1» directed toverde amnuttni the adeorbed faa. Convereely gee ean ba removed from tht voluaa by adaorptlon, a procaaa utilised In the aorptlon pumps. Adeorptica phenomena ara aheaatlcally repreaented by their potential energy-distance diagram (figs.4.16 - 4.19).

BWtnct Mm mrilct —-*•

flg.4.16. - Potential energy of a molecule in (monectlTeted) adsorption.

A molecule Impinging on tba aurl ce la attractad and will : m equilibrium psaltion at aintrwa potantial energy, called tha haat of adsorption. H^. Tha heat of adsorption la equal (In this simple caaa) to tba energy of edeorptlor E_. - ?J7 -

If the adsorption Is purely physical. It involves Van der Wall, Interaolecular forces, like those occurlng In liquefaction of gases. In this two-dimensional liquid, H is larger than the heat of liquefaction. If additional layers are adsorbed, H. decreases until the layer becomes a three-dimensional liquid. In physical adsorption, the atracting forces are comparatively weak, and the heat of adsorption is small (max.8 kcal/mole). Since the forces are attractive, work is done in adsorbing molecules and heat is generated, thus the adsorption is an exothermic phenomena.

In chemiBorptlon the process is similar to the formation of a chemical compound with transfer of electrons. In this case the attractive forces are much larger than in the physical adsorption, heats of chemiBorptlon being correspondingly higher (as large as 250 kcal/mole). The process of chemisorption does not always occur directly from the gaseous state; molecules may be initially adsorbed physically (Fig.4.17) and then, with the provision of a certain minimum energy (activation energy E.) they may become chemisorbed. This is known as activated chemisorption. The energy of desorptloa is the sum of the heat of chemisorption H_, and the energy of

activation EA .

•n ' Hc + h

The process readily occurs during adsorption at a heated surfacei and the total amount of gas which can be adsorbed in this manner is higher than that by non-activated processes.

The over all chemJsorption process of molecules is exothermic.

The inert ^ases cannot be chemisorbed and there are therefore only weakly held on a surface

Moleculei may dissociate and be chemisorbed as atoms (Pigs.4'. 18 and 4.19). - 218 -

Distance Iran surface

Pig. 4.17. Potential energy for activated chealaorption, with •olecular adsorption.

This process can be endothermic or exothermic. It twice the energy of dissociation D , the process la etadofchenic (Fig. 4.18).

If 2ED > D , the reaction is exothermic (Fi*.4.19). Soae values of heats *f adsorption ara listed In Table 4.5. Distance \

Fig,4.18. - Potential energy for activated cheraisorptlon, endotliermic atomic adsorption of 'Uatoralc molecule

atom

s s / ' / / I ™D / t f\ F- 1

ads»b«d 1 mofccuta j

atom* _J

Diibnci from wta« Fig,2.19. - Potential energy for activated chenisorptton,

exothermic atomic adsorption (2En > D). E^-energy oC desorptlon; Ep-energy of physical adsorption; T.-actlvatlon energy of adsorption, l^-heat of clicmisorption; n-cnersy of dissociation. T*bla *.5.

HMH «f Adwrpfon I* KiloufarlM pt not* for OMnfUyri CT-iMMwaliaa: •D «, affi lit 0»o«»•*- «l II, MFa 33 BaiW l«u K, Mft KMIM 4H II, Mir *M• Ac»lU 3fi II, »Ka ai U.MW M 11. Mb ~M Aal 1M II, HII.MW w NH.M HI B,MU. ~*0 n> MNI 3»6 11, alt «i HJ M(X 8 nu.i,ip<„.tii.n x.«.w XI MM. KTMW >~«.« t Xe M1» ~»~t.3 AMW ~1.« annei: G. JOKIM, 4»*. irr. XtM. &*,*•!. igl, art. 3, p. 733, IMS, •wllMl rwtmmSrmr.Tmn,f 131, IMS

DM *n» tiara a certain aoliUty on th« austaca and ttwj aajr -mtftrnf am th« aarfaca, aa th# nacaaaary activation anargr for thl« aotjoa la oaljr D.2 - 0.4 froa tha valua of Eg . Khan too adiorbed atoaa colliaa In thalr notion on tha aurfaco, thai- aay racoabina and daaarfc aa a aolaoula. Ihia proeaaa la known aa aocond-ordar 4ss>s£iaa- 4*33. Monolayer and sticking, coefficient

According to *q.2.46 the number of molecules adsorbed on unit area la given by

~* 3.15 x 1022 aP(MT)"1/2 (4.22)

P is the pressure of the gas (Torr).

In addition to adsorption, molecules are desorblng fro* the surface at a rate given by

dn. N .8 where N_ is the total number of molecules required to form a complete monolayer (see eq. 2.85), 6 is the coverage (i.e. the fraction of possible adsorption sites which are actually occupied). and t is the average tine spent by an adsorbed molecule at a particular site (known as sojourn time).

The sojourn time is shown by Frenkel to be

t - t'.e <4.24) where t* i« the period of oscillation of the molecule noma! to -13 the surface (approx. 10 sec), and E_ is the energy for desorption.

Equation 4,23 la only valid for leaa than a complete monolayer. Similar but more complex equations war* deduced for multilayer adsorption.

From eqa.4.23 and 4.24 it results that

B_ d^"f-«V «-25>

* J. Frankal, 2. Phya. 26 (1924)* 117. - 222 -

Wbe exponential dependence of t and henca of dn./dt , upon both L and T mesas chat t rarlea oror a vide not** from about -13 7 10 to 10 sec fox nail value* of E- (i.e. physical adsorption) -5 30 and loir tempereturea (77*K), and fro* 10 to 10 aac for chemisorp- tion (E- - 10 - 200 kcal/mole) at room temperature. The equilibrium between adsorption (on the noncovererf ' area 1-3) and desorptloa (from covered • area: 9) is found from eqs.4.25 and 4.22

H 6 - 3.51 x 1022 s P(MT)~1/2 t' e0 ° (1-8) (4.26) o where T Is the temperature of the gas, and T is the temperature of the aurface.

This equation can be used to express the aaount adsorbed N 9 as a fuaetion of P for constant T (the adsorption isotherm), as a function of 7 at constant P (the adsorption ifc->bar)> and P - f(T) for constant coverage nj (the adsorption isoatere). Unfortunately, the sticking coefficient s , and the deaorption energy E^ are not constants. In practice, isotherms are observed experimentally and used to determine s , and 6 .

However eq.4.26 predicts the following general features: a) The quantity of gas adsorbed increases with pressure. b) Very little gas can remain physically adsorbed under high vacuum conditions at room temperature. c) At low temperatures the quantities adsorbed (even fox low E. values) arc considerable.

Typically, sticking coefficient* at room temperature lie between 0.1 - 1 , and decline when monolayer coverage

(2 - 7 x 10W «olec./cm2) is approached. Figures ^.20 and 4.21 Illustrate this. 1.0 -&r

CO - w"' — ^ CH, V

V ^ X - V L . f 1 \ \ ^— 10 \ i 11 k v. \ s Mc I 4 i t ?«ia? » Total number of adsorbed raotecuks ptt cm1

Slicking probability vcrwis number of tdMrbwl molecule* on tho 411 pl»BO Of lunKsUm *t 300"K. (Aftfir J. Becker, "Solid RUtfl F%»n," vol. 7, p- 379, Awtonlc Pro* ln

Fig.4.20.

A* it is ahown in fig.4.21, Alpert found that the sticking coefficient ia also low at the low coverages. The explanation has been suggested that tliia can be attributed to the need for oucleation centera. -224 -

Motaults/an?

CmpuiKw of the nmdtm of vimow invntigaton foe tfc* itk&iaf probability ot aitmnni o*. bngrtca. The two nnn liven Tor Efcrifeb «• for Mrfeeateaspcntonwcf 2fO«a4 373°K, All atken an nofainalb* at MM tawi- petatm. CAfttr t>. Lee, H. Tonwhke, and ». Alpcrt, 1961 Fannim 3*ma. Trmn*-, p. 153,1M2.)

Flg.4.21.

4-34. Adaorotion Uotrrr— Laogaulr* used eq.4.26 to express the adsorption lsothem a* •rfc -177 <».27) vher* VRoT. & - 3.51 I 10,22._L^S" m . (•.28) (HT> 1/2 (for notations •« «q.4.26).

. J. tm. Chen. Sac. 40, 1361, 1918. For constant T and T , b Is A constant thus eq.4.27 la an isotherm; b is a constant, expressed in Torr units. Figure 4.22 shows such isotherms for various values of b * Since the value of b ia decreasing with increasing T and T^ , the curves for high values of b , correspond to low tenperatures, those for low b_ values to high temperatures.

b = 1 Torr"1

100 200 300 400 500 600 700 BOO

P(Torr)

Fig.4.22. - Langauir's isotherms.

If the pressure la snail eonpared with lib , the coverage S is proportional with P ,

e % b.p - 226 -

Langmuir'e isotherms VM derived for layer* less than monomoleculsr.

The BET-lsotherme* vu derlvtd foe multimolecular adsorption. This isotherm is described by the equation

vff, - *> vc Vc p^ where v Is th* volume of gu adsorbed *- a given value of '8 v and C arc Constanta at any taaparatu a. is th* saturation pressure of the (as at the given temperature (eq.4.6). The BET isotherm* are represented by curves of the shape shown In Fig,4*23.

The constant C was found to be a fraction of the temperature,

(ET-E.)/ET where E. la the energy of adsorption of monolayer, and t^ the energy of condensation of the adsorbed gas.

4.35. True surface Relating the quantity of gas which tea* found to be adsorbed on surfaces, it was concluded that the true surface is usually Mich larger than the apparent one. This true surface is also known as Physical, surface. C*P) , while the apparent one has often the nee* of •sowstricel surface (Ag).

The BET equation (4.29) is a convenient method to evaluate the true surface area. A plot of the expression Pv~ (PyP) verstii

*/Pv will give a straight line for which the intercept Is l/(vB.C) end the slope (C-l)/(v .C). From these two data, the values of the constants C , and v can be calculated. v Indicates the volume of gtm in a monolayer, thus the number of molecules forming the complete layer.

S. Brunauer, F H. Earn tt, and E. Teller, J. An. (Stem. Soc. £0»

309p 1938. /v,

..L L J I I I 1— 0.1 0.2 0.3 0.4 0.5 0.6 07

P/p

Fig.4.7° B£T isotherms.

The BET method was used by Schran , who determined Ap/Ag ratios « Urge as 900 (Table 4.6).

The ratio Ap/Ag of the physical surface to the geometric one. was also determined by the Mthod of tlectrolytic polarization and these valuta are also listed in Table 4.6.

When a stetsl ia made the cathode la * dilute acid, sad current la passed through the solution, the potential changes, due to accuaulatioa of hydrogen on the cathode.

Schraat A.p Le Vide 103, 55 (1963), mix *.&. - Xaclo of physical (true) «urf*c« Ap to s«oa*trlc (appuant) surface Ag*

. Metal Surface Ap/Ag *f.

¥ery thin foil 6 Al anodlcally oxidized (20 y) 900

Cu Plate (1 asO 14 Schraa Steel - 26 Stainless •te«l Mate (1 wm) 8 bright foil 2.2

Ft Bright foil, cleaned acid, heated flaae 3,3

platinized 1830

polished, new 75

polished, old 9.7

Hi ectlvatcd by oacidetion and reduction 46 Dustman

activated, then annealed 10

rolled, new 5.8

freshly etched with dilute nitric acid 51

etched, after 20 hr 37 Ag finely sandpapered 16

aulgauted, after 1 hr 1.2

C arc carbon rod 328 - 366 - 229 -

The phenomena is described by the equation

KV - E - j~ +• const (4.31>

where E Is the potential, X • - — , V la the mount of hydrogen on the surface of the cathode, and Ap is the physical surface.

4.36. Sorption of pases by absorbents The vain abaorbants used In vacuum technology are: activated charcoal,, zeolites (molecular sieves) and silica a;el. The absorption of theee absorbents is explained by an adsorption* followed by the penetration of the adsorbed gas into the solid by diffusion. It can be considered that the absorption is a phenomenon similar to permeation* but having no desorption surface.

Sorption, by activated charcoal Before the development of other means of pumping to very low pressures, the technique of producing a high vacuum by absorption of the gases on activated charcoal was very frequently used. In the period 1900-1950 there are more than 4000 publications on this subject. The mast frequently used charcoal is that prepared from coconut shell. Pieces of the shell, are destructively distilled at 500-700*C, in Ion containers, until vapour evolution is no longer apparent. The charcoal produced contains tarry residues, which are then removed to increase the gas absorption efficiency. This process is known as

"activfttion"t antl consist* in heating in steam at 800-1000'C for about one hour. The water In the activated charcoal la then driven off by heating the charcoal In a roagb vacuum.

The ratio Ap/Ag was found to be 600-850, which corresponds to specific surfaces of the order of 1000 m /g. Fig.4.24 shows the volumes of various gaa absorbed.

The sorption of water vapour by charcoal exhibits a behavior quite different from that observed for the less readily condenalblc - 230 -.

|HU. (rif.4.23). Below 1.5 Tore th« sorption 1B nail, batwMii 1.5-2.5 Terr it auaUaaly lncre«*e*. Above 2.5 the Increase in Main slow.

Mnnt* charcoal OJOUgtm D.5u'l

A*-1»S o,Vi«r j L«— 7 t-* 1 - -i9sr /> *>* a-. I COM -113*

£2 •*" CO!* -W / _^5ji.-,r. I i" P^ i

Fig.4.24. - Low temperature adsorption on charcoal.

1 1 ; •— 1 1 <" , "i— IN ^ r • 160 - 11*0 . £ i . ^120 i ? • £ 1 >> - «1«|0« 1 . K * 90 - - li» / i . 10 / % - 20 -

—i—r*"T i • Km rig.4.25. - Sorption of water vapour oa charcoal « 0*C. - 231 -

The pumping effect which can be obtained by using liquid-air cooled (-183*C) activated charcoal traps, ia shown In Flg.4.26.

•a i • in ? mi 1 i i i i i in I 2 4 h 10 100 1CC0 10000 TIME.IN MINUTES

Fig•4.26. - Pressure against time curves on pumping H_, N-, 0. by means of liquid-air coded charcoal trap.

So_rp_tion jj^zgp.lltes Zeolites are alkali metal aluminosilicates, having tetrahedrlcal lattices. Unlike ordinary crystals containing waiter of crystalli­ zation they can be dehydrated without any change in the font of their crystal lattice. As a result, molecules of different gases can occupy the apace* l«ft vacant by the removal of water, and the zeolites are therefore very good absorbents. This, ia however true only for certain gases since these materials exhibit the property of pjtraprption. The parsorption may be defined aa adsorption, in pores that are only slightly wider than the diameter of the adsorbate molecules. An exampla of sorption curves is given In Fig,*,27. - S3J -

' " ' u _--^*^-* ** Mft -^L • IM M

• — ia M 0,-ttJth^ 11 - k J*

w i» * • ^"*!5^— *• „-——" " "~ 4 f^r-^" r i 3m 4H CM aw WOO 1200 MOft UM um 39M MOf•l *-

Fig. 4.27. ~ Aaouoc o£ caa VQ

ehabaalte {Ca Al2 Si^ I parcantagaa of dahydratlon (0).

•aaad on tha action of aaolltaa Gaolaeular aiaraa), taa aoroelao puaof «r« abla to puap doim ncuw ayataaa of 1-50 lltara, _2 from uaoapharle j>r«««ur« to th« rang* of 10 Torr. 16*7 an In uaa aapaclally la appllcatlona vbax* tha ayataa) has to aa kaat oil fraa, or afcara tha (aa which la puaoad la daotaroua {a.». railoaetlva). toratloa ar alllea aal flllea gal la a partially dahydrattd Jally of aiUeie aeM. ft la aaad aapacially aa diylnf aianc for pm (rif.*.2S). y f <7 / / s «i / I /

/ /

10OP/P, Fig,4.26. - Sorption-desorption curves for water vapour, of aillca gel

4.4. Degorpttpn - autpaaaXwe 4.41. Deaorpttort phenomena When a material Is placed In a vacuum the gas which was previous-* ly ad-or absorbed begins to desorb. i.e. to leave the material. Th« deaorptlon la Influenced by eh* pressure, the temperature, the shape of the material* and the kind of Its surface. The prassura has a basic influence on the deaorption phenomena sines according to its tendency of increasing over or decreasing balow the equilibrium, the phenomena of sorption or that of dssorption appear*• Nsvsrthelesa the function between the desorption rate and the pressure Is not proved at pressure* much lower than the equilibrium. The difficulty consists in separating the effect of the pressure from that of the pumping time to which is usually connected.

Th* temperature has a clear Influence on desorption phenomena. Desorption is endothermic, thus it la accelerated by Increasing the tempersture. - 23* -

*h* JSJBi, of the material influences tb« deaorption either If the gas !• **~ or absorbed. If the gas Is adsorbed, than only the amount of the surface Is ths influencing factor* but If the g«s has to diffuse fro* tba Interior of the material to the surface, then the third dimension (thickness) la also influencing the rate of Resorption.

Since deaorption phenosene are related to the phTsi^ffl TVtf>Tf Cap» see table 4.6), the deaorption anet alvays be correlated to the history of treataenta (polish, cleaning, etc) of the surtaxes,

4.42. ftttaeseiag The generation of gas resulting from the desorption, is known es ontamsslna;,, end is expressed In tens of the outgesslng constant! The outgaasing constant (or specific outgasslng rats) is defined as the rate at which gas appears to emanate from unit area of surface -1 »2 (geometric), and ie usually meesured In units of Tore liter.sec cm .

lbs experimental obaerrations of outgessing rates, can be represented by the empirical equation of the form

where K. and K_ are the outgesslng rates at h hours end one

hour respectively after the start of pumping; th is the time In hours after the atert of pumping. K is the limiting value of K. and is generally negligible unless t is very large.

At the beginning of the pumping, y Is large end outgesaing ratsa fall very rapidly, but after a few minutes ths fall becomes less marked, with values of Y » lying between 0.5 and 2, depending upon the material. For eetsls, the value of -v la usually near 1. for -nonmetals, y is lying between 0.5 end 1. Values of y greater than 1 are usually sssoclated with sn unusual surface condition, such as a porous materiel or rusty surface. After long pumping times (rarely leas than 10 h) outgesslng rates show e tendency to fall - 235 -

exponentially with tlae, until Halted «t X . A curve allowing * typical tlsa variation of outgasslng constant la presented In Ftg.6.29.

If tba temperature of the Material Is ralaad (baking) tha outgaaaing rate rlsea rapidly to a peak value (Flg.4.29), followed by

a slower fall back to a t.Y variation, but at a higher level corresponding to the elevated temperature. If after a sufficiently long tine, the tenperature Is allowed to fall to Ita original value the outgaaslng rate falls rapidly to a level which Is significantly lower than that Jhlch would have existed IE pumping had been at the lower temperature throughtout.

H—k-\

Pig*4*29. - Variation of outgassing constant with time of pumping.

Together with the acceleration of deaorptlon, heating Bay also have the effect of causing activated cheaisorptlou of physically - 236 -

adsorbed gaa (In particular, water vapour). which can than be deaoxbed only by prolonged heating at Mich higher teeparatures. Chaadaorbad water vapour continues to ba evolved at teaperaturer In exceas of 300*C. It should therefore appear that a degassing progress* should basin with pusplng at room teaperature to reaova physically adaorbed water vapour* before baking is eoamacsd.

H»e theory of the ouceaning process was derived and suaawrlied °y P&&&. • 3,M ceaplete theory of the outgesslng Includes both the adsorption and the absorption siaultaneously. However, in aoat casaa the rate of diffusion Is so small conpared with that of desorption of adsorbed gas that the two processes nay be analysed separately and the resulting occgaaaing rates subsequently added.

The outgasslne, .rata resulting frost absorbed eases, is based on the laws of diffusion, and can be obtained froa> solutions to eq.4»16 and 4.18.

In general the solution consists of the sum of an Infinite aeries, but any be approximated to give the outgasslng race froa the wall of a vessel of thickness L (ca) as

c h where

'»! -2~I~ tt.35) where ? is this diffusion time conatsnt, D the diffusion coefficient (eq.4.16).

When t. < t;/4 , then t^ - tr , and thus K^ varies initially as tT but eventually falls more rapidly to approach an exponential

B.B. Dayton, AVS Vacuum Symp, Trans., 1960, p.101; 1962 p.42, 1963 p.293. - 237 -

dependence as t. becomes large. The theoretical values of K, K are given by

ajaj j-31 MUZ •(3600) #T' ^H where yl is the value of y when t, » 1 , and e is the gas concentration when t. * 0 , measured la cm3 (STP}/a°»3 o£ material.

3 Ku » 2.79 x 10~ T ~ P <4.37) o

The product D.b is the permeation constant (eq.4.20) which is 3 2 measured in cm (STF)/cm of corss section for a thickness of 1 cm, mi pressure differential of 1 Torr. The pressure P is the partial pressure outside the enclosure of the gas considered.

Considering the outgassing at 27*C of hydrogen from a steel vessel of wall thickness 1 cm, the various parameters are: D - 5 x 10~9 cm2/secj e - 0.1 cm3 (ST*)/cm3; h - 10"3 cm3(STP)/cm3.Torr; i - 2 (aq.4.15); f «4s 10~4 Torr (partial pressure o£ hydrogen in the atmosphere).

From eq.4.35 it results that c = 10 hours, t < c/4 - 2.500

Substituting in eqs.4.36 and 4.37 we have

In this particular case the permeation otitgaulng rate will almost certainly be greater than the value of X calculated from eq.4.37 because of the liberation of hydrogen at the outer surface of the vessel by action of water vapour on iron. Xfce real value of K is - 238

-12 -1 -2 about 5 x 10 Torr.lit.s .cm . Experimental value* of K. an •boot one order of magnitude larger than that calculated above, ubaraaa y observed experimentally for this caie la in the ration of Y • 1- Thus, factor* other than diffusion of gases fram the Interior play a considerable part in the outgsssing of Betels. The outgasslng due to water vapour Is believed to be the main additional factor* The ovexaasinfi rate resulting from an adsorbed monolayer can ba found be using eq.4.26; but the results obtained are not always meeting the experimental values, since the coverage 6 Is also a function of the pressure. For an approximation, the following equation can be used:

10~7T 8 . K „ o e-t/cs (4>3g)

where K is the outgassing rate at time t , T is the temperature» t sthe sojouron time, 0 the coverage when t - 0. If 9o * 1 (monolayer)» thin for small values of t (physical adsorption)» the initial outgassing rate is very high, but falls rapidly with time* On the other hand for strong chejalsorption (t high) the initial outgassing rate is low and falls only vary slowly with time. For water vapour, adsorbed in several layers the above approach does not give consistent results. Dayton* suggests that water vapour is held in the pores of the layer of oxide that is inevitably present on the surface of most metals. A eenl-eaplrical analysis of tha distribution of pore size and layer thickness leads to an expression fir the outgasslng rate which varies as t~ . 4.43. Outgassiim retea Outgassing rates were determined by various authors. Figure 4.30 shows the values obtained for various materials, end their decrease with time of pumplag* This figure shows the outgtssing rata* of mecals and plastics.

B.B. Dayton, AVS, Vacuum Symp. Trans. 1962 p.42, Fig.*.30. - OutgMsina rates of various materials at room temperature.

The gas evolution fro* glasses la shown in Kg. 4.31 • Br heating th« (less to 150*C in vacuum, the greatest part of the aaaorbeo' gasea is given off. The curves representing the gas evolution (fig.4.31} have a —XIMMH point at about 140*C for eoda-llae glasses, at 17S*C for lead glasses and at about 300*C for boroslllcate glasses. At atlll high*' tanparatures th« gas evolution la reduced, but after the teaperatur* range between 350-450*C la exceeded additional gaaes are given off due to tbe decoapoaltlon of the glass.

ff4 1 —

• ».-« ..... (\ I \ A / ' A ^y

Fig*4.31* - the evolution of gee frcei various glasses.

Tha oatnaejuln^ rates of various materials dej>onj|on the state of their surface. Figure 4.32 shovs a suasury of these values» for Jfi£X**£*£ surfaces, for degroaacd. polished and bafced ones. The veloes for ueercaaed surfaces correspond to surfaces cleaned by usual 23 \ £

10

? I0 '\ >l J$ ' 3

ID'S 3 I 3 1 3W*

jfoeyeosedj*

| Polishe"ld \b

1 Baked \ (up to 100 hr -Outoossing ratts-

Fifi.4.3Z. - 2*2 - aethods value liquid degreasing agents; the lowest «nd of tl» rant** correspond to vapour eegreasing. Pollahod eurfaeea include aachanlcal polishing, blaatlnsi chaaical or electrocbeaical poliahlng. Fro* th« pebllabed data It vae not poeslble to conclude if one or anothar of these aethods eTBteaatleally raaulta in the lovar outgasalng rates; it rathar appaaxa that each of than can glva tha low values In tha range, If the procaaa la carried out carefully. The values for untreated, degreaaed and pollahad atataa were taken for 4-8 h of puaplug.

Tha bahlne of noaaetala la at teaperaturae of 80-100"C and baking tlaes up to 24h . Fox baked Metals the upper ranges correspond to baking at about 300*C for 24 h, the Middle ranges to baking at 400*C for up to 100 h, while tha lowest values for etainleas ateal also include an additional subsequent baking at 1000*C for 3 h. :£S» IA-1I7*

TECHNOLOGY tjj itf f^ ?A*1 «. m^--«-A' —' -&* •*»-

^V'*'" *•' .'fVr, •i»^**

1A-1271 Israel Atomic Energy Commission A. BOTH Vacutmi Technology October 1972 582 p. Th±s is the text of a Postgraduate Course given by the author at the Faculty of Engineer­ ing of the Tel-Aviv University. After an introduction dealing with the main applications and history of vacuum technology,

the course discusses relevant aspects of rarefied gas theory, and treats In detail molecular, viscous and Intermediate flow through pipes of simple and complex geometry, i Furhter chapters deal with relevant physico- chemical phenomena (evaporation-condensation, sorption-desorptlon, permeation), pumping and measuring techniques, and special techniques used for obtaining and maintaining high vacuum (sealing techniques, leak detection). (Parts I ft II).

0 1U^IUM'M

PART II

A. Roth

Israel Atomic Energy Commission October 1972 I

CONTENTS

Page 1. Introduction 1 1.1 The vacuum 1 1.11 Artificial vacuum 1 *- Vacuum ranges 4 - Composition of the gas 4 1.12 Natural vacuum 6 - Vacuum on earth 6 - Vacuum in space 6 1.2 Fields of application and importance 7 1.21 Applications of vacuum techniques 7 1.22 Importance of vacuum technology 13 1.3 Main stages in the history of vacuum techniques .... 14 1.4 Literature sources 18

2. Rarefied gas theory for vacuum technology 25 Commonly used symbols 25 2.1 Physical states of matter 27 2.2 Perfect and real gas laws 34 2.21 Boyle's law 34 - McLeod's gauge 35 2.22 Chales* law 37 2.23 The general gas law 38 2.24 Molecular density 42 2.25 Equation of state of real gases 44 2.3 Motion of molecules in rarefied gases 46 2.31 Kinetic energy of molecules 46 2.32 Molecular velocities 49 2.33 Molecular incidence rate 51 2.4 Pressure and mean free path 53 2. hi Mean free path 53 2.42 Pressure units 57 II

Page 2.5 Transport phenomena in viscous stat-e ..- 61 2.51 Viscosity of a gas 61 2.52 Diffusion of gases 65 - Diffusion pump (principle) 66 2.6 Transport phenomena in molecular state 68 2.61 The viscous and aolecular states 66 2.62 Molecular drag 70 - Tine to fom a oonolayer 71 - Molecular pump (principle) 71 - Molecular gauge (principle) 72 2.7 Thermal diffusion and energy transport 73 2.71 Thermal transpiration > 73 2.72 Thermal diffusion 74 2.73 Heat conductivity of rarefied gases 75 - Heat conductivity In viscous state 75 - Heat conductivity in molecular state 77 - Thermal conductivity gauge (principle) 82 Appendix 83

3. r'^s flow at low pressures S7 Commonly used symbols 87 3.1 Flow regimes, conductance and throughput &9 3.11 Flow regimes 89 - The Reynold number 90 - The Knudsen number 91 3.12 Conductance 92 - Parallel and series connection 94 3.13 Throughput and pumping speed 95 3.2 Viscous and turbulent flow 99 3.21 Viscous flow-conductance of an aperture 99 3.22 Viscous flow-conductance of a cylindrical pipe-Poifleuille'r? law 103 3.23 Viscous flow-surface slip 107 3.24 Viscous flow-rectangular cross section 108 3.25 Viscous flow-annular cross section 110 3.26 Turbulent flow Ill Ill

PaBe 3 Molecular flow 112 3.31 Molecular flow-conductance of an aperture .... 112 3.32 Molecular flow-conductance of a diaphragm .... 113 3.33 Molecular flow-long tube of constant cross section 115 - Circular cross section 117 - Rectangular cross section 117 - Triangular crosB section 118 - Annular cross section 118 3.34 Molecular flow-short tube of constant cross section 119 - Circular cross section 120 - Rectangular cross section 121 - Annular cross section 121 4 Conductance of combined shapes 122 3.41 Molecular flow-tapered tubes 122 - Circular cross sec tion 124 - Rectangular cross section 125 3.42 Molecular flow-elbows 125 3.43 Molecular flow-traps 126 3.44 Molecular flow-optical baffles 133 - Conductance of baffles with straight plates .. 134 - Conductance of baffles with concentric plates 135 3.45 Molecular flow-seal interface 138 5 Analytlco-statistical calculation of conductances... 142 - Transmission probability for elbows 147 - Transmission probability for annular pipes ... 148 - Transmission probability for baffles 149 6 Intermediate flow 154 3.61 Knudsen's equation 154 3.62 The minimum conductance 155 3.63 The transition pressure 157 3.64 Limits of the intermediate range 158 3.65 General equation of flow 159 3.66 The viscous-molecular intersection point 160 3.67 Integrated equation of flow 164 IV

3.7 Calculation of VACUUM systems .... 168 3.71 Sources of gas In vacuum systems 168 3.72 Pumpdown in the viecoua range ...... 170 3-73 Fumpdown in the molecular range ...... 174 3.74 Steady Qtate with distributed gas load .. 128 3.75 tomographic calculations 181

4. Pbyglco-chjemlcal phenomena in vacuum techniques 187 4.1 Evaporation-condensation 187 4.11 Vapours in vacuum systems 187 4.12 Vapour pressure and rate of evaporation ...... 188 4.13 Vapour pressure data • •. 190 4.14 Cryopuaplng and vacuum coating • • 195 - CryopuKping 195 - Vacuum coating 200 4.2 Solubility and permeation 203 4.21 The permeation process 203 4.22 Permeation through vacuum envelopes ...... 208 4.23 Consequences of permeation 211 4.3 Sorption , 215 4.31 Sorption phenomena 215 4.32 Adsorption energies 215 4.33 Monolayer and sticking coefficients ...... 221 4.34 Adsorption isotherms 224 4.35 True surface 226 4.36 Sorption of gases by absorbants ...... 229 - Sorption by activated charcoal ...... 229 - Sorption by zeolites 231 - Sorption by silica gel ,..,,,,.,.., 232 4.4 Deeorption-outgassing ...... ,,.,,.,,,,...,. 233 4.41 Desorptlon phenomena 233 4.42 Oulgassing , 234 4.43 Outgasslog rates 238 V

Page 5- Production of low pressures 243 5.1 Vacuum pumps 243 5.11 Principles of pumping 243 5.12 Parameters and classifications 244 5.2 Mechanical pumps 248 5.21 Liquid pumps 248 5.22 Piston pumps 250 5.23 Water ring pumps 252 5.24 Rotatlag-vane pumps 253 - Gas ballast 257 5.25 Sliding-vane pumps . - 261 5.26 Rotati&fc-plmgeT pumps 264 5.27 Roots pumps 265 5.28 Molecular pumps 267 5.3 Vapour pumps 269 5.31 Classification 269 5.32 Vapour ejector pumps 271 5.33 Diffusion pumpB 274 - Fumpiag Speed 274 - Ultimate pressure 276 - Roughing and backing ... * 277 - Pump fluids 279 - Fractionating pumps 282 - Back streaming and back-migration 283 - Characteristic curves 284 5.4 Ion pumps 286 5.41 Classification 286 5.42 Ion pumping 287 5.43 Evapor-ion pumps 289 - Small evapor-ion pumps 289 - Large evapor-ion pumps 290 - The Ocbitron. pump 292 5.44 Sputter-ion pumps 294

i VI

£SS! 5.5 Sorption puaips • •••• 298 5.51 Nature of sorption pumping , 298 5.52 The sorption pump .... 302 5.53 Multistage sorption pumping 303 5.6 Cryopuatping 308 5.61 Cryopumplng mechanism • • 308 5.62 Cryopuspimg arrays 316 5.63 Cryotrapping 320 5.64 Cryopumps ...... •>...•. 323 5.65 Liquid nitrogen traps 324 5.7 Getteriag 328 5.71 Gettering principles 323 5.72 Flash getters 331 5.73 Bulk and coating getters 334 5.74 Gettering capacity 336 5.8 Pumping by dilution 337 5.9 Measurement of pumping speed 338 5.91 Hethoda of measurement 33S 5.92 Constant pressure methods 338 5.93 Constant volume methods ...... 343 5.94 Measurement of the pumping speed of mechanical and diffusion pumps 344

6. Measurement of low pressures ...... 347 6.1 Classification and selection of vacuum gauges ...... 347 6.2 Hechanical gauges 349 6.21 Bourdon gauge 349 6.22 Diaphragm gauges . 349 6.3 Gauges using liquids 354 6.31 U-tube manometers 354 6.32 Inclined aanoraetera 355 6.33 Differential manometers 356 6.34 The Dubrovin gauge 356 VII

Page 6,35 The McLeod gauge 359 - Sensitivity and limitations 359 - Raising systems 365 - Forms of McLeod gauges 367 6.4 Viscosity (molecular) gauges 371 6.41 The decrement gauge 371 6.42 The ro tatlng molecular gauge 373 6.43 The resonance type viscosity gauge 374 6.5 Radiometer (Knudsen) gauge 374 6.6 Thermal conductivity gauges 377 6.61 Thermal conductivity and heat losses 377 6.62 Pirani gauge 379 6.63 The thermocouple gauge 382 6.64 The thermistor gauge 384 6.65 Combined McLeod-Piranl gauge 385 6.7 Ionization gauges 385 6.71 The discharge tube 385 6.72 Hot-cathode ionization gauges 386 - Principles 386 - Common ionization gauge 389 - Bayard-Alpert gauge , '.'. 392 - Lafferty gauge 392 - Klopfer gauge 395 6.73 Cold-cathode ionization gauges 396 - Penning gauge , 396 - The inverted magnetron gauge 397 - Redhead magnetron gauge ,: 398 6.74 Gauges with radioactive sources 399 6.8 Calibration of vacuum gauges 401 6.81 General 401 6.82 McLeod gauge method 401 6.83 Expansion method 401 6.84 Flow method 402 6.85 Dynamical method * 403 vtix

Page 6.9 Partial pressure measurement 404 6.91 General $04 6.92 Magnetic deflection mass spectrometers 405 6.93 The trochoidal mass spectrometer 408 6.94 The omegatron . • • • - 409 6.95 The Farvttron 410 6.96 the quadrupole 412 6.97 Tine-of-f light mass spectrometers 413

7. High vacuum technology 415 7.1 Criteria for selection of safeerials .....; 415 7-11 General 415 7.12 Mechanical strength ,...... ,. 415 7.13 Permeability to gases 41? 7.14 Vapour pressure and gas evolution 417 7.15 Working conditions 417 7.16 Metal vessels and pipes 418 7.17 Glass vessels and pipes 41? 7.18 Elastomer and plastic pipes 420 7.2 Cleaning techniques 422 7.21 Cleaning of metals 422 7.22 Cleaning of glass 428 7.2.3 Cleaning of ceramics 429 7.24 Cleaning of rubber 430 7.25 Baking 430 7.3 Sealing techniques 430 7.31 General, classification ,....,..,..... 430 7.32 Permanent seals 431 - Welded seals • < • 431 - Brazed seals 438 - Glass-glass aeals 446 - Glass-metal seals 449 - Ceramic-aetal seals 459 IX

Page 7.33 Semipermanent and demountable seals 460 - Waxed seals ..... 461 - Adheaives (Epoxy) 461 - Silver chloride 474 - Ground and lapped seals 475 Liquid seals 479 7.34 Gasket seals 481 Sealing mechanism 481 - 0-r iDg seals 493 - Assembly and maintenance of C ring seals 502 - Shear seals -- - 504 - Knife edge seals 505 - Guard vacuum in the seals 506 7.35 Electrical lead-throughs 508 7.36 Motion transmission 512 7.37 Material transfer into vacuum 518 - Cut-offs 513 - Stopcocks 520 - Valves 521 - Controlled leakB 526 - Vacuum locks ...... 526 7,4 Leak detection 531 7.41 Leak rate and detection 531 7.42 Leakage measurement 537 7.43 Leak location 543 7.44 Sealed unit testing 544 7.45 Sensitive leak detection methods 547 - Halogen leak detector 547 - Detec tora using vacuum gauges 548 Principle of operation 548 Single gauge detection 551 Barrier leak detection 552 Differencial leak detection 554 - Mass spectrometer leak detectors 554 - Ion pump as leak detector 555

8. Vacuum systems < 559 8-1 Basic criteria of design 559 8,2 Evaluation of the gas load 560 - Leakage 562 - Out gas sing • 566 X

- Faraaatloo...... 569 - fMf Irn r»mi1 •—ill • 572 8.3 Vacua ciuatcra 572 ft.* fuMelag coHbtnatlon* 573 8.5 Kulas for operating vacuus ayatcaa 576 - bfarmeaa «or W»a. 8.1 - 8.3) 579 - 243 -

* 5. PRODUCTION QF LOW PRESSURES

5,1 vacuum pumps

5.11 Principles of _pumpiiig,. Since vacuum technology extends on so oany ranges of pressure (Sec. 1,1), no single pump has yet been developed, which is able to pump down a vessel from atmospheric pressure to the high vacuua or ultra high vacuum range.

Although all the vacuum pumps are concerned with lowering the number of molecules present in the gas phase, several different prin­ ciples are involved in the various pumps which are used to attain lov pressures. Vacuum pumping is based on one or more of Che following principles: Compresslon-axnansion of the gas. In piston pumps, liquid column or liquid ring pumps, rotary pumps, Roots pumps; Drag by viscosity effects, in vapour ejector pumps; Drag by diffusion effects,in vapour diffusion pumps; Molecular drag, in molecular pumps Ionization effects, in Ion pumps Physical or chemical sorption in sorption pumps, cryopumps and gettering processes.

5.12 Parameters and classifications The selection of the pumping principle or of the pump to be used la defined by its specific parameters. The main parameters are: the lowest pressure, the pressure range» the pumping speed! the exhaust pressure, In the ultra-high vacuum range two other parameters are addedt the selectivity of the pump and the composition of the residual £*S.

* For a detailed treatment of the subject refer to; B.D. Power; High Vacuum Pumping Equipment, Chapman & Hall, London, 1966. - 244 -

The lowest pressure which can be achieved by a pump at its inlet* ia determined either by the leakage is the pump itself, or by the vapour pressure of the fluid utilised in the pimp. This pressure de­ termines the low pressure end of the pressure range in which the var­ ious pumping types sre effective (fig. 5.1).

Th* pwattg* range of a single pump is that rang* iQ which the pumping speed of that pump can be considered useful (Fig, 5.2). Pumps of the same type but of different sizes or constructions nay have ad­ jacent pressure ranges, so that the pressure range of a specific pum­ ping Method can be larger (Pig, 5.1.) than that of an individual pump (Fig. 5.2).

Tha puaping speed of the pumps Is not constsnt (as it was consi­ dered in Sec. 3.7), but is s function of the pressure. The pumping speed vs. pressure curve of pumps has either a shape of a curve decrea­ sing as the pressure decreases (e.g. rotary pumps), or of a curve in­ creasing first with decreasing pressure, reschlng a maximum and then decreasing as the pressure decreases (e.g. diffusion pumps* Root's pump).

The classification of the pumps, according to the pressure ran^e, la summarized In Fig, 5,1, while the typical variation of the pumping speed Is shown In Fig. 5.2, expressed as percent* of the maximum pum­ ping speed of each type of pump. RovfhvMwjcn HlgtivtOMMI ll*n-h*o*i IIOMB 0— hOwtfiwy | t°~i MM* 1 . MKutoM B-O DI^M M pumps

HLRWV «i«cMrpMnp«

V«Mx#pum» 1 DfNMton paw* G«WDJ*P.' » ' c^>i««„

1 1 1 1 n* m" •»' TO itr -w* KT* w* u» nr* -HT* nr* «r *r" u" nr* «*>

Pretajna ton

Fig. 5.1 Prrssure ranges of vacuum pumps.

io» HP to* to* io» 19" W it-« it» it-« s ^A^ ^A \ DtttMtofi a Slost»-M»0« 1 pump *- lOMMt rotmrww pun* V in wfthout PM biMH 1 —wW»t»t M»t- -A —-»0*J| bn \ \ / L flM»l wrt»y \ / \ IT \ -*>* 1 \ 1 y^(\\ 1 \ 10. ^i ^o- *Mt K,- ^i *-» *•« TC'

Fig, 5,2 Compnrntive pumping speeds of several pumps, In terms of their maximum speed. - 246 -

The exhaust pressure Is the pressure against which the pump nay b* operated. Fro* this point of view the vacuum pumpe may be broadly divided Into three classes: - Pumps which exhaust to atmosphere, usually known as roughing or imckiac pomp** The removal of the atmospheric air from the system to some acceptable operating pressure is refered to as roughing out the system. The maintenance of a required low pressure at the outlet of another pomp, is refered to as bacfclap. Hechanical rotary pumps, end ejectors are the typical roughing and backing pumps. - Pumps which exhaust only to sub-atmospheric pressures, require a backing pomp (In series) to exhaust to atmosphere. Diffusion, Root's pwps, and molecular drag pumps are of this type, which require a backing pump. - Pumps which immobilise the gases and vapours within the system re­ hire no outlet. These are the p ips based on ionization or on sorption. A typical laboratory vacuum system, with roughing and backing stages is shown in Pig. 3.35, Figure 5.3 show* an industrial vacuum system, in which the process chamber is maintained at a low pressure by a pumping system consisting of three vacuum pumps in seriest a 3- stage diffusion puep, backed by a Hoot's type, which is backed by a

rotary plunger pump. The rotary piston (Kinney) pump t being capable of unassisted discharge to atmosphere, is used initially to reduce the system pressure to about 10* Torr and then is used to back up the diffusion pump. The diffusion pump can reduce the pressure in the clean chamber to less than 10' Tort, but when the process la outgasslng or there is an sdmlssion of control gas, the pressure can be asln- talned at only about 10 Torr. Fig. 5.3, Schematic cross section of typical industrial vacuum system, and the graph of tlie pressure ,. o various points In the ays tern, 248

5.2 Hacbanical pumpa_.

5,21. Liquid pumps.

Host of the vacuum pumps using liquids to compress and exhaust have only historical Interest. He will mention here only the Sprengel, pomp, the vater-jet pump, and the Toepler pump, the later two being •till used in laboratory.

The Sprengel piap, has only the hiatorlcal Interest of being used In the first lamp factories.

This pump was hased on the principle whovn la Pig, 5.4. The mercury drops Introduced In the vertical capillary T, capture between them air bubble*. In this way the system evacuates air from the side tuba C and exhauata It through the mercury at the bottom, to the at­ mosphere*

The vater-jet pump la « familiar practice In laboratory work, especially In filtering operations. Water supplied from a fast- running tap Is fed Into the nozzle at A (Fig, 5.5). This water stream WJkTfR SUPPLY 'A

Fig, 5.4. Sprengel pump. Fig. 5.5. Water ejector pump. - 2M -

then emorged nt Mp.lt vrtncKy from ttu* conversion jnt B, Tlie- \rt li surrounded by a cono to prevent splnshinp, and also f.utde tlin water scream tfovn to waste nt C. A Bide tulx* n, 1B connected to the vc"-.«;«»l to be evacuated. Molecules oF the gas .-ire trapped by die UJph speed jet and forced out Into the atmosphere. Hy this means;, pressure: dixr. to 10-17 Torr are attalnahle, the ltnlt hoinp, due to the vapour prt"-.'-nr.- of the water {see TnMe. 2.2).

Kip,, 5.6. Toapler-pump, The principle of the Toepler pump Is fundamentally Che same as that applied by Tsrrlcelli In his famous experiment. The air from E (Fig, 5.6) is "pumped" by alternately raising and lowering the Mer­ cury reservoir R, which Is connected to the tube of barometric length placed below B. At each upward "stroke" the gas in B is closed from E and forced, through the tube F, into the atmosphere at M. Then, on the downward "stroke11, the pressure in E is lowered by expansion of the gas into B. The glass valve 6 prevents the mercury from entering the vessel £, in the upward stroke. With the Toepler pump, pressures down to 10** Torr can be obtained, except the mercury vapour pressure which fa about Ifl" Torr (see table 2.2).

The great disadvantage of the Toepler pump is its very low pumping speed. This was somehow increased recently in the "automati­ cally operated" Toepler pumps.

5,22. Piston pumps

The piston pump (Fig. 5,7) have valves so arranged that air is pumped out of vessel A. As the piston is raised from the lowest posi­ tion, the valve V- closes, and the motion of the piston then reduces the pressure in B. The pressure difference between A and B, will open valve V, and gas will pass from A to B. As the piston descends, the pressure In B increases, V. closes, V, opens and the gas in B escapes through V..

In one stroke the volume of gas V, Is expanded to V. + V_, thus the pressure is reduced from P to P.,

(5.1) -V2 y

Fig. 5,7, Piston pump principle.

and after n strokes to r .

!» i v*

Tlie minimum attainable pressure is limited especially by the dead space below the piston i.e. tlie sp*>ce between the valves V. and V_ (Fig. 5.7) vhen the piston is In Its lowest position. If V, repre* sents the dead volume, then Che minimum attainable pressure ia

Po - 760. Vd/Va (5.3) since at the end of the stroke the pressure in B oust be atmospheric in order to open V„,

Piston, pumps have V./V «• 1/6 - 1/10, thus their lowest pressure is- about 100 lorn - 252 -

5.23. Water ring pumpa.

Water ring pumps arc constituted by a miltl-blndc Impeller, which Is eccentrically mounted relative to the pump canine. (Fig. 5,8),

DkcMgi

MpMfer..

Sfctfl „

Dttdwrgt pwi

Suction pod _

Fig. 5.8. Cross section of a water-ring pump.

When the impeller rotates the liquid is thrown outwards to form a ring which rotates Inside the pump casing. The pockets between the blades of the impeller are completely filled with liquid when at the top position, but as the impeller rotates, the liquid moves away from the axis and draws gas through the suction port. AB rotation continues the liquid returns towards the axis and forces the gas out through the discharge port.

The sealing liquid, which is generally water is heated by the action of the pump. It is either run to waste and replaced or circu­ lated through a cooler.

Water ring pumps have ttNominal operating pressure of about 30 Torr, and are used in large systems where such pressures are sufficient. - 253 -

5.24, Rotatlnfl-vane pumps .

The rotatinR-vane pump, known also oi "rotary pimp", ti consti­ tuted of a stntor and an eccentric rotor which has two vanes (bladrs) in B diametral slot, (Figs, 5.9 and 5.10). Tl>e stfltor is a stee] cylinder the ends of whicli ore closed by suitable plates, which hold Che shaft of the rotor. The sttitor is pierced by the inlet and exhairt porta wllich are positioned respectively a few degrees on either side of tUe vertical. The inlet port is connected to the vacuum system by suitable tubulation provided usually with some kind of dust filter. The exhaust port is provided witli a valve, which may he a metal plate

moving vertically between arrester platest or a sheet of tteoprene, which is constrained to liinfie he twee n the stater and a metal hackLnR

Fig; 5.9 Cross section of a rotating-vane pump. - 254 -

Fig. 5.10, Exploded view of a rotating-vane pump. - 255 -

plnte.

The rotor consita of a steel cylinder mounted on n driving shaft. Ite axis is parallel to the axis of the Btntor, but Is displaced from this axis (eccentric), such that it makes contact with the top surface of the stator, the line of contact lying between the two ports. This line of contact known as the top seal between rotor and stator must have a clearance of 2-3 microns. A diametrical slot Is cut through the length of the rotor and carries the vanes. These are rectangular steel plates which rake a sliding fit in the rotor slot and are held apart by springs which ensure that the rounded ends of the vanes always make good contact with the stator wall. The whole of the stator- rotor assembly is submerged in a suitable oil.

The action of the pump Is shown in Fig. 5.11. As vane A passes the inlet port (Fig. 5.11.a), the vacuum system is connected to the =3pace limited by the stator, the top seal, the rotor and vane A. The

fa) (M

Fig. 5.11. Action of rotating-vane pump. - 256 -

volume of this space increases as the vine sweeps round, thus producing a pressure decrease in Che system. This continues until vane B passes the inlet port (Fig. 5.11.b), when the volume of the gas evacuated is Isolated between the two vanes. Further rotation sweeps the isolated gas around the stator until vane A passes the top seal {Fig. 5,ll.c). The gas is now held Between vane B and the top seal, and by further rotation it is compressed until the pressure is sufficient (about 850 Torr) to open the exhaust valve, and the gas is evacuated from the P<**P- Since both vanes operate, in one rotation of the rotor a volume of gas equal to twice that indicated In Fig. 5.11.b is displaced by the pump. Thus, the volume rate at which gas is swept round the pump* refered to as pwp displacement S is

St - 2V. n (5.A) where V is the volume between vanes A and B (fig. 5.11.b), and n is the number of rotations ps'r unit time (usually 350-700 r.p.m.)

The contacts of the vanes and rotor with the stator form three separate chambers each containing gas at different pressure. These contacts must therfore make vacuum-tight seals, especially for the top seal which must support more than one atmosphere pressure diffe­ rence. For this reason the inner surfaces of the stator that of the rotor and vanes are very carefully machined. Hence, great care must be taken to ensure that no abrasive aaterlal or gas which is Likely . to corrode the octal surfaces enters the pump chamber.

In theory, the lowest pressure achieved by the pump is deter­ mined only by the fact that the gas Is compressed into a snail but finite "dead volume". When the system pressure becomes so low that, at maximum compression, the gas pressure Is still less than that of the atmosphere it cannot be discharged from the pump. Subsequent pumping action reexpands and recompresses Che same gas without further decreasing the pressure in the system. The ratio of the exhaust pres- - 257

sure to the inlet pressure is termed the pomp compression ratio (aee also eq. 5.3) „ Thus, to produce pressures of the order of 10" Torr, pumps having compression ratios of the order of 10 are required. In addition to lubrication and sealing, the oil also performs the function of filling the dead volume, thus Increasing the compression ratio,

The lowest (ultimate) pressures achieved by single stage rotary —3 pumps is about 5 x 10 Torr, as measured by a Mc Leod gauge (permanent gas pressure). If the pressure is measured by a Plrani gauge (total —2 pressure), pressures of about 10 Torr will be recorded for the sane

single stage puap0 This higher reading is due to the vapour pressure of the sealing oil or its decomposition products in the pump.

Parallel connection of two Identical rotor-stator systems will

provide twice the displacement but the same ultimate pressure. Series connection provides the same displacement but greater pimping speeds

at low pressures and lower ultimate pressurea A two-stage pump way -A -3 reach 10 Torr

The pumping speed curves (Fig. 5.12)plotted for rotary pumps,

do show a fairly constant speed at the higher pressures (760-lOTorr)s but this speed falls off noticeably at the lover pressures and becomes aero at the ultimate pressure, as described by eq_. 3.255.

GaBii|ballast.when a rotary pump ia set to pump condensible vapour, like water vapour, the vapour is compressed and Its pressure is

increasedp thus It condenses, The liquid (wafer) mixes with the pump oil, and as oil circulates in the ptaap, it carries some of the contaninaCing liquid with it to the low pressure side where it will evaporate, limiting the attainable pressure. In order to avoid this

phenomena( Oaede (1935) provided the pump with a gaa ballast valve

(Flg0 5.13), which admitR a controlled and timed amount of air into the compression stage of the pump. This extra air is arranged to provide a compressed gas-vapour mixture, which reaches the ejection pressure before condensation of the water vapour takes place. The

principle of the gas-hanaat InH hown In Fig. 5. 13, where ABCDEPrapifeaent successive positions of the leading edge of vane V.

PRESSURE, IN rnRR

Fig. 5.12 Pumping speed-pressure curve of rotary pumps.

flo atmosphere

r^Oll level

ONt-VMV SAIL VALVE IBtUStiK GAS HOT' ftf-EKPCLUP THROUGH G«S SALLAsr IdtCT DURIHO COMPIIESSIDN

Fig, 5,13 The principle of the gas-ballast pump, and one way gas ballMt-valve. Consider a gas-ballat pump with: p. - the total pressure of the ballast air,

P. - the partial pressure of vapour in the ballat air

P - the partial pressure of the permanent gas (air) at the pump inlet,

P - the partial pressure of vapour (water) at the pump inlet,

P - the saturation vapour pressure of the vapour

P - the ejection pressure required to raise the exhaust valve against

the springg atmosphere and oil above it,

S - The pumping speed at the inlet

S, - the speed (rate) at which air Is admitted through the gas ballast

T - the pump temperature

T - the ambient temperature

C - compression ratio p - the density of the vapour at P

The compression ratio C 9 i»ea the ratio of the maximum Co the minimum swept volume between the rotor and the stator is

r Pe (5.5)

The maximum value that P.. can have without condensation of the vapour during compression, results from P (5.6) C -^- and from eqs. 5.5, and 5*6, It results

PB ' Pg (5.7) P -

For example if the pump temperature is 60* C, P • 150 Torr. With P - 1.4 atm - 1060 Torr, (5.7 e)

It follows Chat in this puntp condensation of water will occur if the partial pressure of water vapour at the pump inlet exceeds 16 per­ cent of the air pressure.

If gas ballast is used, the equality between eqs. 5.5 and 5>6, gives P P P, , S. , P P s g . h ~b a (5.8) *V * P -p s (p - P ) e s **e s

(5.9)

Equation 5e8 is slightly changed if the vapour content of the gas ballast (p,) and temperatures T, T are also considered. In this

v + y^.y^ (5.10)

The use of gas ballat increases the ultimate pressure of the pimps (Pig* 5,14), However this disadvantage is unimportant In practice because the gas ballet valve la ususally open only during the initial stages of pumping. . PRESSURE: (otr Fig, 5,14, Pumpinf, speed curves of Rome rotary pumps,

5,25. SlldioR-ynne pumpa.

These pumpa have a sinp.le vane which slides in a slot cut in the stator between the inlet *md exhaust ports. There are two types of this kind of pump (Figs. 5.15 nnd 5.1o).

In one of these types (Flfl, 5.15) the vane slides in its casing and on the excentric cylindrical rotor. The reciprocating vane mounted in the caning of the stator is maintained by springs in contact with the rotor, and provides a seal between inlet and outlet ports. - 262 -

Fig. 5.IS* Sliding vane pump. Vane sliding both in casing and on rot

Another type of sliding vane pu«p is shown in Fig. 5.16. In thia type the vane ia fixed hy a hearing tc th& outer sleeve of the rotor. The rotor rotates eccentrically* which Hakes the vane slide In Its slot in the casing.

The whole intttemhly is submerged In oil which completes the vacuiM seals and provide lubrication. The punning cycle is shown in Fig. 5*17. The volime of gas swept around the pump at each rota­ tion, la that between the stntor and the rotor at the tnatant when the rotor passes the vane slot. Fig. 5.16 Sliding vane pump, Vnne sliding in casing.

Fig. 5,17. Mode of action of sliding vane pump, a) induction; b) iso­ lation • c) compression; d) exhaust. 5,26 RoftinR-plunp.er pumps

In these puapa (Fig. 5,18) the sliding vane la replaced by a hollow tub« which la rigidly attached to the outer sleeve of the rotor. The tube rolls and slides In a bearing, and an appropriate hole cut In the side of the tube allows gas to be drawn Into the Inlet side of the puap.

These putps are designed for large pumping speeds. The heat of compression of the gas can ha considerable, so the stator Is usually provided with a cooling water jacket.

The shape of the pumping speed curves are similar to those of rotary vane puaps. (Fig, 5.14)

Fig. 5.18. The rototing-plunger pump, a)Induction; b) exhaust. 5.27. Roots pumps»

The Roots pump consists of two double-lobud Impel lr c. Flfi. 5.19). These arc rotated in opposite directions within tl'^ pnr. lioUHinp. Hie directions of rotation belnf, those shown l»y the .irrovp the Intake and exhaust will be as shown in FiR. 5.20,

DISCHARGE'

Fig. 5.19, Roots pump.

The impellers have identical cross sections and are dimensioned and arranged so that an enough large part o£ the surface of S, is a close fit to a part of the surface of H„ throughout the rotation.

The impellers are also a close fit inside the pump housing H (Fig, 5,19). The rotating impellers do not, however, touch one another, nor do they touch the houslnp,, hut there is a small clearance (about 0,1 mm) at the joints 1,2 nnd 3 (rig. 5,19). As point 1 moves around the Inside wall of- the pump housing, points 2 and 3 266 -

movs correspondingly (Fin* 3.20).

Fig, 5.20 Action of Root A pump..

Since tin? Inlet port is isolated In fact from the outlet hy a narrow gap {clearance between parts) there la a back flow of fias fro* the exhaust region to the Inlet region, and therefore the efficiency of compression In much lower than In the case of ail sealed pumps. However, the absence of rubbing contacts means that higher speeds of rotation (1000-3000 rptn) are possible, leading to much higher pimping speeds,

The conductance of the clearance gaps decreases as the average pressure in the puap falls; the pump efficiency is expected to Increase (Fig. 5,21), Maxima efficiency occurs when the pump la operated at a conpressloa- ratio of about 10 at a pressure of the order of 5 x 10~ Torr, and thus the putap must be provided with a suitable backing pump. - 267 -

20.000p

- 15.000 Is. I I- - 10.000 t) /p \ ' 5,000 • 1 \ : i X o|_ / ^_^ io - io' T7P ~W ~< io1 io

Fig. 5.21. Typical pumping speed curve of RnotB pump.

5.28. Molecular pumpB.

The principle of the molecular drnp; pump, based on the direc­ tional velocity imparted to gas molecules which strike a fastmovinp. surface is described in Sec. 2.62. This principle Is applied In modern pumps which contain alternate axial stages of rotating and stationary discs and plates. The discs and plates (Fig. 5.22) are cut with slots (Fig, 5.23) net at an angle so that gas molecules caught in the slots of the moving disc are projected preferentially in the directions of the slots in the stationary plates. The runnlnn clearances between the rotating and stationary plates are of the order of 1 mm. The rotational speed for a pump having a rotor dia­ meter of about 17 cm, is IfiOOO rpn.

Although variation of the pitch angle of the slots varies the zero flow compression ratio and pumping speed, a pitch angle of 20* appears to be a good compromise for many applications. Since a compression ratio per stage of about 5 can he achieved, A pump having 9 stages Bhould maintain a zero-flow compression ratio of the order S^BL

Fig. 5.22, Molecular pump (A. Pfelffer).

S3 1

1 1

Fig. 5,23. Detailct of the rotor and a tat or plates of the molecular nump (Fig. 5.22). - 269 -

of 5 : 2 x 10 , For this compression ratio, the pumping speed in

constant below 10~ Tort (Fig. 5.24), but above l(T2Torr the speed dependB upon tlie aize of the backing pump. The dotted line indicate* the theoretical Hpeed for air only. Normally the presence of hydro­ gen, which back diffuses, limits the total ultimate pressure to abouq -9 10 Torr, although the maximum speed for hydrogen ia some 20 per cent more than that for air. The great advantage of molecular pumps compared to diffusion pumps is that molecular pumps are free of (hydrocarbon) vapoutB.

Fig, 5,24. Pumping speed of Pfeiffer molecular pump for air. 3 3 I) with 45m /hr backing pump; II) with 10m /hr hacking pump. •

5.3, Vapour pumps«

5.31. Classification

The general term of vapour pump applies to ejector pumps as well as to diffusion pumps. The distinction between these two kinds of pumps la described fundamentally, by considering the mean free path of the gas molecules at the Intake port (nouth) of the pump, in relation to the throat width (nozzle clearance Fig. 5.25). H»t 1 VACUUM wmj

( *-""1 ^ WWt*CCWW3fct w

WRSWMJET—<( ^ -^

1

Sl-SIEM rat J mi CASWD 'W,^ f "«J f n i DirFtisiort rniMP (b) l/AFOHfl FACTOR PlffliP

Fig. 5.25. Vapour pumps.

Both the e jtscttfr and the diffusion pump have at their base a boiler (heater) which supplies the vapour (e.g.ail). In the diffusion puwp, (Fig, 5,25) the vapour (oil or mercury) travel up the chimney and is deflected by the umbrella placed at the top of the chimney. The molecules of the vapour stream collide with the gas molecules entering through the intake port. Since the mean free path A of the

gas molecules is Greater than the throat width t_t the interaction between gas and vapour 1B based on diffuaionCsee Sec, 2.52), which is responsible for the dras of the gas molecules towards the fore-presslon region. This effect establishes a pressure gradient between the high and fore vacuum sides.

In the ejector pump (Fig, 5,25,b)t the mean free path A of the gas molecules at the intake Is less than the clearance t_, ThUB the gas is entrained by the viscous drag and turbulent »l»li»g **ich carries the gas (at high speeds) down the pump chamber of diminishing cross section and through an orifice near the fore-vacuum side. - 271 -

Combinations of the diffusion and ejector principles, are encountered in diffusion-elector pumps, sometimes called vapour booster pumps.

5,32. Vapour ejector pumps.

Ejector pumps work with oil vapour or steam. Figure 5.26 shows the diagram of an oil ejector pump. The pump fZuid is contained in the boiler (10) and the vapour flows through the jet chimneys (8,9) into the nozzles (2,4)„ Due to their special shape each of these nozzles produces a supersonic jet, which enters the nozzles (diffusers 3,5) and condenses on their cooled walls„ The air to be pumped enters the pump through the high vacuum connection (1), and is carried with the jet and compressed,, The process is repeated in the second

stage0 A compression ratio of about 10 is achieved in each stage. The air compressed to a pressure equivalent to that of the fore- vacuum line (6) is removed by the backing pump. The condensed fluid flows back through the return pipe (7) to the boiler. The fluid co­ lumn in the return pipe counter balances the vapour pressure in the boiler.

Oil vapour ejector pumps are constructed of glass (small sizes), for pumping speeds of a few liter/see at 10 Torr, and heaters of hundreds of watts (Fig* 5,27), Metal constructions reach pumping speedB of thousand (s) of liter/sec at 10 - 10~ Torr, and have

heaters of many kilowattse Figure 5.28 shows the pumping speed curve of two oil vapour ejectors (9 B3, 18B3) compared to a 9 in diffussion pump (903B) and a rotary pump (100 cu ft/min). 1 HBHMWUUMCONM z aeOETCR N02ZIC. J ni»T HozzLe 4 BOOSTER N02ZLI S prroT NOZZIC. 6 FORE-VACUUM CONN 7 PUMPFLUOBETUftNnPt * jfir CHIMNEY JIT CHIMKlKV K> BOILER

® ©

Fig. 5.26. Double stage oil-vapour ejector pump. - 273 -

INTAKE. PORT

— LAGGING

Fig. 5.27. (;lass oil vapour ejector pump.

1000 yf- 903 B-\ Hog \~-y 2iooo 1 H V-18B3 =: KH3 z W983 sM0 If KOMIIY \ "* «0 J/ VACUUM fUMP^ > ^ V^— •v^N KTl 10-* PRESSURE, IN TORR

Fig. 5.28, Pumping speed curves of oil vapour ejector pumps (9B3, and 18B3). - 274 -

sti-.iw ejector puetpa are able to produce rough vacua* at high f't.Bj.imt 8f».>edii„ A four stags stew ejector In able to product 0.5 u

Fig. 5.29. Typical item ejectors.

In th« tingle *tage of a steam ejector

5.33 Diffusion puypa.

Puaiplnp speed.In a diffusion pump the puaping speed is determined by the size of the intake clearance and the Ho-factor. The area A 2 (cm ) of the Intake annulus Is (Fig. 5.25a):

(5.11) where d^ is the diameter of the intake port, and 4&is the throat width.

In accordance with eqs 2.48 it is impossible for a gas of molecular weight M and temperatire T to pass through this area at a flow rate exceeding 1k >.«fl A liter/sec. or for air at 20 C exceeding

S - 11„6 A Liter/sece max

The ratio between the admittance (i„e„ the true pumping speed S across the throat of the pump) and the maximum flow rate S , is

0.45„

The pumping speed of the pump is thus given by

S « H.S - 3.64|^-| H ~ t (2d - t) (5.12)

With H « 0,4 and t - d/3 (large diffusion pumps) the pumping speed for a

S - I1'6 B 0,4 ^- (2--|)d2 Z 2d2 liter/sec (5.13) where d is the diameter of the intake port. A pump with a pumping Bpeed of 1000 liter/sec would need to have an inlet port diameter of

1 " 1000 I ! d - • 22,4 cm 9 in and thia is about the diameter of diffusion pumpa having auch pumping speeds« - 276 -

Assuming that the Mo-coefficient la independent of the molecular weight of the gea being pumped* eq. 5.12 implies that the pumping speed of a diffusion pump should be inversely proportional to the aquare root of the molecular weight of the gaa. This proves to be approximately true for BOB* pump designs.

Equation 5.12. also implies that the pumping speed of a diffusion puap is independent o£ the preaaura. This la indeed the case for a range of preaaurea of away deeedea, the pumping speed curve being as shows in Fig. 5.30, At the maximum pressure at which the vapour pumping action begins (A„ Fig, 5,30} this action reduces the system

10.000

5 1,000 s 5 100 I t- - — -N 10 1 _ _ __ w1 io-* «r» JO** ior* ID* 10"' Pwssur*, lotr

Fig, 5,30, typical diffusion pump characteristics, pressure, the* decreasing the density of gas Molecules entering the vapor stream. Thla in turn reduces back-diffusion and hence the pumping speed ris^s with decreasing pressure. The pumping speed conti­ nues to increase until the pressure la such (B, fig, 5*30) that the rate of back-diffusion at the top jet. is not determined by the inlet gas density but by the rate at which gaa, la removed from the jet. At this, and lover system pressures* the speed remains Constant at a maximum value of S,

the ultimate pressure obtainable is theoretically determined by the vapour pressure o£ the pump fluid. With additional refrigerated baffles much lower pressures can be obtained, In practice however the ultimate pressure is governed by the characteristics of the system, lit that the pumping speed In the system must necessarily become zero when the gas load (leaks, outgassiug) is equal to the maximum rate at which the pump can hanxtle gas. This Is Illustrated by Fig. 5.30. Curve &BCD allows the theoretical characteristic, while curve ABCE shows a typical practical characteristic in which Che pumping speed becomes very small at 5 x 10~ Torr, when the throughput is 5 x 10~ Tort, lit/sec. The characteristic of a larger pump (S-10,000 lit/sec) Is

also shownB The section AS of the characteristic is reasonable linear on a log plot and hence may be represented by the eaplrical equation

( P \z (5.13)

where S is the maximum pumping speed (B,C), while p is the pressure corresponding to point B„ The slope Z takes values between 0,8 and 1. The section B C E can be represented hy an equation similar to that governing the variation of a rotary pump (eq, 3.255)

S - Sm C 1 -~) (P

where p Is the ultimate pressure (point E),

Plfflip sizes* A wide range of sizes of diffusion pumps are available with Inlet port diameters from 1 in to 36 in., with corres­ ponding pumping speeds from about 10 lit/sec up to about 45000 lit/sec. Small sizes o£ diffusion pumps are also constructed of glass.

Roughing and backing. The single jet pump does not function very efficiently in practice. For efficient operation two conditions must be fulfilled: - The system pressure must be initially reduced below a certain value (roughing)which tti most practical cases. Is of the order of 10"* Torr. - The pressure below the jet must be kept reasonably low (backing - 278 -

puap) to reduce the probability of back-diffusion.

To Hake these conditions easier to achieve by the external roughing and backing pumps, vapour pumps are constructed with several jet stages In aeries, one acting am a backing pump to another. The •win function of the top jet (Fig. 5.31) is to give a large pumping speed and thus this jet has a large admittance area. On the other hand the lower jets have smaller admittance areas, and hence smaller escape areas for the vapour stream. Consequently the pumping speeds of the lower Jets become successively smaller whilst the pressure dif­ ferences which they can support hecome larger* It should be noticed that the throughput la necessarily constant throughout the pump.

Speed Prey ire »0 I s«»e 1 5 x 11 ' ion

f'i baching' pump : / \ J

Fig. 5.31, Multistage pump. J'miip fluida. Diffusion pumps Use cither mercury or organic n»f'i- Table 5.1 gives data on several of the organic Fluids used. T.ihie 5.? lists tUe properties required from these fluids and the extent tn which they are satisfied.

Table 5,1

t-lufe for aw In tipour stream pumps ,- 1 *••*—'! xr.££! ?

owl; ""'i*; • ta:­ XIW"11 \* JI ire •• 0*11 il . 10 ' .. SM* £ SKSS? tKSia iif! 1 (MM > . 10-' Hpanaro Nareuil Hi* Man-nil J"* .».»• I)d ml' S3' r.iidcoor^Hnii ,><*)•* OtfWtS' fl-HOi s SMitntif l>TJni' MOW J10 101 ;:r Silii-Piit I1"TO)' if"" «CIW»f.iW»f*». m - «»-

Mil*. 5.2

H i MJNIIVMI at* *f ffWM ftr IWI fa VlfNMP |MM|Nf

Xmwh HM*I frmw !••* (wr to Ifr-Mim t»iiw5^~

j Mtrt*>w»l|Jnmfcin HUthw iitowkhwt,

•lb* w wtf rwtfww pmf'

ificSI]I l»MI HHMI 1* lM* *fcritt.w am w»dfofhWf»ihn pmmimvitt CMMMV ti WMHiMiiiiiMitwftKi MJIilWtiWfDimMtWIIHMKIlMllMnllWIMMItvMiaiMNllIlMmit cnmlwlMtr, **, rtwntwlww. trw, kmi, tirw mi tin ft mta&tti by nit, hH thf IMHIMMI h*tl»n4 nk*»l mm*. < WvwWn* nwl wft-MUrrhti ihnulil tfHnftm ht *wW*J tn pump Jnh*ii> wtd JH u*(tmt. CVIWMHH «W HMMI

AM tkUt iu«l lit prat* » dKMwtlllun (a birth emt, HM » MCIM» dw w. hMk>ii(w«MthitrtirifiiR<«r(nntri({*it|iwMnrfeM))Piel*c«llK)iliind coW<*i»i«l« iirfM art HfrHci I* lir* mxi,

()f HwlJt rtnwn In imvlnw UM*i Mtr> KIJMMI k AMy IHHWWH'I''

KM nth OlMnlw IMH and vMtwn In MKM tit* IHtiiiira nMMm bm-jM nTtwtk (IIRMIM '" rtuldi HIK) IN ftHnrwUii (ml In ntMf af irttfarfihi tnmmtMtytmMUtrt In vKuwn H«tki, Untt m*\, hnwtmt wHK tabum. Ml* WWMIII,

iiM t« MW *W*M* «

Otiiittiibk vriili MII ihiiik IMM. vkMMiir riwioi »wnlWi*«»U'C »«C .(f. nrci .. ll'C ss » •ffift, - 281 -

The relative advantages and disadvantages of the varioun oils result from Table 5,2. A comparlfion between oils and mercury la shown In Table 5.3.

Table 5.3 - Comparison between mercury and oils.

• Advuntngia disadvantages

An clement—cAnnol dccnmptnc Nigh vapour pressure: MB y. ID a turr Ones nnt oxidize when hcnied lit Atmo­ n( 20' V spheric pressure uulcus henting k violent Liquid nir imp or inevitably low mndiict- and prolonged fancc Is essential rT dw total prevnrre ft Dues nut react willi or dissolve commonly to be attained CIKOUII It red gas an J vapours Amalgamates wiih many metals; restricts ivasily purified by distilliitinn chttlec of puinp cimslruclioa matcml; High density facililntcs iclurn lo Mgh pres­ steel and glass is used Tar pump jjcfccti sure boiler and jet system* High fore-pressures -up to 40 torr can be Sensitive as a pump fluid In small trace* used in nuiItUslagc pump of oil or ircisc

Advantages Disadvantages

Low vapour pressures nbttiin- Complex compounds or mixtures of compounds: de­ able: 10 * lorr passIMc compose slightly on heating i/t vaata in baiter Cold trap unnecessary jn Except Tor silicones ind Narcoil, readily nsklircd on mnny vacuum processes heating fit air High conductance haflle Dissolve all kinds of gases and vapours to form solutions which prevents most of and nrctrnfaiWc compounds, Dccnrapoyiioa acceler­ back-slrcnmlng is readily ated by catalytic action of some metals, cspccURy used copper Inert lo metals chosen for Decompose nn hot filaments, and in electric diacharae. pumu construction

Fractlpnatlnfl PUMPS. If the flump oil is a mixture of chemical coMpound* the interest .is to separate them so that the top jst Include those o£ the lowest vapour pressure. This la achieved by Incorporating into the design the principle of fractionation (Fig, 5.32), Fluid returning from the jets to the boiler flows radially Inward towards

Fig, 5.32 three-stage fractionating oil diffusion sump. the center of the boiler. If this flow is impeded &y barriers with snail openings (shaded parts on base plate. Fig. 5.32), the fluid is heated substantially while It la still near the outer portion of the boiler, so that high vapour pressure constituents are boiled off near the outside.

As the fluid flows coward the center it Is further heated and - 283 -

lower vapour pressure components are vaporized. The jet system Is constructed of concentric tubes arranged such that each jet receives vapour from only a specific annular region of the boiler. The backing (lowest jet) receives vapour from the outer portion of the boiler where

the vapour pressure is highestt and the top jet receives vapour from the central section where the vapour pressure of the fluid is the lowest «

Coolings The cooling of the diffusion pump is a part of its working principle, as is the heatings Diffusion pumps are usually water-cooled, buf air-cooled pumps are also available (small size pumps). The cooling should be arranged so that the coolest region is where the vapour stream strikes the pump wall. The rate of cooling is also fairly critical since if it is too low the vapour is not entirely condensed and thus the back-streaming effect is increased. On the other hand, if the cooling rate is too high the vapour is not only condensed but also considerably cooled. This results in a slow flow back to the boiler, and necessitates a high boiler power to re- evaporate it.

Back-streaming and back-migration, (see also Sec, 5,65) Dif­ fusion pumps suffer from two defects wherby the pump fluid enters the vacuum enclosure: the back-streaming, and the back-migration.

Back-streaming is due to a small fraction of the molecules of the working fluid (oil or mercury) travelling from the top jet in the wrong direction (towards the vacuum chamber). This undesired direction is either imparted to the molecules as they issue from the jet or it is acquired after impacts with gas molecules or with other vapour molecules in the stream, A well designed nozzle should be dry during pumping.

Back-migration is due to re-evaporation of the working fluid from the walls of the pump and the connecting tube to the vacuum chamber.

Back-streaming and migration can be decreased by using baffles and cold trapa.

Characteristic curves. The characteristic curves of diffusion puapa arc generally plotted In the form of pumping speed (lit/sec) against the fine aide pressure, Figura 5.33 shows some at such curves.

MttsiiM, in TORS trump iinMdfcti; typical rcsutfsar e §iwn; speeds wil! ttepeml lo some csicni 011 Hie fcniter deafer w;Mfci$« ami the pump fluid used) MC 3000: Consolidated lilcclrudynumics oil diffusion pump, 14-in. I.D. Inliihc port, Octail S Pfi: Vacuum Imiwstriitl Applications oi! dilFuatafl pump, 6l-m. I.D. intake port, Silicone 703 OT 400: leybold oil diffusion pump, 5 in. I.O. intake port, Lcybatdoil f-' D1FP 250: Baker; oil diffusion pump O 203: BJwurdit High VacBtmi Ltd. oil diffusion pump, 2|-in. !,n, iiiliikc port, Silicorw 702 F 203: scCr-purlfymjt version nf O 20.1 (see Fig. 2.22) O 102: lldwirds High Vacuum f.lri. off diffusion pump. MM. I.D. »*&« P»ri- SHIcodc 702

Pig. 5,33. Pumping speed-fine side pressure curves of diffusion pumps.

The diffusion puup must always be backed by a forepump, and the chrpuRhput of the forepump must be equal or larger than that of the diffusion pump. Pumping Bpeed Li-1 \ \ \ ^^^ \*rf *~ \ ^ \\ \ \ /'^v \ \ \ \ \ V \ \ \ \ \

Throughput torr.Li-1

10 10

Fig, 5.34, Speed and throughput curves for the same diffusion pump. - 286 -

For the MM on of ehocslttg the aproprlate backing pump, it la useful to plot the characteristic earn of th« diffusion pump la C«M of chroughput-preeaur*. Figure 5.34 thous auch a pair of curve*.

5.4, Ion-pumps

5.41. Classification

If a gaa la Ionized and the resulting positive ions are attracted to a negatively charged plate, atoms of the gas are effectively removed from the system, and thus a pumping action la produced. The general term "ion-pomp" Includes thoae vacuo* pumps in which gas molecules are pumped by being ionized and transported In the desired direction by an electric field. The ionization may be produced by collisions of the gaa molecules uith electron* emitted either from a hot filament or from a cold cathode discharge. The first type of pump ia refered to as hot-cathode ion pump .while those belonging to the second type arc known ae cold cathode Ion pumpa. Both the hot and cold cathode ion gauges (Section 6.7) act as pumps In this manner, and can be used to pump small dosed systems, which have been previously evacuated to about 10~ Terr. Pumping speeds for hot cathode gauges are typically of the order of 10~ liter/sec. While cold cathode gauges can reach about 1 llt/aec.

In order to increase the pumping efficiency sorption and getterlng phenomena ate combined with ionization. Pumps which combine the action of an ion pump and the sorption of the ions in a sorbent, are known as lon-aorptlon pumpa. lon-aorptlon pomps in which a garter* is contlnously or intermitently vaporized and condensed on the trapping surface to give s fresh deposit of sorbent, are termed getter-Ion pumpa. The vaporization is due to thermal evaporation In evapor-ion pumps and to cathode sputtering** In sputter-ion pumps. Although ion pumps

* Getters a material which is included in a vacuum device (cube) for removing gaa by sorption (see Sec, 5,7), **SputtcrlngKproccss of ejecting atens of a cathode by bombarding It with heavy positive ions. without lorptlon or gettering vera alao constructed, ccccuerclal piraps ate either of the evapor-lon or sputter-ion type,

5.42 lon-pwping.

An electrical gas discharge, in vhleh ions are formed, la basi­ cally capable of pumping gases, and one assumes that the formed Ions are either bombarded into a metallic collector provided for the purpose or that these ions are trapped vithin the surface atoms of such a collector, due to a chemlsorption effect. The pumping efficiency is expressed by the ratio 1 (? between the ion current and the pressure oi the gas in the device. The ion current 1 is proportional to the number of molecules entering the device per unit time, thus to the throughput Q. Therefore

i Q (5,15)

and the pumping speed S is given, by

S - S liter/sec (5.16) P

ere B » j£ ls the constant 0j the pimp expressed in lorr.licer/sec.

Ampere, i is the lea current (Ampere), P is the pressure Cforr).

The value of the pump constant B was determined by plotting the pumping sfseed S (for a determined gas), as s function of the sensitivity i+/P (Fig* 5.35).

The pump constant Bof practical pumps was found to be about

0-0007, while the maximum value of this constant is

•- 0.191 (5.17) io4 ' ios lit/a

Ftg. 5.35. Pumping speed VB sensitivity of sputter-ion pumpa. v/liere; e - clmtfiC of nn electron • 1.6 x 1(1-I T A nee n - number of molecules In 1 liter gas, nt P • 1 torr, 20*c,

Unlike tlie dlfFualon pump, the Ion pumps (evapor-lon and sputter Ion) does not require a forcpump to pump the collected gas up to atmospheric pressure, since the pumped RBB IS In effect digested. However, an Auxiliary pump In needed to reduce the system pressure -3 -4 to the ranee of 10 - 10 Torr, where the Ion-pumping action will commence. Mechanical or sorption pumps are usually used for this initial purapdown. - 289 -

5,43 Evapcr-ion pumps

Evapor-ion pumps combine Che ion-pumping effect with the gettering process of evaporated active metal. The gettering effect is used both during evaporation (dispersal gettering), and in the form of a fresh film on a surface (contact gettering). The gas is ionized to ensure transport by electrical pumping of the inert gases (which are not gettered) to the getter-coated vail at which they are made to arrive with energies of a few hundred electron volts. At these energies, about 20X of the ions are retained, and embedded in the film as fresh getter is vaporized. The most widely employed getter in these pumps is tita­ nium.

From the various types of evapor-ion pumps we will describe here: a) The small pumps, designed to perform a single pumping operation. b) The large high speed pumps with continuous feed of titanium.

c) The Orbitron pump.

Small evapor-ion pumps„

These pumps (FigD 5.36) consist typically of a hot cathode ioni­ zation gauge of the Bayard-Alpert type (see Sec. 6.7) containing an extra filament around which is wrapped a wire of getter material (usually titanium or Zirconium),

Usually, these pumps are first evacuated to diffusion pump pres- —3 sures (below 10 Torr), On firing the titanium getter, sorption of the common gases takes place; on operating the ionization gauge, positive

ions of the gas, and in particular the inert rare gases, are driven Into the titanium-coated wallB, where 10 - 20Z of those incident are retained and buried In any subsequently deposited titanium. A puaping speed of about 5 liter/sec for air is possible and ultimate pressures -9 of 10 Torr are attainable. A typical application of such pumps is the improvement of the residual vacuum in special vacuum tubes. An auxiliary small getter-ion pump in a glass bulb is attached to the main tube, *" After sealing-off the na'in tube from the conventional pumping system at a pressure of about - 290 -

KM OXLECMQ CUTCTROOE ro*MCD IV CWMRATCO TlftUHUW

O 10 3D 30 AO TIME. IN MINUTES

Fig. 5.36 Small evapor-ion pump.

10~ Torr, the petter-ion pump (evapor-ion pump) is operated to reduce the pressur to in" Torr. Tfie evapor-ion pump is sealcd-off, and discarded.

Large cvapor-ion pumps.

The construction of such pumps is schematically shown in Fig. 5.37. A spool, carrying titanium wire, ia externally controlled so that the wire is Led downwards onto a post of refractory conducting material maintained at 1000 V positive with respect to the pump wall. Electrons produced at the circular filament (100V positive with respect to the rail) bombard the post and heat it to about 2000'C. This causes rapid evaporation of the titanium wire which then condenses on the cooled pump walla. This continuous evaporation of the wire insures s conti­ nuous pumping action both by dispersal and contact gettering. A wire mesh grid, at a potentlnj of 1000 V positive (with respect to Wire feed spool

Fig. 5,37 Evapor-ion pump, the valla) also attracts electrons from the filament and these cause Ionization of the gas, resulting in positive ions travelling to and being retained by the pump walls, Using a titanium evaporation rate of about 5mg/min, pumping speeds are of the order of 3000 1/s for H_ at 10-6 Torrj 2000 1/s for N at 3 x 10~6 Torr; 1000 1/s for 0. at -5 -5 10 Torrj and 5 1/a for argon at 5 x 10 Torr, Satisfactory maintenance-free life times of at least 1000 hours for these pumps are recorded. Xlie component most likely to fail early. la the titanium feed device due to built-up of tltanlua on the tip of the wire guide, Difficulty is also experienced due to peeling of the titanium film on the pump vails after repeated evaporations, especially If air at atmospheric pressures la frequently admitted. the large titanium evapor-ion pomp has been applied in those cases where a high pumping speed, coupled with oil vapour free residual - 292 -

-6 **7 glues at low pressure (10 - 10 Torr) are required in cliamfiers of considerable alees, such as accelerators, larjie X-ray tubea etc. The nost extensive use of the evnpor-ion pump is on the 30 BR V (30 x in e V) proton-synchrotron, tic tlrookhaven, where over 50 onltR have lieen in use for several years, Ihe_ Qrbltron pump. The Orfiitron Is a device in which electrons are injected inCo an electrostatic field between two concentric cylinders, the central cylinder being the anode and the outside cylinder being grounded. Electrons injected Into this field with sufficient anpular momentum, have paths of several tons of meters, thus a hifjh efficiency to ionise inert gases.

V ->SK.

Fig. 5,38 The Orbitron pump principle. - 293 -

The central cylinder (Fir.. 5.38) i.q a tungsten rod of sra.ill dln- mftCer supporting titanium cylinders oF relatively larp.e diameters. Mont of tlie electrons ore collected by the titanium cylinders, because they intercept electrons of relatively litf.h annular momentum, and the titanium is heated to sublimation temperatures. Active gases are pumped by the fresh titanium on the walla of the pump. Inert p.nscs are ionized In the electrostatic field, driven to the walls, and hurried by fresh titanium.

The electrons must lie injected with an anj*ulnr momentum such that they move on orbits between anode and cathode. For this purpose tungsten filaments are placed parallel tn the cylinder axis at a dis­ tance of about 1/3 of the radius of the cathode. One of the lead-w1re*s of the filaments 1B placed to he parallel with the filament and is situated between the filament nnd the .-made.. This arrangement avoid? direct path of electrons to the anode. This arrangement avoids direct path of electrons to the anode. Tn order to r.ive a slight axial compo­ nent to the electrons in their orbiting motion, a plate Is provided (it tlie upper end this plate is at the same potential as the filaments.

Fig, 5.39. Pumping speed of Orbitron pump. - 294 -

The pumping speed of en Orbitron pump la shown In Fig, 5,39. The ultimate pressure Is of the order of 10* Torr. It Is difficult to obtain lower pressures, since the outgassing rate of the hot titanium is relatively high.

5,44. Sputter-Ion puapa.

The sputter ion pumps are designed such thft ut electrical discharge occurs between the anode and the catode at a potential of several thousands of volts in a magnetic field of a few thousand gauss. Since the magnetic field causes the electrons to follow a flat helical path, the length of their path is greatly increased. A. high efficiency of ion formation down to pressures of 10 Torr and less is assured by this long path length. The gaseous ions so formed are accelerated to the titanium cathode, where they are either captured, or chemisorbed. Due to the high energiesthey are propelled into the cathode plate and sputter cathode material (titanium) some of which settles on the surfaces of the anode where It also traps gas atoms.

Sputter-ion pumps consist essentially of a stainless steel vessel, containing an anode of honeycomb construction (Fig. 5.40), and a tita­ nium cathode mounted opposite each end of the anode. A potential of about 3000V is maintained between the etetrodes and a magnetic field of about 1500 G is applied by external permanent magnets along the axis of the electrode system. Positive ions of System gas which are formed in the region of the electrodes are accelerated to the cathode and acquire sufficient energy to sputter titanium.. The sputtered tita­ nium condenses mainly on the open structure anode and in so doing pumps active gases by both dispersal and contact gettaring. Gas mole­ cules which reach the anode by either of these processes are rapidly burrled beneath succeeding layers of titanium and are thus permanently removed from the system, On the other hand, gma which reaches the cathode as positive ions has a high probability of being desorbed by succeeding ion bombardment. This is particularly so in the case of 295 -

Titanium cathodes

Fig. 5.40 Sputter-ion pump (principle).

the inert gases since they can only be ion-pumped and then held at the cathode by the relatively weak forces of physical adsorption. Sur­ prisingly, helium, for which normal sorption by any material is insig­ nificant, la pumped quite well* apparently by being rather hurried in the cathode material. Argon Is the most troubleaome in this respect (Fig, 5.41) and is the main factor governing the ultimate pressure attainable. The lew pumping speed for argon is frequently of concern, since the atmosphere contains 1 percent argon. In addition to their poor argon pumping speed, simple diode sputter-ion pumps exhibit re­ gular pressure bursts (Fig. 5.42), at time intervale of several minutes.

In order to obtain stable pumping for argon, either a third element, the sputter cathode is Interposed, or the pump cathode Is slotted. In the first solution an electrode In form of a grid (cathode " " lyUtegen n ~- _ A V T'i :: Nitrogen r.l _ ; s A \ Oxygen j

- •> Argon *

Fig. 5.41 Characteristics of a sputter-ion pump (5 17sec).

| mo-1 £ i*io'*-

Fig. 5,42. Pressure vs. time curve of a sputter-ion pump exhibiting the argon Instability. - 29? -

Fig. 5#43) ia incorporated between the anode and the outer plate elec­ trode, euch that the new RTid becomes the true cathode and the side plates become auxiliary electrodes (Fig. 5.A3). This arrangement Is refered to as the trlode-pump.

Ita principle ia that titanium is preferentially sputtered fran the cathode and ion burial and noble gas pumping occurs on the auxi­ liary electrode.

CalMe "°-*§ B11 0 0 011 B B11 ^

'0^ V J / -Co Sputtc atoosot ^

Fig. 5.63. Triode ion-pump.

Blhiialnognttic IsM)

Fig. 5.44, Slotted cathode diode pump. _ 298 -

The triad* design raeylte Is a stable pumping of argon* but bring• CD a 1MS la tb« cathode Ufa tleme, or to a reduction of tha pumping speed for ether gases. The slotted cathodes in tha diode pump (Mf. 5.44) appear to brine a bat tar lolutloo to tha airgon Instability.

A further difficulty with ion-pumpe, is tha cara which wtt ha —2 taken whan starting tha pump at high presences (mat. ID Torr). the Ion current at high pressures la large and causes heating of tha pump. If tha pump has previously handled much gaa the temperature rlae leads to outgaaalng which In turn cauaca a larger Ion current. Such e process rapidly becomes Mrua-a»ayn leading to glow discharge between the elec­ trode* and a rapid riae in system prasaura. Even if the gas evolution la not rapid the pumping speed la reduced giving what is termed "alow starting**, these troubles can ha largely overcome, be initially puaping -4 to pressures of the order of 5 x 10 Torr before operating the sptitter- iou pomp.

Sputter-ion pumps are available (Fig. 5.45) from 1 liter/sec to 5000 liter/see* Tha Ufa of a sputter-Ion punp la limited by sa­ turation of the titanium or build-up of eputtared material (flaking). Commercial pumps ar* quoted aa having a life up to 50,000 h at a pres­ sure of W* Torr, bat at 10"5 Torr the life la only 3000 - 5000 h. If pressure burets occur the life time of the pump may be considerably less,

5*5 Sorption pumps.

5,51. Mature of sorption pumpine,.

The sorption samp consists of a refrlgesated enclosure containing

an activated sorbent (Sec. 4.36), On opening tha pump to tha ayatemt gaa im sorbed until tha sorbent is saturated, therefore sorption pumping la a batch ngpeeaa.

Tha materials used aa aorbtntt in commercial sorption pumps are the zeolitei (Sac. 4.36) also known as 'molecular slavee". The ability 299 -

Fig. 5.45. Ccmercial aputtar-lon puap.

of zaolitaa to abaorb largo quantitiaa of gaaea reata on thalr unique porouf cryatal structure. The atructurc of aoltcular a Uvea la made of vary fin* roughly apherical cavitlai, connected by minute c* annals, of about ttia ease diameter aa that of ga* molaeulae (Tabla 2.5). The ratloa batwacn the diameters of the channeLa and the, molecular £ia*et4r' of varloua gasee may make highly aelectlve tha aorption pumping proccifit of a gea mixture. - 300 -

The adsorption character la tie* at a typical Molecular slave (wltli • Mao pore aise of 5 A*) la shown in Pig. 5.46. The curves nhow the euantlty of gas which can be sorbet), In Torr. liter of gas per gran

Nilrogen -t95*C

Fig, 5,46. Adsorption laothema, *olecular a lava Linda type 5 A. of aorbant, as a fraction of the residual g*» pressure. It Is apparent that as the sieve temperature in lowered, pore gas Molecules can be eorbed. Neon and hellua, however, are sorbed to a lesser degree than nitrogen* In the atmosphere, the partial pressures of nitrogen, neon and helltsi ara 595} 1.4 x 10~2 and 4 x 10-3 Torr, reapectively. If the alave la exposed to a volune of atmospheric air, the nitrogen par­ tial pressure will be reduced to a *uch greetsr extent than that of neon mod hellua.

The isothara* (Fig. 5.46) indicate that a partial pressure of -2 -3 10 Torr le obtained after pumping 100 Torr-11tare of nitrogen, 10 Tore* liters of .aeon or 6 x 10 Torr. liters of helluat .«r great of zaollt*.

Figure 5.47 ahowa tha equilibrium oresaure after atncwphertc air tint been punpad. Commercial sorption pu*pa normally operate In Che raufX 10" - lo" liter (chamber volume) per Rriun (of sorbent}, For smalt values of this range the Uniting pressor** in governed by the quantity

Pressure oi M3 rn ftfr

id" 16* ttf1 \ -Chambtr Volume/Sorbtnt Weight, liiers/grofr

Fig. 5.47, Predicted equilibrium praaauraa of the constituents of air «• a function of puaplng load. of nitrogen serbtd at row temperature {branch A). It can be Icwcrcd by preheating UJ» aeolite to 300*C and cooling frost this temperature (branch B). Oxygen, argon and carbon dioxide are pimped quite effectively and do not present the problems encountered with neon and helium.

5.52. The Sorption p«p. The sorption pimp conslRts of a stainless steel body with Internal copper fins to facilitate heat transfer to the zeolite charge* \ liquid

-3*S-~1

Fig. 5,48,a,Typical commercial sorption puaip (Varlan); fe*fwo sorption pumps connected for Multistage pumping. nitrogen container can be Attached to three support backets (Fig.5.48 a). The sorption pump is valved into the system, and lmamed In - 303 -

liquid nitrogen. As the temperature of the zeolite (molecular sieve) fall*, It aorbfl (Fig. 5.47) more gas from the system to csuee a reduc­ tion in the pressure. After pumpdovn to the equilibrium pressure has been accomplished, the valve to the system is closed. At this stage the molecular sieve Is saturated. The re-aetlvatloc can be carried out by allowing the pump to vara to roon temperature, care being taken to vent the pump. A removable, rubber stopper (Fig. S.48) allows easy release of the gas which 1B evolved from che molecular sieve as lc warms. The stopper also acts as a safety pressure-release valve, since a high pressure may develop in the pump on its heating.

Normally, warming the molecular sieve to room temperature Is all that is necessary to prepare It for the next pump-down. However, since adsorbed water is not readily evolved from the molecular sieve at rots temperature, it occasionally becomes necessary to hak^ the molecular sieve for several hours to drive off water vapour. For this purpose a bakeout unit i& used, wh^ch fits tightly around the sorption T3u=rp, and heats it up to 300°C.

Sorption pumps are generally used to pump fromt atmospheric pres­ sure and ultipate pressures of the order of 10 Torr are achieved provided tV • eorptive capacity is correctly matched to the volume of the systen). Typical performance characteristics of the pump from Fig. 5.48 a, are shown in Fig. 5.49. In order to pump larger volumes several pumps may be used simultaneously. In order to obtain a sort of continuous pumping it is posaiblt Co use two punps alternately, one pumping whilst the other is valved off froo the system and Is being re-activated (Fig. 5.48.b).

5.53 Multistage sorption pumping.*

The final pressure achieved by sorption pumping can be inproved by prepumping the sorption pump with another sorption puap or a aecha-

* F.T. Turner and M. Felnleib, Tram. Second Interest. Views Congress, Pergaaon Press. Oxford, 1962 p. 3G0 - 306 * A.E. Harrington! High Vacuum Engineering, Prtntice-Hall, Englevood

Cliffs, N,*p p. 132 - 136. - J04-

;\

f 100

i zo 01il«r l*«W 20 £ 10 E E

1™

PwBpm? imw [Ma] *[ftr lOmin pf»-e***

Fig. 5,49. Pressure vs. piMp-down tiae at constant volune. Cor sorption pump (Fig. 5,48 B).

aicml puep. The effects which can he achieved amy he shown by considering the exaaple of a sorption pump eonnacttd to a vacuum systen, where the ratio of the chsaber vclune V to the aorbent weight w la V/w - 0.02 liter/g. If the partial preaeure of a particular gee In the ataosphsre of the systea 1* P,» than the quantity of thia gsa present. In tha ayataa la p. . v/w Torr, llter/g. An the punp la cooled to

-195*C, a louar equilibrium preaattre ?2 2a achieved, at nee a quantity

of gaa (Fj - P2> V/w waa eorbed. Thin quantity of gaa la equal with that shown by the -195*C adeorptlon isotherm (rig, 5.46). If we denote

this later quantity by Q_/w# the equtlibriw is described by the aqua- Fig. 5.SO. Multistage norptlon pumping.

(S.18) which can be arranged an

(S.l?)

Flgur. 5.50 ahowa the plot of t^L. for „/v . „ „ (llne u and

V/w . 0.01 (line 5). Curves 2 and 3 are plota of qT/w + 0.02 P2 for nitrogen and neon uilng the values of Q_/w (ac-195*c) shown In rig. 5.46. Line 4 ia the value of Q-Vv n1tror.cn at ram temperature. Point A (Mg. 5.50) results £roa the intereectlon of ^ - 595 Tone (partial prceaure of N_>, with Una I, thus defines the value y *

? «•£_ . 595 x 0.02. The horizontal through A, intersecta curve 2 1 * -3

at B, and the vertical through & definea the pressure P£ - 3 x 10 Torr, and Intersecte line 1 at C. According Co «q. 5.IB, the vertical distance BC is the quantity of nitrogen aorbad. By adding an additional sorption pimp V/w Is practically halved to 0.01 (Una 5,Fig. 5.50), and for both pimps cooled siMultaneougly point A aoves to E. Since 0.02 F or 0.01 P are snail values compared

to Qj/w for H2 at - 195*C, the curve 0.01 P +• Qj/w la practically the sane with curve 2. the horizontal through E, intersects curve 2 at F, thus the new equillbriua pressure is about P, - 1.3 x 10~ Torr, which cannot be considarad MM a significant improvement.

By cooling the first pwp and leaving the second pomp connected but uncooled, the pressure drops fro* A to B as before, and the quantity of M„ remaining at C i* about five orders of magnitude smeller than ths initial quantity at A. If the first pump is now valvad off and the second PUMP IS cooled, the pressure drops fro* C to D» thus

the final pressure of »2 is of the order of 10 - 10 Torr, which I* a serious improvement. Since the two pumps have already adsorbed a certain quantity of nitrogen at row temperature which Is given by point 0 (line *> this must ba added to that represented by point

A (Pig. 5.50). thus the quantity of *, present is increased to 1 1 A , which raises the intermediate pressure from B to S , moves C

If neon t* considered, point W (rig. 5,50) represents its ~2 partial pressure (1.4 x 10 Torr) in the atmosphere, and the quantity of neon present. The first puatp lovers the neon pressure to Zx 10* Torr (point X), and the second pump to 2.6 x 10 Torr (point Z), which is a pressure Much higher than that of nitrogen (point 0, or

D ). In practice, adsorption of «2 can Interfere with neon adsorption and the actual equilibrium pressure of neon may he fifp,tier th.in point 2. *f helium is considered, the SIM*13 fsorrvtivc capacity of seoltte for this gas (F1.&. 5,46) explains the fact tliat Its part iaI pressure of 4 x 10 Torir is practically not changed l.y sorption purapinR.

Tlius sequential pimping Is able to reduce the partial pressure of N_ (and active gases), but cannot achieve total ultimate pressures lower than a fev microns due to neon and helium,.

Fig. 5,51 Two stage pumping, with two different cotnbiaatio»s of pu»p.«*.

IE s Mechanical punp 1B used to prepump eha system, although - 308 -

the result It not entirely oil free, it reduces both the partial pres­ sure* of neon and helium. By cooling the sorption pump after this prepumplng, the ultimate preasure In the system should be much lower (Fig. 5.51).

5*6 Cryopumning.

5*61* Cryopuaplng mechanism.

Cryopumplng is the process by which gases (vapours) are condensed at low temperatures In order to reduce the pressure. The ultimate pressure which can be achieved by such a process was described by eq. 4.9, while the pumping speed which can be obtained was expressed by eq. 4.12.

la order to understand the cryoputnping phenomena Moore* analyzed it In some detail. He considered the molecular flow (see Sec. 3.11) of a gas between two Infinite paralled planes; one a gas source and the other a cryopump condenser (gas sink), as shown in Fig. 5.52. For this analysis Moore made the following assumptions: 1. The distance L between the surfaces is small compared to Che mean free path (molecular *Iow),

2. The condensing surface is covered with a deposit of condensed solid formed from gas from the source surface, and the deposit has an exposed surface temperature I,.

3. Of the stream of molecules that strike the solid on the con­ densing surface ( the mass flow rate w.) the fraction JE^stick and the rest are dlfusely reflected. 4. The reflected molecules leaving the condenser constitute a mass flow (1 - f)w , and have a velocity distribution corresponding to the temperature T«, i.e. the accommodation coefficient (eq, 2.104) is unity.

* R.W. Moore, in. Trans. 2nd Internet. Vacuum Congress, Pergamon Press, Oxford, 1962 p. 426-438. - 309 -

Fig. 5.52. Model for analysis of cryopumping between Infinite parallel planes,

5, In addition to the reflected molecules, the solid deposit also emits molecules by evaporation at the temperature T-, and at the sane rate We» as If It were In equilibrium with a gas at tempera­

ture T2.

6, The mass flow u, from the source consists of the flow H. c- imiited by the source and that of diffusely reflected molecules w. which strike the source surface. The mass flow w- In constituted of molecules having a velocity distribution corresponding to the tempe­ rature T.«

7, The velocity distributions of all molecular streams are Maxwelllan (eq. 2.38).

According to these assumptions, the gas between source and condenser (Fig. 5.52) can be considered composed of two streams moving In opposite directions; w. flowing from source to condenser, and w, flowing from condenser to sourcet w, and w_ have velocity distribution!: corresponding to temperature* T. and T£ respectively.

Sine* the net mass flow Input to the system Is given by:

Wl - wl " w2 - wl f " We2 (5.20) the masa flow for the two opposite streams results as

(5.21)

We s + -— (5.22) 2*\[r-j

densities n. and n?, according to eqs. 2.47 and 2„38 It results:

"2 " °2 v2»v f° k T2 ^

A 5 -2[-H ] (5-24) where w/A is the mass flow per unit area, m la the molecular mass,

and v&v la the average molecular velocity (eq. 2.38). Equations 5.21 - 5.24 can be used to express the densities of the hypothetical gas streamst

. V- + We? t 7t[ h W + We.

+ v2av ( ISTTT) —*r (5.26) The total molecular density 1B the average of these two expres­

sionst but the gas between uourco and condenser will be far different from Isotropic, thus thci usual meaning of pressure cannat he uaed here. The situation can be illustrated by considering the pressure which

(T) 0 CASE A

rp = njR i, / —- 'I

CASE Q Wl 0 s.

CASE C

Fig, 5.53, Pressures Inside an open-ended prohe. would be sensed by on open ended probe in various orientations (Fig. 5.53). It is assumed that the pressure sensed is the pressure ? inside the probe, the gas being At the temperature of the probe T. The presure P reaches a level so that the efflux of molecules through the opening equals the Influx from the environment, which is the sum of the molecular fluxes incident on the probe opening from surfaces 1 and 2 (Fig. 5.53).

Based on the expressions marked on Fig. 5.53 and eqs. 5.25; 5.26 the pressure sensed in various orientations is:

Case A : 2»k T i>j H, + Be, ,-F^A (5.27)

15.28)

fork I 1% Kx (l-£> + Ifej "I • ' J A~^ (5.29)

Equations 5.27-5.29 show that regardless o£ probe orientation the contribution o£ the nolecules evaporating from the deposit is the same and equal to

T '-NH- 1 * "e, (5.30)

According to assunption 5:

1 I* (5.31)

where P, Is the vapour pressure of the condensed gat at T,t and from 5.30 and 5.31:

<5.32) thus the contribution due to reevaporatlon fron the condensing surface is equal to the vapour pressure of the deposit corrected for the probe temperature. The contribution of the net mass flow input U. to the systen

Is dependent on both f and the probe orientation (eq. 5.27-5.29). This dependency is shown in Table 5.4 for extreme values of f.

Table 5.4. Probe pressure P , for We, = 0

2TI k T £ (2TT lc T 1* W. 1

Obviously the values i«\ represent the cases without appreciable cryopumping, thus the pressure is independent of the orientation of the probe, i.e. the gas is isotropic.

When f»l the portion of the probe pressure contributed by H. is highly sensitive to the orientation of the probe, varying from a maxi­ mum value for case A to zero for case C. Thus when W is large compa­ red to Ue_ the gas is far from being isotropic.

The pumping gpetd of an isotropic system is defined by S - Q/P. Since the expression of the pressure in cryopumping (Fig. 5.53) is different according to the orientation, the pumping speed is also - different.

In the case of space simulation a pumping speed per unit area S ./A based on the molecular density n . sensed by the source of gas - 314 -

is anst significant. This is equivalent to case C (Fig. 5.3S) with I - t,. Ihua frosi eqs. 5.29, 5.31 and S.32; pi

2»k T. Y «, U-f) (5.33) 'si

si k I, (5.34) the puaping speed per unit area is

As Pgl approaches the value P _ Cc^/I^lh the punping speed diminishes to zero.

It results that as long as Pv2 «^ml the pumping speea apparent to the gas sours* is independent of the pressure and eeaperature of the condenser and has the value

"art A 1-f (5.36)

When £«1 (reflecting)

(5.37) while when f - 1 (good sticking):

This later Is the case which is desired to be achieved in space environment simulation.

A second definition of the pumping speed is that based on case B (Fig. 5.53), which corresponds more directly to that used for diffusion pumps. In this case the pumping speed S is given by

S W. -E_: L- A Am n #- •^&JW- •u- :-*-fcr

To get maximum pumping speed it is required that P - «P

(5.40) A 2-f

Here the^pumping speed Is Independent of pressure, and the significant temperature is T , the probe temperature. This temperature la prac­ tically equal to the ambient temperature If conventional vacuus systems are concerned, but is usually very different in syBtenia using cryogenic surfaces*

From eq. 5.40 and for f«lt the pumping speed per unit area Is S V _E_£*f -*V£_ (5.41) A 4

and for f <

(5,42)

The result In eq, 5B41 Is identical with that shown by eq. 5.37, because probe orientation Is not important when f«l. The result in

eqa 5,42 differs markedly from that given by eq. 5.38, because the probe inlet In position B will still receive molecular flux, even when f • 1.

5.62 CryopuapJnR arrays,

Cryopuaping surfaces cannot generally be exposed directly to a source of gas at room temperature because the heat load due to radia­ tion would exceed that due to the condensation of gas molecules. There­ fore, the cryogenic surface is protected on the side facing the gas source by an optically opaque baffle (eo,. Fig. 3.14, 3.15) at an Intermediate temperature to act as a radiation shield. Figure 5.54 shows some common arrangements used In space simulation.

The radiation shields also impede the gas flow to the condenser and limit the maximum achievable pumping speed. Fig. 5.55 shows a cryopuaping stray which offers a reasonable compromise between pumping •p«*d and radiation losses. Here both the chevron baffle and the back shield are cooled with liquid nitrogen to 77 - 100'K, while the conden­ ser is operated at 20°K, so that nitrogen and all less volatile gases are cryopumped. Additional means, such as diffusion or ion-pumps must be provided for removing helium, hydrogen and neon. The emnissivlty of each surface Is chosen primarily to minimize the heat load on the con- - 317 -

(cl

Fig. 5.54, Ctyopuroplng arrays where 100'K surfaces are interposed between 20I>K surfaces and the room temperature surfaces (veliicle, clu-rmliflr). denner, and secondarily tn minimize that on the I'adlation sUi«=j '..i. The relatively high value of the condenser emisslvity taV.es into account the presence of frost.

Tlie pumping effectiveness of the array will he described by ltd over-all capture probability G, rather than by the sticking coefficient of a simple condensing surface. The capture probability G la the - 318 -

llMglltM 090 050 020 300*11 CL50 ,.,....*„... „.»„,„•

Wit HHT Jim

Fig. 5.55. Portion of a cryopumping array (Hoore). fraction of the total number of molecules Incident on the inlet side of the cryopumping array which is finally captured In the array.

Moore analysea the model array ahown in Fig. 5.56, by using : g - the probability that a molecule Impinging on chevron shields will pass through (see Sec. 3.5); f - the condenser sticking coefficient, and % - the probability that a molecule impinging on the condenser will pass through (equal to the ratio of open area).

Acording to the notations on Fig. 5.56, the conditions of equili­ brium are:

g w. + (1-g.) v. 's U. 1 1 ¥Vi w, « I

1

"w? .'v'1: ''A 1 1 *7 ;^' *r 1 """ S_; 1 K' I CltVHON CONDfNF_R RADIATION SHIELD

Fif,. 5,50. Model for analysis nF cryrwnir nrrnv.

(5.'.3)

ami tl»e simultaneous solution of thflsr I'qnntinns, p.ivc the over-all capture probability as:

' " "l " . i1" U"fc) (1_f' C2~','s) + d-r.^fO-r.^'d-n^-c'l (5.W

Annunlnf; p - 0.23 (FIR. 3.22), and R •» 0.25, clip capture probability clvi'ii by ec. 5,44, is n function of f, as shown by I'IR. 5.57. Tlic pumping npecd of such an array, as defined by a probe oriented as case U CKIR. 5.53), is given by eq. 5.40 witli f replaced by G, tbus

S 2 G V _E_. avp (5.15) 2-G i - 320 -

oA i — i 1 1 1— O OZ 04 06 06 IO

f - CONDENSER STICKING COEFFICIENT

Fig, 5.57, Effect of condenser sticking coefficient on array capture probability.

5.63, CryotrappinE*

Oases may be trapped on coaled surfaces on which a condensible vapour (eq. water) lias been condensed, with the result that tlie partial pressure attainable may be significantly lower than the equilibrium vapour pressure at the temperature of the cooled surface. This pumping action in known as cryotrapplnp,. It offers the possibility of cryo- ptunplng of gases such as N„, H-, Ar mucli more effectively in the pres<- ence of a contaminating agent (such as water vapour) than in a system from which all such agents have been carefully removed and excluded. Cryotrapplng Is believed to be due to the non-condensable gas being carried down by a condensable vapour and trapped within the pore struc­ ture of the condensate.

The cryotrapping of nitrogen and argon by water vapour condensed

at 77*Kt has shown* that the water vapour forms a porous deposit with 2 an effective area of about 600 m /g of water. The quantity of N_ and Ar required to saturate the surface deposit of water, is proportional to the quantity of water deposited (Fig. 5,58). The number of mole-

* F.W, Schmldlln, et. al. in 1962 Vacuum Syrup, Trans, Macraillan, Neq York, 1962 p. 197. - 321 -

M,qiams of HjO

Fig. 5.58, Quantity of nitrogen and ar^on required to saturate .in ice coattnRt culea of nitrogen trapped per molecule of water condensed on tlie sur- _2 face line a constant value of about 10 for partial pressures of N, above about 0,1 Torr, and then decrear.es with decreasing partial pres­ sure to a value of about r) x 10 at 10 Torr.

The cryofcrappinft of hydrogen and helium Uy condeased arrcon at 42°K was also achieved**. The result a are shown if Fip,. 5.5*1, For the curves (a) and (b) tbe hydrogen flow rates are so low tli.it the equilibrium pressure is below that for saturation at 4,2*K, ao that no condensation occurs nnd the pump trip action is only that of the diffusion pumps. In each of these cases, when the argon partial pres­ sure reaches about 1/10 of that of the hydrogen, cue hydrogen partial pressure drops suddenly by a Factor of 10or more and then remains &t this lower value as the arpon-ilow rate and partial pressure are increased. The decrease in hydrogen partial pressure is due to oddt-

**J. Ilengevoss and E.A. Trendelenburg, in 1963

Vacuum SympM Trans> Macmlllfln, N.Y., 1963 p. 101, - 322 -

|a»«|il| KtMtlttM/ca1 tie

AitMptiKltriuwi.rMr

Fig. 5.5v, Hydrogen cryotrapping by argon for different valued of hydrogen and argon flow rates. tlonal pumping resulting from the trapping of hydrogen by the condensed argon* From the trapping rates and the nrea of the cryosurface, the sticking coefficient for hydrogen on the argon deposit was determined to be 0,4, The 45" line on Fig. 5.59 corresponds to the case in which one hydrogen molecule is trapped by one condensed argon molecule. Curve (c) is taken at a hydrogen flow rate which 1B great enough that the hydrogen partial pressure exceeds the saturated value at 4.2*K so that the condensation on the cryostatic surface occurs even in the absence of argon. Thus in curve (c) both condensation and trapping occur at the same time. Therefore the break in the curve (cryotrapping by argon) occurs at an appreciable lower argon flow rate than that corresponding to the intercept with the 45" line. Tit 1B indicates that about 10 times as many hydrogen molecules are deposited by the combi­ nation of condensation and trapping as are argon molecules.

Experiments for cryotrapplng of helium showed that the sticking probability of helium on argon deposit is about 0.03 and that about 30 argon molecules are required to trap one molecule of helium. 5.64, Cryopumpa,

A practical cryogenic pump is illustrated in Fig. 5.60. It consists of a helix made of stainless steel cube, which acts as the condenser surface, mounted directly in the chamber to he evacuated. The coolant (liquid nltrop.on, hydrogen or helium) is supplied from a Dewar to the helix through a vacuum insulated feed tube, and is

;-.-T© roughing pumps

Temperature sensing element \~f**C'

JhrottU valv*-

Fig, 5.60. A cryopump (schematic). made to flow through the coil by means of a REIS pump at the outlet end of the coll. The coolant boils as it passes through the coil, hence cooling the tube. The rate at which coolant passes through t>ie system , and hence the temperature to which the condenser surface is cooled, Is controlled by a throttle valve mounted in the gas exliau.it line. A temperature sensinp clement mounted on the condenser coils automatically controls the throttle valve setting. _2 The cryogenic pump is not used at pressures above 10 Torr, partly because of the large quantltieH of coolant that-would he requi- - 324 -

red, and partly because the thickness of solid built up during high pressure pulping would seriously reduce the pump efficiency at low pressures. The rate of built-up of solids is typically of the order

of 10 cm/h at 10 Torr and 10~2 cm/h at 10 Torr.

Pimps having speeds for nitrogen up to 5000 1/s (liquid helium consumption * 2 1/h) are commercially available. Pumps with speeds

of the order 10 1/s are feasible, but for these high speeds the coolant would be fed directly from a gas llquefier rather than from a storage vessel.

5.65. Liquid nitrogen traps.

Liquid nitrogen traps are cryogenic devices which function primarily to prohibit the transfer of pump oil vapour Into the vacuum system, and to pump by condensation, water vapours and other vapours which originate In the system,,

Hell designed traps incorporate the following featurest

a„ The trap offers minimum impedance to the diffusion pump (see Sec 3,44 and 3,5) and to the vapour condensation or cryopumping sur­ face (Sec. 5,62)„ However the trap should be effective in keeping pump oil from the chamber. b. A constant low temperature of the cryopumping surfaces is maintained to prevent pressure bursts resulting from the reevaporation of condensate. c. There 1B no warn surface path shunting the cold areas which would permit the diffusion punp oil to migrate. d. Additional considerations include minimum liquid nitrogen consumption, a trap interior accessible for cleaning* A water-cooled baffle is required to prevent holdup of the pump oil on the trap during long-term continuous operation. The condensed oil should return to the pump along the trap walls rather than dripping onto the hot jet assembly„ W.B.

Oil Molecule Polh

Fig. 5.61. Examples of liquid nltrop.cn traps. ],N_ - liquid nltropcn; W,l), - water baffle; T.fi. - thcrtnnl gradient.

Figure 5.61 illustrates several trap designs which meet some of the above criteria. Typos a,b,c are generally used with pumps jf small and medium sizes. An elbow trap (Fip, 5*61.d) may be used on large vacuum systems where n liquid nitrogen circulation system is available. A vacuum system pumped down to Its operating pressure exhibits molecular flow through the trap, thus for an optically tight trap, en oil molecule must undergo at least one cold vail col­ lision (Sec. 3.5) before nntnrlng the vacuum chanher. Since the sticking coefficient is practically less than unity, and some oil-to- oil and oll-to-g*n molecular collision also occur, traps can be more effective If they »re designed such thot each oil molecule impinges on the cold *urface» .1 j*reater number of tiroen (Type a,b) tlwn the •iuiMM of one contact required by simple optical ti&htneHB

I-' 1 ' 1 ' r "I " ... , V..9.C on Cotfficttnt

.59 " .999 - S*m „,—011- HI H) pteg '•---/ V - / V

•/» -Got cfni i'.. \N - : \ - 1 \ - - 1- \ 1 - D *f -'« - Gal 3w«epln< - •» - Mote., Br —* •—Troni Hon—* —ViK mi .

1 .1, 1 1 1 . 1.

loo P, Ce» Prtssurs* (un­ rig, 5.62, Dil-backBtreaminf, (calculated values) through a 36" dlam. optically tight elhow tr.lp,

D.W, Jones and C.A. Taonls, J, Vac, Scl, Techn. 1, 19, 1964 - 327 -

at 300°K at Che trap bottom, neglecting any cracking effects. Solid horizontal llneB show the hackstrearning due to a sticking coefficient a less than unity, each line corresponding to a value of a. resulting from the conclusion of a calculation which indicates that 1.09 (l-o) percent of the entering molecules pass through the trap. Unfortunately the sticking coefficient Is not accurately known, and small differences In Its value can cause differences of decades in the backstreaming rate.

The interrupted line (Fig. 5,62) Is the oil transfer due to oil-oil collisions,, From the calculation Jones and Tsonis established that the oil transfer due to oil-oil collisions is

,, 3 N « 2 x ID* 2^- molec/second (5.46) ir where a (era) is the radius of the pipe of the elbow, M is the molecular weight of the oil„ and p is the partial pressure of oil vapour at trap inlett (Torr). It results that this mode of transfer is significant

only if a>0o9999o

The probability P of an oil molecule escaping the trap due to an oil-gas collision, was found to he

5 L Pr - 4.5 x 10" (1-e ~* ) (5.47) where % is the mean length of escape path from trap base to knee (in

the case considered % - 2,7a)( and L is the mean free path of the oil molecule. Since L Is an inverse function of the gas pressure, the transfer due to oil-gas collisions is rising with gas pressure. It results that this mode of transfer is significant only if ct>0,99.

The right-hand side of Fig* 5.62 shows the oil transfer resulting from diffusion and sweeping action in the transition and viscous flow regions.

Traps are ueuelly not meant to operate in the transition region - 328 -

(see Table 3.1). although condition! existing in Che transition re­ gion can be expected during pu&pdown. The trap configuration shown in Fig. 5.61.b. suppresses thiB oil transfer by having two regions

of trapping which differ significantly in dimensionst the skirt and til* chevron, Thus, transition flow in the two sections will occur at two different pressure ranges, and at any pressure at least one of the sections will be efficients

Regions of the trap where the cold surfaces are in contact with ambient temperature parts, experience temperature gradient shifts resulting from changes In ambient temperature, or in the cooling con­ ditions. These gradient shifts result in the reevaporation of a part of the condensate. If the reevaporated gas is not retrapped prior to chamber reentry, a pressure rise will occur in the chamber. The type of trap ahwon in Fig. 5.61.C has no allowance for retrapping capabil­ ity. The other designs shown In Fig, 5,61 have a good possibility of retrapping such molecules. The amount of gaa condensing on the thermal gradient region can also be minimized by shielding these

regions from the chamber with cold surfaces (Fig. 3,61 a, b, d)0 Gas condensed on the intermediate temperature regions during the higher pressure phase of a pumpdown will reevaporate at lower pressures and thus cause an increase in pupdown time. Figures 5,61 a and c indicate that troublesome temperature gradients may exist on the top of a partially empty liquid nitrogen reservoir «hen it is in sight of the chamber. If the reservoir walls have a sufficiently high thermal conductivity, this temperature gradient can be avoided.

5,7 Gettering}.

5.71, Gettcring principles.

Any sealed-off vacuum device (lamp, electronic tube) In which the pressure remains essentially constant, contains a chemically active getter. - 329 -

While effort is made to effectively degas the tube parts during construction and evacuation before sealing-off, there will always be some evolution of gases during operation. To avoid a build-up of pres­

sure a getter i»ee a material able to chemisorb gases, is Included In

the tubet

Getters may be classified into three groups according to the form in which the getter material is active: flash, bulk and coating getters,

Flash getters are chemically active metals which can oe easily volatilized. Flash getters may work by the process of dispersal Kettering iii which gas is sorbed whilst the getter Is being evaporated, and contact getterine in which the film of getter material deposited by evaporation on the walls continues to sorb gases„ Dispersal getterlng is limited in time, but the getter presents its maximum possible sur­ face to the gas; contact gettering is a continuous process where sorp­ tion on the outer layer is followed often by diffusion into the bulk.

Bulk getters are heated metals in sheet or wire form.

Coating getters consist usually of powders sintered on the

surfaces of electrodesD

The general conditions for getters can be summarized as:

a. The getter must he resistant to storage prior to use. b. It must not be affected by the process of manufacture of the tube (seallng-ln of the electrode assembly, baks-out of the tube, see- ling-off of the tube). c. The getter should absorb during its evaporation (if flash getter) and after the tube has been processed, d. The vapour pressure of the getter material and its reaction products with the residual gases in the tube should be negligible.

Operating temperatures and uses of various getters are summarized In Table 5.5» - 330 -

Tibia 5.5

II

|#l Vii

m

2fKM*

|jl!e - 331 -

5,72 Flash getters.

The active Ingredients In flash getters are chemically active metals that can be easily volatilized, such as Ba and Ba-Al alloys. An alloy of rare earth metals (Cerium, Lantanum) known as mlsch metal Is also used. Phosphorus is the getter used in most Incandescent Limps,

Barium getters are made in a form that can he easily evaporated and deposited as a thin film on the vacuum envelope or on other cool inactive parts of the structure. In the processing of an electron tube, the procedure is: evacuation arid bakeout at 400-500"C, cathode heating, induction heating of electrodes, and flashing the getter just before the tube is sealed-off or immediately afterwards. As barium is very active it must be protected against atmospheric action during storage and within the vacuum tube prior to flashing. The pro­ tection Is done by incorporating the Ba in metal pellets or tubes, or alloying it.

Nickel, copper or iron clad tubes or pellets are obtained by filling a tube of these metals with liquid barium under vacuum. The tube is then drawn down to size obtaining a wire, or cut into small cushion-shaped pellets, Ni, Cu and Fe tubes with Ba filling are commercially available in a wide range of diameters and in "infinite" length under such names as Hiba, Cuba, and Feba wire. The right amount of Ba for a particular use is nipped fron the clad wire with blunt-edged cutters, which causes self-sealing of the ends and protects the Ba from atmospheric oxygen to a one extent. The tube (cladding) wall Is often purposely thinned along an axial strip to facilitate exit of Ba vapour and to direct it in a predetermined direction. The surface is ground flat (Fig, 5,63a) on one side in the getter known as KIC (Kenet Iron-clad), or given a longitudinal notch, known as Kerb getter wire (Fig. 5,63,b). These getters are to ba degassed at about 750"C, and subsequent heating to 850°C or more causes the tube or pellet to burst. - 332 -

Fig. 5.63. Barium getter wire, a) KIC; b) Kerb.

To avoid getter spoilage during storage or processing, alloys of Da with Al are used. These are known as the Stahil type getterfl of SAES-Mllano, and consist of Ba-Al alloy either in a straight, grooved Fe or Ni tube which Is welded to a -nickel or iron alloy wire bridge (Fig. 5.64 a,b) or within a groove in a ring of stainless atcll (Fig. 5.64 c). These getters are chemically stable in air up to temperatures of 600*C, but require a higher flashing temperature (above 1000°C> than the pure Ba getters,

A gettering method used in all-metal electron tubes is the Hatalum process. In this process a thin tantalum strip which can be heated by special connecting leads, in coated with a mixture of Fig. 5.64. Ba-Al alloy p.ettere.

Ba CO- and SrCO- which Is fully stable In nlr. This strip Is heated first to 800-110QaC (it a suitable stage of the evacuation process, whereupon the carbonates decomposes to their respective oxides. CO. is formed wheih is pumped off, the atrip temperature raised to 1300°C, the Ttt substrate reduces the oxides to Ba which evaporates onto the tube wall. A disadvantage Is tltat the tantalum txlde formed, slowly dissociates and increases the pressure in the sealed-off tube. The replacefttentof the carbonate mixture by barium berylllate (BaBeO.) la a subsequent development which avoids the evolution of CO,, Two other types of these (reaction getters) are the Alba in which barium oxide la reduced in presence of aluminium, and Bnto in which barium is eva- - 334 -

porated from a Mixture of Ba-Al alloy, icon oxide and thoriun powder.

Red phosphorus is used as a getter in both vacuum and gas filled lamps. The tungsten filament is coated with red phosphorus from an alcoholic suspension. At the first incandescence of the filament, the phosphor is flashed and it gives rise to a deposit on the glass enve­ lope which is transparent to light.

Magnesium is used as a getter in mercury vapour hot cathode discharge tubes, and Al-Mg alloy (the so called Fornier getter) is used lo oxide-cathode tubes.

5.73. Bulk and coating getters.

Bulk getters are usually operated at elevated temperatures to promote diffusion of the adsorbed gas into the solid. Since evapora­ tion is not essential to the operation of bulk getters, they are usually selected from chemically active elements with low vapour pres­ sures and high melting points. These include Ti, Zr, Ta, Th, U, W, and Mo, used in a variety of forms! sheet, ribbon, wire, rod (bulk getters) or powder (coating getters).

Titanium is frequently used as a bulk getter in electron tubes. After Ti has been outgassed in vacuum at 800*C for a few minutes no additional gases are liberated when it is heated to higher temperatures.

At temperatures above 700*C Ti continuously sorbs N,, 0? and C0_, since at these temperatures C, ti and 0- readily diffuse into the TI, leaving an active surface. Sorption of these gases from 10 to 90 atomic percent is possible, and the compounds formed do not dissociate at higher tem­ peratures or upon subsequent reheatlngs. Hydrogen, is sorbed by ti from room temperature up to 400*C. Above 500*C, H. is released by

the Ti and at 800'C it becomes practically free of H2> Water vapour and CH. nay be gettered by a two-step process: the TI is heated first

at 1000-1200*C where H20 and methane are dissociated, the 02 and C diffuse into the Ti, leaving the H„ In the gaseous phase. The Ti tern- perature is then lowered below *tOO°c, and the H is gettered.

again at 880°C or more- Water vapour is cleaned up between 200-250°C;

°2* N2* °° and C02 are eorbcd at ^OO'C, An efficient getter can therefore be arranged by means of two Zr wires or strips! one at AQD°C and the other at 14GQ°C. Another usual procedure is to spray or paint the zirconium powder in a binder (nitrocelulose in anyl ace­

tate) onto the electrodes0 Zirconium hydride (ZrH,) may also be applied

as a paste, and at 800fiC in vacuum t'ie H is released leaving the Zr coating»

Tantalum behaves like Ti at elevated temperaLures. After flashing

D at 2000°C> it getters H2 at 800 C and 0,, and N2 at 1500"C0

Thorium sorbs 0 t and H in the temperature range of 400-500°C. It is also used in a getter called Ceto which is an alloy of 80% Th with 20% misch metal Can alloy of 50-60X cerium, 25-302 lanthanum and other

rare earth metals). Thorium powderB and Ceto powder are also used as coating getters=

Tungsten filaments getter 0- at temperatures above 1500*C. A fraction of the 0. molecules striking the hot filament are converted to W0,. and then immediately evaporated off leaving a clean surface for further gettering action. Nitrogen and CCJ cannot be gettered in this way, but they can he gettered by evaporating W onto a cool sur­ face* When W is heated above 2300°C, a fraction of the ft. molecules striking the filament will dissociate Into atoms„ The atomic hydrogen produced is very active chemically and may reduce glass or ceramic In the vicinity of the filament to form metal oxides and water vapour. On striking a cool metal wall the atomic hydrogen may recombine to H-„ Water vapour ment above 2300*0,, the surrounding walls as WO. (white deposit). Atomic hydrogen evolved 336

deposit), with the formation of water vapour. The cycle can then be repeated indefinitely until the filament is etched to the point where it burns out.

An interesting inverse cycle occurs when a B filament is heated above 2800*C in the presence of iodine vapour. Tungsten which evapo­ rates from the filament and condenses on a hot surrounding wall at about SQO'c will combine with the iodise to form 81, vapour. This vapour is then dissociated on coining in contact with the hot filament and iodine vapour is released to react with the W deposit on the wall again. The phenomena is used to increase the life time of incandescent lamps.

5.74. Bettering capacity.

The sorption capacity of getters ran he evaluated in terms o£ pumping speed per unit area of getter deposit, and in terms of total sorption per unit weight of getter. For the same getter both figures differ for various gases, and depend in some limits on the history of the getter (degassing, flashing, etc).

It is obvious that the pumping speed S of a getter film of area A, can be calculated by a formula of the type of Eg.. 3.72 or Eg.

4.12 (for Pu - 0)„ thus

(5.48) M where • is the sticking coefficient, T and M the absolute temperature and molecular weight of the gas. By comparing the maximum theoretical pumping speed

for 02, 0,001 for H and 0.003 for N^. Although these values can be changed If a mixture of gases are - 337 -

present, it results that the maximum Initial gettering speed for CO is 4,75 liter/sec. cm , and for 0- is 0.2 lit/sec, cm . All measure­ ments show that this speed will fall off with time.

The total sorption capacity of several getters is shown in Table 5,6.

Table 5.6„ Sorption capacity of several getters„ (Torr.liter/g).

«^^ H H co2 Ge °2 2 2 Barium 15 - 50 45 - 88 3-36 5-60

Magnesium 20 - 200

Misch metal 20 - 50 46 - 64 3-16 2-45

Thorium 7 - 33 19 - 54

5,8. Pumping by dilution.

The partial pressure of residual active gases can be lowered very effectively by alternatively diluting the active gas with an inert one and repeating the pumping process. This procedure is known in the lamp industry as "flushing".

Suppose that the atraosphere(760 Torr) existing in a vacuum de­ vice is constituted of 20 percent (152 Torr) oxygen (active gas) and 80 percent nitrogen (inert gas)- By pumping the device to 0.76 Torr, the residual gas la still 20 percent (0.152 Torr) oxygen and 80 per­ cent nitrogen. If we fill up the device with nitrogen to a total pres­ sure of 76 Torr, and pump it again to 0.76 Torr, the residual gas con- -3 -3 tains 1.52 x 10 Torr oxygen, thus only 1.52 x 10 x 100/0.76 - 0,2 percent oxygen, Repeating the filling up and pumping once more, the partial pressure of oxygen is Lowered to 1.52 x 10 Torr, and so on. The partial pressure Pti vhich can be obtained by flushing ie

(5.49) ° IP,

where P is the partial pressure at the firet evacuation to P_, P is the filling pressure with inert gas, V is the pressure to which the device is repeatedly evacuated, vhile n is the- number of flushing cycles. For the previous «xaraple, after 3 cycles the partial pressure

of 02 will be:

7 ?u - 0.152 pSg&j - 1-52 x 10" Torr

5.9. Measurement of pumping speed

3.91. Methods of measurement.

The pumping apeed can be defined either by the steady state condition (eq, 3.27)

S - Q/P (5.50)

or by the transient pumpdown condition (eq, 3.269)

V Pl S - 2,30 —log •—— (5.51) Z r2

Equation 5.50 is the basis of the constant pressure measuring, roethoda. in which the pumping speed S is determined by measuring the throughput q at a constant pressure P.

Equation 5,51 is used in the constant volume method in which the pumping speed S is determined by measuring the change in pressure (from P^ to Pq) during the pumpdown of a constant volume V.

5.92. Constant pressure methods.

The throughput is defined (eq. 3.24) as the product of the dla- - 339 -

placed volume of gas by les pressure, thus by measuring these rwo values, eq. 5,50 permits to establish the pumping speed.

Two constant pressure methods ofmeasuring the pumping speed are based on this concept: the method of the moving mercury pellet, and the inverted buret.

In the moving mercury pellet test, a large vessel A (Fig, 5.65) is being evacuated by the pump under test and air is admitted through the calibrated capillary tube via the needle valve. An appropriate gauge is connected to the vessel at T, and the needle valve is adjusted until the pressure indicated by the gauge 1B constant. The volume of air v pumped out of the capillary tube in time t is given by the movement of a mercury pellet in the capillary tube. In this measure­ ment the pumping speed can be determined at various pressures P (in the vessel A) and is given by:

2 n P 0V P Tra i

where F is the atmospheric pressure, a is the radius of the capillary and % the distance by which the mercury pellet was displaced during the time t.

It must he mentioned that S is the pumping speed in the vessel A, end includes the conductance (eq. 3„28) of the parts between the pump and the vessel,,

The inverted buret measures both the displaced volume of gas and ltB pressure» The outlet at the top of the buret is connected to

a T leading to the needle valve (Fig. 5a66) on the vessel vhich is

pumped0 The other side of the T connets to atmosphere through a valve so that when this valve is closed the gas evacuated by the pump, sucks oil up into the tube from the beaker at the bottom.

If Vo is the volume of gas at atmospheric pressure initially - 340 -

Fig. 5,65. The movltif, mercury pullet method. existing above the oil lcv-1 .it the in^t.int the valve to atmosphere is closed, then when tin? nil lcvi_*l has risen h (cm], the volume i.s

V = Vo - IiA (5.53) where A is tho CTOHM nt-ct J..ii"il area of the buret.

TnrifrrJI* volvp •f*J rid.

Fig. 5.66. The inverted buret - The pressure F at this moment is given by

* . p _ .. £_2il P u «-M>

the oil and mercury respectively. The quantity of gas present is

PV - P Vo - P hA - £§£i (V h - h2A) (5.55) a a pa . o

If the oil is raised to the height h, during the time t, then the throughput is expressed by:

•-HP„. A + £jii (Vo - hA) (5.56)

2 If Mem), A(cm ), t(sec), and P (Torr), the;: Q results from eq. 5.56

When the diameter of the burett is large

In usual cases hA<

(5.58) in which the correction term may be quite large,so that the use of a simple expression (eq„ 5,57) will lead to a considerable error.

In the Inverted-buret method the valve connecting to atmosphere is opened after each measurement, to allow to the oil level to drop - 342 -

back to that of the beaker. It 1B enscntial to wait long enough before starting another measurement to permit the oil to drain down from the wall of the tube.

lo order Co shorten the waiting tine low viscosity liquids should be used. Water Is for some purpose a better choice than oil. The error due to humidity is in this case about 2.4 percent of the ataospheric pressure. With a 0.2 cm buret tube callibrated In units -3 3 -4 of 10 cat a throughput as small as 10 Torr. lit/sec can be measured with acceptable accuracy.. Room temperature variations can produce errors; a change of 3"C results in 1 percent change in volume. An alternative constant pressure measurement Method consists of Measuring the pressure difference across a known conductance and using eq. 3.26. The conductance can be an aperture (eq. 3.72) or a tube (eq. 3.108), Fig. 5.67 shows the arrangement for Its use.

Liquid • topi*

•oK

To tftrtpufttp To totpvinp

/ 0,»l0.5cr L« 305 cm

fig. 5*67. Arrangement for determining throughput by measuring the pressure drop across a known conductance. The gas entering the test dome of a vacuus pump (Fig, 5.67) flows through a metering tube or aperture of known dimensions from an auxi­ liary chamber (D.). A controlled leak (needle valve) for admitting the test gas and a separate diffusion pump are connected to the auxi­ liary chamber, so that the pressure P. can be adjusted to any desired value. The conductance C of the metering tube-can be calculated from its demensions (eq_„ 3„108), the pressures P. and P. are measured, and the throughput is given by

Q - C

In order to avoid errors the apropriate gauges for P. and P. are to be used, they must be calibrated compared to each other, and the flow must be kept in a range where the same flow regime is In the whole metering tube (see Sec. 3,63),

5.93. Constant volume method.

This method Is based on eq.. 5.51, and consists of recording the pressures P. and P_ at the beginning and end of given time intervals. The method is usually less accurate than the constant pressure ones, because all pump speeds vary in fact with pressure.

The pumping speed determined by eq, 5.51 for the pressure drop — pl + 2 from P. to P„ is in fact related to the average pressure P - ,

Since eq. 5.51 is based on the assumption that S is constant (see Sec 3,73), the time intervals used In this measurement have to be as short as possible.

The constant volume method has the disadvantage of requiring quite large vessels (see eq, 5,51) and of excluding the use of McLeod gauges because of their time-lag. 5*94• HoMurcecnt of the pumping speed of mechanical and diffusion pumps. Typical arrangements for measuring the punning apeed of mechanical pumps are shown In Fig, 5.dfi.

LMtMhc "i- " r -in .o"™ ^-^ lODmn . J UKIID *~* it. —©Vixwm qtWQt

* P,.-nn ' Pump (b)

Fig. 5.68 Arrangements far measuring the pumping speed of oil scaled mechanical pumps, a) for inlet larger thnn 2 in. inside diameter; h) for inlet smaller than 2 in. inside diameter.

Some of the condition*? of the test dome nre: a) it should he of same diameter 1) as the pump inlet; l>) the height of the test dome. and the place of the inlet*; and their direction should he as shown in Fig. 5.68; c) the gas inlet should not he oriented directly towards the gauge Inlet; d) Mc Lend gauges should he preffored; e) the ranges of the gauges should overlap.

It was found that consistent results arc more readily obtained by first pumping down the test system nnd then increasing the throughput from zero upwards and taking pumping speed readings at successively higher values of the pressure. This method is able to check if the leaks In the system arc not excessive and to drop the outgnssing rate 4o a low enough value.

The pressure range of interest for diffusion pumps is generally less than 10" Torr (molecular flow). 7n this range the geometry and dimensions of the connections to the pump inlet affect critically the measured values of the pumping speed. For example, adding a tubular extension (done) of the same diameter as the pump barrel and of length equal to 3 times Its dlatter to the inlet of a diffusion pump will reduce the net pumping speed to about half that measured directly at - 345 -

the. puiap inlet. The configuration of the test rleme, the location and orientation of the gas inlet and gauge connections, all influence critically the measured value of the pumping speed. The American

Fig. 5.69. Arrangement for measuring Che pumping speed of diffusion pumps.

Vacuum Society recommends: n) The teat dome should have a diameter D (Fig. 5.69) equal to time of the pump barrel, and a height of 1,5 D. b) The connection of the gas-inlet tune be on the nxis of the test dome 1.0 B (Fig. 5,69? above the face of the pump and oriented directly toward the top of the done. c? Gauge-connection tube be oriented so that the piano of the open end is parallel with the axis of the test dome and located so that the open end is just above the top surface of the pump and protruding inward from the wall of Che test dome ahout 0.25 In, ta wold plugging by diffusion pump oil condensed on the wall. d) The fiauge should not he connected through a low temperature vapour trap, co avoid the need of temperature corrections.

-347-

6. MEASUREMENT OF LOW PRESSURES*

6.1, Classification and selection of vacuum gauges

As it was stated in Sec, 1,1., the range of vacuum technology extends nowadays about 17 orders of magnitude of pressure below atmos­ pheric. Consequently vacuum measuring techniques have had to be deve­ loped to measure inw pressures of widely differing magnitudes, from a -14 few Torr to aboi't 10 Torr. It is no single gauge which is able to cope with such a range, although tne ideal of vacuum scientists and engineers is to develop such a gauge,

The measuring techniques are made all the more difficult because pressures can only be measured in the range from 760 Torr - 1 Torr by using the force resulting from the pressure to set some form of mecha­ nism in motion„ In the range below about 1 Torr it is necessary to use some other physical properties of the gas (compression, viscosity, thermal conductivity, ionization). Table 6.1. shows a classification of the gauges according to the property of the gas used to measure the pressure.

Each type and kind of gauge is sensitive to variations of pressure

in a specific range (Fig8 6.1), Most of the gauges measure total pres­ sure (Table 6,1), but some of them (Mc Leod gauges) indicate only the partial pressure of noncondesible gases. The reading of mechanical and liquid column gauges is independent of the kind of the gas, while in most of the other gauges the reading is a function of the kind of gas.

According to their over all shape, gauges are usually of the en­ closed (tubulated) type, which are characterized by a sensing zone lo­ cated in an enclosing envelope which in turn is connected to the vacuum space to be measured. These gauges can be regarded as a means of meas-

* For a detailed treatment of the subject refer to: J.H. Leek: Pressure Measurement in Vacuum Systems.Chapman - Hall, London, 1964, -348-

Table 6.1. Classification of vacuus gauges.

Physical property Kind of pressure involved Kind of gauge recorded

Keeh«ni«l Bourdon Diaphragm Pressure exerted Total, independent _ . .. u-tube Liquid of kind of gas. by the gas - Inclined column Differential

Partial; only non- Gas compression - Mc Leod condenBibles

Viscosity of the Decrement gauge Total; depends 8*a Rotating molecular on kind of gas Resonance gauge

Rate of transfer Total; roughly inde­ Radiometer - Knudsen gauge of •omentum pendent of kind of gas

Thermal Pirani gauge Total; dependB conductivity Thermiator gauge on kind of gas Thermocouple gauge Discharge tube Ionization Total, depends Normal hot cathode gauge Bayard - Alpert gauge on kind of gas Lafferty gauge Klopfer gauge

Penning ^auge (cold cathode) Inverted magnetron gauge Redhead gauge

Alphatron

Partial pressure analysers Partial -349-

uring the incident flux density of molecules entering the gauge mouth • The gas entering the gauge may be modified by the envelope (adsorb, desorb, heat, cool). For the lower ranges of pressures, nude gauges can be used; these gauges consist of the sensing element which is mounted inside the vacuum space to be measured without using any gauge envelope.

When selecting a suitable gauge for a definite purpose, consi­ deration must be given to the following points:

a) The pressure range for which the gauge is desired, b) If the total or the partial pressure is to be measured. c) If the gauge reading can be dependent on the kind of gas. d) The accuracy of measurement required. e) Rind of mounting {panael, table).

6.2, Mechanical gauges

6.21. Bourdon gauge

The Bourdon gauge consists of a helical coil of hollow tubing of eliptical cross section sealed at one end and connected at the other to the vacuum system to be measured. A pointer is attached by a me­ chanical linkage to the free sealed end of the helix and moves over a calibrated scale. If the pressure inside the tubing decreases below atmospheric, the tube cross section tends to become more flat, which causes the radius of Che helix to decrease, and moves the pointer.

The readings of the Bourdon gauge are dependent on the pressure difference between the inside and outside of the tube thus on the external atmospheric pressure. Variations in atmospheric pressure can be up to 40 Torr. This limits the lower end of Bourdon gauges to about 20 Torr,

6»220 Diaphragm gauges

Diaphragm gauges measure pressure differences by the deflection of metal (or glass) diaphragms (aneroid capsules) or bellows. The rea­

ding is amplified mechanically, opticallyB or electrically (capacitance PRESSURE (TORR)

102 10° 10"2 10~4 10"6 10"8 10~10 10"12 10 "u 10"16 \ BOURDON -•DIAPHRAGM •*• CAPACITANCE •——•INDUCTION •U-TUBE -LIQUID-OPTICAL •DUBROVIN -s»McLEOD •VISCOSITY-DECREMENT -•VISCOSITY-ROTATING •VISCOSITY-RESONANCE -KNUDSEN •PIRANI, THERMOCOUPLE •DISCHARGE > HOT-CATHODE IONIZATION — BAYARD -ALPERT LAFFERTY -»- -*• KLOPFER — PENNING "-REDHEAD •• 1 ' "-ALPHATRON •* "-FARVITRON I i -> o-OMEGATRON I i < t-QUADRUPOLE [MAGNETIC GAS ANAL.-5 •-» -351-

strain gauge, inductance). For the measurement of pressure in vacuum technology, the refe­ rence pressure of interest is not the atmospheric pressure, but "zero", compared to the sensitivity range of the gauge. In tbe gauge shown in Fig, 6.2., an evacuated beryllium-copper capsule (pressure sensi­ tive element) is mounted in the gauge chamber, which is connected to

Fig, 6.2. Diaphragm gauge with mechanical indication (The Wallace- Ticrnan gauge).

to the system within which the pressure is to he measured. Distortion of the capsule due to the pressure is transmitted through a mechanical linkage (Fig. 6.2) to a rotating pointer which indicates the pressure on a circular dial viewed through a scaled window in the front of the gauge chamber. These gauges operate over a pressure range 0-50 Torr and can be read to 0.2 Torr. -352-

In order to •••sure smaller displacements of the diaphragm than

la ?oasible by using mechanical linkage, optical or electrical methods

are used. With gauges using bellows as sensing elements, and amplifying

the deflection by a light beam reflected on a snail airror which is til- -4 ted by notion of the bellows, pressure changes of 5x10 Torr were de­ tected. These gaugeB were not adopted for general vacuum use, probably because of their delicate mature.

By using electrical methods of detecting changes in the position

of the diaphragm, gauges both sensitive and robust have been developed.

One such method depends upon the capacitance between the diaphragm and

a fix flat electrode. Movements of the diaphragm, in response to the

pressure, changes the spacing between diaphragm and electrode, and there­

for* the capacitance, which can be measured with a capacitance bridge. -4 Such gauges reach sensitivities of about 10 Torr.

The gauge constructed by Alpert (Fig. 6.3) uses a null deflection

technique, and is bakeable to 400-500*C„ The instrument uses a liquid

manometer for absolute calibration^ but the pressure- sensitive element

is the metallic diaphragm. The Kovar cup is divided by the thin corru­

gated metallic diaphragm into region A (Fig. 6.3) connected to the vac­

uum system, and the region B in which the gas pressure 1B recorded by

the liquid manometer. If the pressure in A equals that in B the dia­

phragm is undeflected. This null position is recorded by measurement with an a»c. bridge of the capacitance of the diaphragm relative to an

inserted metal probe. The pressure in A is compensated by that in B, -353-

Fig, 6.3 ' Absolute muiioniclcr

1, Prcssure-scnsilivc ilinphsiitim; 1. IwAine-pUHe of h «v Kovar perforated sheet; 3, Kovar cup—0-to in. wall ihfcfcnKa; 4. Kovnr-to-etoss tubHig;5,capaciIa- tivc probe—J-in. ilia liruss disc moinilcil on Mycalcx post; 6, Ncoprcnc pnskct; 7, Kovar diaphragm—001 in. lliick; 8, tlilferunliM screw for adjusting prohe position

Fig. 6.3. -354-

th* exact anil point of th« dlaphraga being determined by the capaci­ tance. The aathod facilitate tha production of ultra-high vacuus, ami snbsequeat Introduction of pure gee saaplee up to 50 Torr, tilth m accuracy of about 0.1 Torr (ba ualng an oil aenoaater).

_2 Tba •T

Torr. Tba dlaphraga undergoes a movement of 3x10 In for a pressure difference of 1 Torr, and the capacitance method la able to detect de­ flection of about 5x10 in.

Other electrical aethoda are capable of acasuring diaphragm de­ flections down to a few nlcrolnches. By using atraln gauges pressure chances of 10~ can ba detected. The Mutual Inductance is used in aoaa gauges, In which two colla are arranged near the dlaphraga, one coll la energised by high frequency a.c. and Induces e.a.f. In the other. The aanltude of the induction depende on the moveaent of the diaphragm, and tha electrically indicated readings are directly pro­ portional to tha pressure. Induction aanoaeters can aeaaure in the range 10 -10'3 Torr with a sensitivity of about 10 Torr.

6.3. Caugaa using liquids

6.31. U-tube aanoaetars

Tba currently uaed preaaure unit of vacuus technology (see Sec.

2.42), the Torr (millimeter of mercury) results froa tba concept that pressure is expressed by the height of a liquid coluan. Hanometera using liquids consist of s U-tuba partly filled with liquid, having one end connected to the eyatea in which the preasura la to be aeaaurad. -355-

The other end la either open to some reference pressure (usualLy accas-

pheric) or is closed off. A closed end manometer ia first thoroughly

evacuated and then filled to the proper level while still under vacuus

so that, the gas pressure over the liquid ia the closed arm Is negligible

as compared with any pressure to be measured. The open end is connected

to the system, so taht the difference In level between the surfaces of

the liquid in the two arms will be just proportional to Che total pres­

sure in the sytem. The difference in level li is related to the pressure

according to:

P - g . P . h (6.1) 2 3 where F is the pressure (dyne/cm ), p the density of the liquid (g/as ), 2 hCcn), and g-980.7 cm/sec . When the liquid is mercury h expressed In

millimeters is by definition aqual to the pressure in Torr (Sec. 2.42).

Differences of level of 0,1 m can Just be detected by eye, and

a pressure of 1 Torr can be read with unaided eye with a probable error of about 10 percent. For lower valuta of the pressure, varioua causts produce errors: liquid sticking to glass (variable capillarity), irre­ gular light refraction in the glass, dissolution of gasas In the liquid (especially in oil)» and temperature differences.

6.32. Inclined manometers

The scale of manometers can be extended by constructing it on an inclined aide of the U-tube (Fig. 6.4). tn this arrangement the pres­ sure P (Torr) ia given by:

P * n (1 +^i\) sin a (6.2) P where n-number of milllmetric divisions on the inclined scale, A. and A are the croas sections of the inclined branch and of vertical (prea- surlxed) branch respectively, and a the angle of the Inclined branch to the horizontal.

The ratio A./A has usually values of the order of 1/200, thus it can be neglected

Fig. 6.4. Inclined manometer.

6.33 Differential manometers

If the tuba diameter la sufficiently large (about 1 cm) and the tuba and mercury are kept clean, a manometer can give accurate readings down to 10 Torr, by the uae of optical weans of magnifying small differences In level. Two arrangements of this sort are shown in Fig. 6.5. The column height difference due to preasura difference between C an B (Fig. 6.5.a), ia measured by tilting the framework on which the bulbs are fastened and observing the deflection on a mirror 0, Another system (Fig* 6.5*b) includes the mirror K In the vacuum system. Change* in the level of the mercury, act on the floater b and rotate the mirror on a The range of these gauges la 1-10" Torr. 6.34 The Dubrovln gauge

This gauge consists of a glass cylinder partly filled with mer­ cury and a stainless steel tube closed at the upper end and open at the bottom, floating vertically in the mercury (Fig. 6.6). The gauge is prepared for uae by laying it on its side so that the open end of thr steel tube is exposed and evscuatlng the gauge so that the residual T° Irl PUop

FIR. G.5. Differential manometers with optical a fin ifIcaclon. a) Titling system; b) Lever system.

pressure In the ^augc and ntcel tube be very low. While sttll eva­ cuated tho gauge la returned to the vertical position* Mica low prcs- iurt gal Is admitted through the connection at the top of the gauge, tha steel tuba Is pushed down more deeply In the mercury. For a pros- aura P in the gauge the balance la reached when the weight of the tub" plua the force exerted on the cloned end of the tuba by the s*a pres­ sure la equal to the change In weight of the displaced mercury. If D. and D» >re the Inner and outer diameters of the steel tube, and P Is tha steal density, then 2

2 2 -£- P + J

Proa 6*3. It reeulte that. 2 2 B2 ""I 1 T • * z g [p. «.-h) - P. I] -

Dl (6.4) D22-Dl2 V. which gtvea for thtt zero paelcion, I.e. P-0,

(r..s> o P. thus eq. 6.4 say ba written 2 2 D2-Dl P - Z a1 « P. > (6.0) The sensitivity of the Dubrovin gauge fros eq. 6.6. in: 2 •"• 1 1 ,, ,.

Since 1/g . p is the sensitivity of an usual U-tube oanose:cr

(eqt 6,1), the sensitivity of the Dubrovia gauge is greater by a factor 2 of °1 . For D - 1 cm, D - 1,05 (vail tbickBess of 0.025 aa), 2 2 1 z 2 1 *» this gaug-e is ten times nore sensitive than an U-tufae. For such a gauge a change in pressure of 1 Torr shows a position, change of h of 1 cm, so that pressure changes of 0,1 Torr can he detected easily. The Dubrovin gauge is a convenient instrument for measuring in the pres­ sure range below that of an U-aanometer and above that in which normal­ ly a McLeod gauge is used,

6.35, The McLeod gauge

The principle of the McLeod gauge was explained in Sec. 2.21 as an example of the use of Boyle's lav. The McLeod sauge also known as compression ^au^e^ls the instrunent most frequently used tax absolute -6 pressure measurements in the range 1 Torr-10 Torr. The sensitivity of the HcLeod gauge aay be defined as dh/dP, the change in height of mercury for unit change in pressure. As it vas shown in Sec, 2,21, the McLeod gauge can be used either vitfa a quadratic or with a linear Bcale. In the first case the pressure Is given by eq. 2-5, thus the sensitivity will be

dhl V <6 8) W " A . (l4h)1 - therefore it is increasedas the volume V of the bulb is increased and aa the cross sectional area, of-the capillaries i* decreased. The sensiti­ vity is also a function of (Ah),, thus it is different at different points of the pressure scale, the maximum sensitivity occuring at snail values of (Ah)., i.e. at Lou presButes, -360-

Fox th« case using a linear scale, th* pressure Is given by eq.

2.6, and the sensitivity is

dh« «

dT-t.fto4) «•» thus it is independent of (Ah), i.e. constant et all points over the

range. The sensitivity nay he Increased by making h -h smell.

The ultimate limitation on sensitivity is determined by the prac­

tical limits to the values of V, A, (Ah)^ and ^-h^. If V Is very

large the quantity of mercury required to fill the gauge is excessive,

and the weight of mercury tends to distort the glass and thus falsify

the readings. The reasonable values of V is about 500 cm • In capilla­

ries of lass than 1 MI diameter it is found that surface ten*ion forces

hold the mercury in the capillary tuba even whan the mercury in the bulb

is lowsred. The limit of (Ah)1 and hQ-hs Is set by the difficulty in measuring these length accurately when they are lass than 1 mm. Thus maximum sensitivities of KcLaod gauges are in the range of 3-6x10 mm/

Torr.

The range of pressures which can be measured by a KcLaod gauge

is also determined by A, V, (Ah^ and the total length of the capillary

tube. Since the minimum length la about (Ah). - 1 am, and the maximum

la about (Ah)~ • 100 am, the pressure range which can be measured with

a quadratic scale (eq. 2.S) is of four decades, and that of a linear

scale (eq. 2.6) of only two decades.

The HcLeod game does not measure accurately the contribution of the vapours in the system. It Is only accurate for gases obeying Boyle's law. If the vapour does not become saturated during the compression in the capillary, it can be considered that Its behavior Is veil approximated by Boyle's lav, tliua thot the gauge vlll read tine total preayure «!uo t.. both gas and vapour. On the other haati, it condensation does take pin«c,ch« gauge will read the pressure due to the gas plus the saturation vapour pressure, due to the condensed vapour. A nethod to determine If con­ densation ha8 occured la to compress the gas-vapour nixture Into a length * of the closed capillary, by a head of mercury of height h (Fig. 6.7). If v Is the volume per unit length of the closed capillary, then -*v is the volume of the compressed gas.

With a mixture of gas (P > and vapour (P > g » h - P + P (6.10)

fig. 6.7. Estimating condensation by a McUod gauge. -362-

thas

lb - tr^ + i , ?v (6.11) end ftllK* I P -|-- const <6.12> eg. 6.11 is written

Hitk no vapour present

The most ccessoa vapour present in a vacuus system is water va­ pour. Assuming that water vapour obeys Boyle's law down to its conden­ sation point, than condensation occurs if

m V *v' *v M . (Ah)i i sat. vap. pres. at gauge CMP.

Similarly, the ninlausi value of the head (Fig. 2.7) oE mercury (&h), which can cause condensation is also numerically equal to the value of the saturation vapour pressure Is Torr. This would be the casa with oftly water vapour In the system, which is unlikely to arise in prac­

tice. Thus it Pv' and Wh)j *re set equal to the saturation vapour pressure (s.v.p*), the minimum partial pressure of water vapour which 2 can condense on compression is equal to

P for condensation to occur Is greater than. 10 Torr. the calcula­

tion must be performed for each individual gauge, but it can be seen

that the greater the value of V or the smaller the value of A, the lo­

wer is the water vapour pressure which may result In condensation. The

maximum practical value of V - 500 en and the r-loimum capillary dia- -4 meter of 1 mm. lead to minimum critical value for P of 4x10 Torr. " a v

Further limitations on the use of the McLeod gauge is set by the

connecting tube and the outgassing of the gauge bulb.

The conductance of usual connecting tubes Is not more than 0.1

liter/sec, so that a small leak at the gauge end of the tubing can

give rise to an unexpectedly large discrepancy between the pressure in

the system and that measured by the gauge. Such an error car lie esti­

mated by closing the gauge connection next to the system, and measuring

the pressure rise in the gauge due to leakage (for a time interval of

about 5 din.). As an example assume that:

_2 P (the reading at time t after closing th« gauge) * 10 Torr

V (volume of the gauge) • 300 en

D (diameter of connecting line) - 0,5 cm

L (length of connecting line) - 100 cm

The volume of the connecting line is irD2 . 3 V - —r- L cm -344- ud th« leak rate Is given by

Q . .*, °, (V + V) 10J Torr. liter/eec (6.13) t g c and the conductance (eq. 3.94) by 3 C - 12.1 |- liter/s (6.14)

If * leakage axlsts near tit*gaug e the pressure In the system

is r different f ron ?0> and

AP - £ (6.16) For the nuaerical values of the above example:

!_j0*.2oa5 inn . TO £ «33 V. . * •i 100 - 19.6 CM

2 1 0 -1 3 6 Q - J30 0 319.6 x 10" - 9.6 x 10" Torr liter/sec

Z C - 12.1• STJ 164P -' 1.5 x 10~ liter/sec and the error Is B.?'6'^ - 6.4 x 1(T4 Torr. 1.5 x 10"'4 _3 If with this arrangement a pleasure of 1x10 torr ts read, 64 per­ cent of the reading la the gauge error. Errors of this aagnltudc or such greater frequently appear When the pressure rise test is carried out. Therefore leakage free and large conductance connections are very important for Mrf-eod gauges. Outgeeaing of the gauge bulb and pipe connection may give error* of the same order of magnitude as leakage. Since a spherfcel bulb of -365-

: Vg - 300 cm , has a surface S - 200 cm , in order to obtain Q - 9.6 x

10 Torr. liter/sec from outgassing an outgassing rate of

Q 9.6 x 10-6 a Torr. lit * • s SoT5— "4 '8 x 10 8 — cm . sec, is necessary,, According to Figs, 4.30; 4,32 this is quite possible for undegasBed surfaces, Another source of error is due to the cold trap which is normally interposed between the gauge head and the system to prevent mercury va­ pour entering the system. The trap acts as a mercury condenser causing a steady stream of mercury vapour to flow from the gauge to the trap. This streaming creates a pumping action similar to that in a diffusion

pumps producing a lower pressure within the gauge than that in the sys­ tem. Various systems may be used to raise the mercury. These systems are based either on changing the level of the mercury in its container or on changing the volume of the container. The level of the mercury can be changed either by changing the position of the container rela­ tive to the McLeod gauge, or by exerting pressure onto the surface of the mercury in the container. The first node may be done by raising or lowering a concentric container. (Fig. 6,8.a), or a container con­ nected through an elastic pipe (Fig. 6„8.b)„ or by reversing the con­ tainer connected by a ground joint (Fig. 6.8,c). Figure 6.9 shows means of exerting pressure onto the free surface of the mercury. This may be done by pressing a piston on the surface (Fig. 6.9.a), immersing (magnetically) a plunger Into the mercury^ or hy. changing the air pres­ sure above the mercury (Fig. 6.9 c-e). The volume of the mercury container may be changed by using a rubber ball (Fig. 6.10 a), a rubber pipe (Fig. 6.10.b); diaphragms (Fig. 6,10.e) or bellows (Fig. 6.l0.d), fa» Ibi

Fig. 6.8, Devices lor raising the mercury, baaed on the change in the position o£ the container

J)ig. 6.9. Devices for raisijiR the rap-rcury, by the change of its level in the container -367-

Fig. 6,10. Devices for rai;;£nj; the mercury, by i_hanj;ing cl»e volume of the container,

Forms of McLeod gouges

McLeod gauges are commercially available in bench typer or short­ ened models.

In the bench type (I'tj;. 6.11) the overall height of the gauge ts reduced by using a subsidiary pump to raise and lower tke mercury. To raise the mercury, air is admitted to the reservoir by suitable adjust­ ment of the two-way stopcock. To lower the mercury the stopcock 1P turned to the pump position.

Two shortened models of the tIcT.eod gauge are extensively used: the Heasuvac and the Vacustat.

The Heasuvac type (t*iR. 6.12) uses only a small quantity of mer­ cury which is contained in a flexible reservoir.

The level of the mercury 1B normally below the lower ends of the capillary and reference tubes. In order to raise ther mercury, the le­ ver ia turned. The reference tube also acts as the tube connecting the gaugd Co Che system and is of relatively large diameter to ensure reasor nable high pumping speed. Allowance for the different capillary effects in the reference and capillary tubes is made by the manufacturer by Fig" 6.11. Itencli type NcLeod Fig. 6.12. The Measuvac gauge, geuge setting the reference mark nt the appropriate distance above the cloaep enda of tha caplliarlen. Tim use of two capillaries, respectively associated with different volumes provlden this gauge with a range 150-10-3 Turr.

The Vacua tat la a compact McLeod gauge in which the gauge head la mounted on a panel and can be rotated about ltd centre point (Fig. 6.ID). Because the volume of gas (which 1B compressed) la small and the capillary tube la of fairly wide bore, thn gauge Is only suitable to measure pressures in the range 1-10 Torr. Wlien placed In a horizontal position, the Mercury flown Lnto the reservoir and the rest of the gauge la evacuated, fin rotating to the vertical position (Fig. 6.13) Lite Mercury rises into the capillary tubes. The quantity of Mercury used in the gauge ia Just sufficient to rise to the fixed nark on the refe­ rence tube at the lowest mc.i.iurable pressure; slight tilting may be Fig. 6.13. Vacustat. necessary at higher pressures. The gauge uses Che square law scale (eq. 2.5).

Multirawfie McLeod gauges, which extend the range of measurement, are constructed (Fig. 6.14). The single bulb of the usual McLeod gauge

Is replaced by a series of bulbs of volumes V.t V., V. ending In a bulb and a capillary tube B of volume V.. For very low pressures the gauge is operated on the square law principle (eq. 2.5), the mercury being brought to a fixed nark on the capillary tube E and the tube B being calibrated to read the pressure.

For higher pressures the gauge is operated on the linear scale principle (eq. 2.6). Marks are provided separating each of the volumes

V., V_f v. and V, at points K, L, M, and N. Thus for the next pressure range-, the mercury Is raised to fill the gauge to point K, and the height Fig4 6.14, Hulfcirange McLeod gauge. dlCfareoc* 1, between K end the mercury level In D Is measured. The tab* D bat the sane diameter as the connecting tubes at K, L, M> N. The pr««iur« is given by: vt.i p - C6.U) v',2 T +' v3 , + v.. A further extension of the range Is made by raising the mercury to level X,, where the pressure Is given by

If the Mercury is raised only to level M, the pressure Is

P - - 3 3 In a gauge for which V- - 300 mm ; V„ - 19,700 mm* 3 J -5 and V, - 120,000 mm , would cover the range of pressures 2 x 10 Co 100 Torr. Cleaning a McLeod gauge. Before filling the McLeod gauge with double distilled mercury, the gauge must be thoroughly cleaned. A recommended procedure Is Co clean with nitric acid, followed by ammo­ nium hydroxide, distilled water and then alcohol. The alcohol can then be easily removed by passing clean, dry air through the gauge* After filling the gauge with mercury and sealing to the vacuum system, the lowest possible pressure is obtained and then the glass is heated to release water vapour, 6*4. Viscosity (molecular) gauges.

6.41. The Decrement gauge

The principle of the method is to observe the rate of decay in the amplitude of a small light pendulum swinging freely in the vacuum. Provided the friction at the pivot is low, the damping forces and hen­ ce the rate of decay, are functions of the gas pressure.

The vertical quartz-fibre (Fig„ 6,15,a) is illuminated from one side and viewed through a telescope with an eye-piece scale. The fibre is set vibrating by drawing the small mass of iron at the lower end (0, Pig. 6.15, a.) to the side wall of the gauge by means of an exter­ nal magnet. To calibrate the gauge, the tine taken for the amplitude of vibration to decrease to half of its initial value is plotted vs. the pressure on a log-log diagram. (Fig. 6.16)

The theoretical study of the decrement gauge leads to the equa­ tion:

' " PHTTB <6-19> where t is the time taken for the amplitude of vibration to decrease by half; F is the pressure of the gas; ti the molecular weight of the gas, A and B are constants determined by the geometry of the fibre used. m .I ft $

fig. 6,15. Decrement type gauges, a) vibrating quartz-fibre; W ro­ tating suspended disc; c) torsion system 1) fixed plates batman which th« disc rotates (oscillates); 2) fibre fixsd at two ends; 3) oscillating (rotating) suspension.

V 5r" \ e u N ^ -- i V * e*' °' s ^-* f i 1 . >

Fig. 6.16. Calibration curves for a viscosity gauge. -373-

The useful range of the gauge Is 10 Co 1 Torr. The low pres­ sure limit is set by the large values of t Involved. As M Increases t Is less for a given pressure (eq. 6.19), thus the operation of the gauge at low pressures lraporves with the heavier gases.

6.42. The rotating molecular gauge

In this gauge a horizontal disc (l.Fig. 6,17) (3 cm diameter) rotates about its vertical axis at controlled speeds up to 1000 rpm, with a second disk suspended coaxtally at a small distance above it (2,Fig. 6.17). The molecular drag (Sec. 2.62) due to the rotation of the lower disk exerts a torque on the upper one. This torque is balan­ ced by the restoring force of the suspension {4.Fig. 6,17), the resul­ ting torsion being measured by the displacement of 'he mirror 3.

The equation describing the effect is

3 - K . W. itf^~ (6.20) where B is the angle of rotation shown by the displacement of the mir­ ror 3; K Is a constant of the construction for a specific gas, W is the angular speed of the rotating disk (1), P is the pressure, M, R , T the data of the gas (eq, 2.91), These gauges can measure pressures from 10 to 10~ Torr.

Fig. 6.17, Rotating type gauge. -374-

G.43. The r—oninct type viscosity (ini

In this gauge* * light suspension is mounted vertically and al­ lowed to vibrate at its resonant frequency (normally between 30 and 300 e/e), tha damping totem* obviously balng a function of tha gaa pres­ sure. By Mans of * photoelectric sensing device an electromagnetic driving signal is created which Maintains the oscillation at a constant amplitude, the whole systesi £o-ming a closed-loop servo-system. The driving signal is equal to the damping losses and hence a function of pressur*.

A commercial version manufactured by Ffeiffer, operates over the _3 pressure range 10 - 100 Torr.

6.5, Kadiotetcr (Knndsen) gauge

The basic element of the radiometer gauge consists of two parallel pistes (Fig. 6.18.a) one of vhich is heated, separated by a distance* The t"ih»"i-*d plate is supported on a sensitive suspension so that a small force acting upon it can be measured by its deflection.

The plates A, and A- at temperatures T, and T, respectively are separated by a distance d (cm) which is small compared with their linear dimensions and with the molecular mean free path (Fig. 6.18,*). It is assumed that the molecules coming from A, (travelling towards A.) have an average velocity V, dependent upon the temperature T. (eq. 2.38),

and those leaving K^t have V. dependent upon T-; V. p and V. repre­ senting the corresponding root-mean-square velocities (eq. 2.40). At any instant there are n. and n^ molecules per cm with V, and V, respectively*

In the sorroundlng gas the average velocity is V , the root-mean- iquare velocity is V , and the molecular density is n, 2 The pressure P. (dyne/cm ) in the space between A. and A. is given

* W. Becker: Vacuum U, 195., 1961. Fig. 6.18. Arrangements of Che radiometer fixed and movable vanes. bf (eq. 2.34);

(6.21)

and a^ and n« can be related to n, by (eq. 2.46):

i n V -in, V, + i n, V, 4 av 4 1 lav 4 2 2av (6.22)

(6.23) it follows front eq. 6.22, that

nl Vlav " "2 (6.24) Substituting for n. and n- In eq, 6.21,

P V 2 V V + (6.25) l ' 5 • I T » lr • av' lav T " 'l/ Vtar* For a Maxucllian velocity distribution V /V -V3B/8 (eq. 2.38; 2.39), thus making

(6.26) -376-

bj using «q. 2.34, and 2,40.

Since the pressure on the outside of the plates Is P, the resul­ tant pressure AP on either plate tending to force then apart is

nP - P, - P - (6.27)

1, + 2 a ne/co 2 •*' Iff Vr " ] y ' Thus it la independent of the kind of gas.

By introducing a third plate Aq

r -3-Pll/^ +V T* I C6.28) 3 # m thus the resulting force on A. per unit area is P. - P-, * iP, i. RE. 1/V

that Is the force op the central vane is independent of its temperature.

A Bore exact treatment of thia theory, taking into account the accommodation coefficients for the vane surfaces and the inside surface of the gauge tube leads to results which differ for various gases, the response to helia and hydrogen being particularly low.

The Knudsen gauge consists of a light vane C (Fig* 6.19) supported vertically at it* centre point by a torsion wire, and of two plates A and B heated to temperature T.. Surfaces E and F receive molecules of velocities corresponding to T while surfaces G, H molecules fron the walla of the vessel, at T. If T. > T there is a net couple on the vane, and tha resultant torsional twist in tha suspension wire is measured by the conventional mirror, lamp and scale. This case corresponds to

T2 - T in eq. 6.27. -3 —5 The useful range of the Knudsen gauge is 10 -10 Torr, but can be extended down to 1Q~ Torr, -377-

ISuspenslon fibr*

JH I CZL

Ftg. 6.19. Knudsen gauge.

6.6. Thermal conductivity gauges

6.61. Thermal conductivity and heat losses

thermal conductivity gauges are baaed on a filament mounted In a glass cr metal envelope attached to the vacuum system, the filament being heated by the passage of an electric current. The temperature the filament attains depends on the rate of supply of electrical energy, the heat loss by conductivity through the surrounding fias. the heat loss due to radiation (and convection), and the heat loss through the support leada to the filament.

Xf the rate of supply of electrical energy is maintained constant, and radiation plus support lead losses are minimized, the temperature of the wire depends primarily on the loss of energy due to thermal conduc­ tivity of the gas, whichfln a specific range of pressures) is a direct function of the pressure (see cqs. 2.111; 2.115). The temperature va­ riations of the filament wich pressure are measured in terms of the change of the resistance (PlranX gauge, Thermistor gauge), or the turn- -378-

perature of the filament is recorded by an attached thermocouple (ther­ mocouple gauge)*

In Sac 2.73 Is shown that In the viscous range of pressures* the thermal conductivity of the gas is Independent of its pressure, while In the Molecular range the thermal conductivity is proportional to the pressure of Che gas.

In connection with eqa 2*113 it was calculated that the heat conduction per unit area from a surface at temperature T « 100'C to a surface at 20*C by air at 10 Torr. (acctmnodation coefficient a - 0.7) is E - 8*87 x 10"3 watt/cn2. If the filament of the gauge is 1 ail <0,OQ25) ia diameter and 4 in U0 cm) long, the surface area -2 2 ia about 8 x 10 en •, and the gas heat conduction of the order of

3 2 4 (8.87xl0" > x <8K1Q" ) - 7 x 10" watte For perfectly absorbing surfaces (black body), the ecergy lo&s by radiation* is W » 5.6 x 10"12 (T 4 - T.4) - r s i - 5.6 x 10"12 (373* - 2934) - 6,8 x 10"3 watt/cm2; thuB the total loss by radiation ia

3 2 4 Er - <.6»axi0" } X CSxlO' ) - 5 J* x 10" watt, which is comparable with the loss due to theraal conductivity of the gas at 1? - 10 Torr, However, since the emisivities of surfaces of dean metals at temperature! in the range 0 - XOO'C ate generally of the order of 0.1, the true loss due to radiation would be of the order of 5 x 10~ watt, so that rtdlatian loss and gas-conduction loss would become about equal at a pressure of

5 x 10 i •• x 10"2 »7 x 10~4 Torr 7 x 10'4

* For tungsten filament the radiated energy is H • 7,5x10 T•'watt/cs . -379-

Slnce radiation Increases Caster with temperature than does gas conduction, an equality between then occurs at higher pressure as the temperature is raised. Therefore, for measurement down to lowest pres­ sure , the filament should be operated at the lowest temperature for which the heat IOBB due to gas conduction can be measured.

The third loss, that by thermal conduction to the support leads to the filament, can be kept sufficiently small by using a filament of small cross section and low heat conductivity„

The losses by end conduction are

W - 2 x 0.239 kA? - c ( dL

6 3 (10 20) - 2 (0,239) (0,14) (4.9 x 1Q" ) °Q" - - 7.9 x 10"6 watt where k is heat conductivity (for nickel k»0.14 cal„ cm/"C in the

o temperature range j-200 C)t A is the cross section area of the wire

,r (diameter-0s002 j, and the factor 0,239 converts from calories to watts. For the temperature gradient dT, it is assumed that the central third of the filament is at the maximum temperature (100°C) and the third at each end have a uniform temperature gradient 3(T - T,)/L. For the example given here the loss by end conduction is much lower than that by radiation.

6.62„ Pirani gauge

The Pirani gauge is one of the most widely used vacuum instruments. It consists of a glass or metal envelope containing a heated (see Sec. 6.61) filament of a metal with a high temperature coefficient of resis­ tance, such as platinum or tungsten. As the pressure in the gauge tube increases, the temperature of the filament and therefore its electrical resistance tend to decrease. The usual control circuit for a Pirani gauge is the Wheats tone bridges, in which (Fig. 6.20) one leg of the bridge is the filament of the gauge tube R 'and the other three legs have resistances nearly equal to that of the gauge tube. It is advan­ tageous to us* two identical gauge tubes in the circuit, one of which Fig. 6.20. Circuit for Pirani gauge.

R* i" evacuated to a low pressure and Bealed off. If the sealed-off dummy cube 1B Mounted adjacent to the gauge tube, fluctuations due to changes In anblent temperature and bridge voltage are to some degree compensated.

In Fig* 6,20, resistances R, and R, are fixed, While R, and R is variable. With the milllampermeter G connected in the VAC position, the balance condition of the bridge is

"2 ' R3 " - Z - (6,30) P It* One Method of i isuring the pressure in the gauge head R , Is to ba- P lanes the bridge by varying R, and calculate the resistance It , a pre- •i p vious calibration permitting to convert the values of the resistance into pressure.

Another method la to keep R_ am* Rj. constant and preset R- and to Measure the out-of-balance current through 6. In this case it is essential to keep the voltage across the bridge constant. The bridge nay be balanced initially at atmospheric pressure, then an increase that the lowest pressures correspond to full-scale readings of C. Al­ ternatively the bridge can be balanced at a fixed low pressure (<10~ Torr) and then as the pressure falls from atmospheric the out-of-ba- lance current decreases. In both these cases the high pressure end of the scale 1B very compressed and the scale becomes more open towards the low pressure end. The usual useful scale extends from 5x10 Torr ta 5x10 Torr.

In commercial forms of the Pirani gauge the control unit (Fig. 6.20) Includes the appropriate power supply, which supplies the recti­ fied a.c. By switching to the position SV (set voltage) the milliamper- mecer G can be used as a voltmeter and the voltage can be set to the standard value marked on the scale. In some Instruments the bridge voltage varies with the pressure in the gauge head and set voltage must be controlled before each pressure reading.

Fig. 6,21. Plran gauge head*

The Pitanl gauge head Includes a tungsten, nickel or platinum filament wire (0.005-0.1 mm diameter) wound in a helix of 0.5-2 mm outside diameter (Fig. 6.21), with a pitch of at least 10 wire dia­ meters to prevent any one turn from shielding its neighbours. This -382-

filamant is atreched between supports usually 6-8 CM apart to which It la spot waldcd.

It Is not possible to calibrate the Pirani gauge from first prin­ ciples, and the calibration is made against another gaug.; (e.g. McLeod). Typical calibration curves at constant temperature (A) and constant- voltage (B), are shown in Fig. 6.22. I ( / V

3, / „• " X , •/

0] I 0 45 60 80 100 Relttire meter reading

Fig. 6.22. PIrani gauge calibration curves.

6.63. The thermocouple gauge

In this gauge a filament Is heated electrically and its tempera­ ture Is measured directly by means of thermocouple. The heating current which is passed through the hot filament is kept constant at a standard value independent of the temperature of the filament. As the pressure increases, the heat conduction through the gas increases (eq. 2.113) and fhe temperature of the filament decreases. The thermocouple (usually spot welded to the midpoint of the filament) responds to tin temperature of the filament and provides a direct reading of the pressure.

From the many thermocouple gauges built, it is useful to show, the original gauge of General Electric (Fig. 6.23), and the more refined -383-

type manufactured by l!as tings-Kay dint (Fig. 6,24). In the CE type gauge Che filament consisted of a platimtm-iridium ribbon 0.0234 by 0.O078 cm in ctoss Bcction and 3.66 cm In length with a Nichrorae-Ad- vance thermocouple welded to its midpoint. The heating current used was 30-50 mA.

(a I (bl

Fig. 6.23. Thermocouple gauge (flE type) a) the gauge head; b) the electrical circuit.

In the Hastlngs-Raydist gauge (Fig. 6,24) the sensitive element consists of two thermocouples acting in parallel and a third thermo­ couple in series to compensate for variations in ambient temperature. The two thermocouples A and B arc heated in series by alternating current from a transformer. Thermocouple C (Fig. 6.24) connected frora the midpoint between A and D to the center top of the transformer pro­ vides temperature compensation. Since thermocouples A and B ore connec­ ted "back to back" in the a.c. circuit, they act as parallel sources of

electromotive force for the d.c» circuit for which die lead from C( through the d.c meter to the center tap is the common return path (see equivalent circuit Fig. 6.24), These gauges are available for ranges of G.1-2Q Torr, 5xlO-3 - 1,0 Torr and IxlQ-3 - IxlCT1 Tore. «- oproximately. These gauges have a speed of response shorter than other -384-

Gguqt tubi "COUM"I»6»"

Tig. 6,24, Hastings thermocouple gauge and Its equivalent circuit.

I 10 SO 100 600 1,000 Pre»ur«,mi trout

Fig, 6.25, Calibration curves for Hatings gauges, heat conductivity gauges.

Figure 6*25 shows some calibration curves Eoc different gases.

6.64, The tharalstor gauge

Iha thermistor gauge Is a PIrani-type gauge which employs a semi­ conductor elenent having a high negative temperature coefficient of resistance. The principal advantage of this type of gauge is that the response curve of the bridge current as a function of the pressure nay be essentially linear over a wide pressure range i£ plotted on a log- log scale. -385-

6,65. Combined McLeod - Pirani gauge

IC was suggested to seal a Plrani filament (or a thermistor) In the top of the McLeod capillary. Thus the gas compressed in a certain ratio by the McLeod gauge is measured on the thermal conductivity gauge, recording pressures as low as 10 Torr.

6.7. Ionization gauges

6.71. The discharge tube

The discharge cube (Fig. 6.26) is an elementary form of ionization gauge in which a potential difference of several thousands volts is applied between two electrodes in a narrow glass tube connected to the vacuum system. The ionization in the tube produces a glov discharge,

J^^HLELECTW F^ TUNGSTEN x o , E> WIRE 1 1 MM. D1A. I \

BEING MEASURED

Fig. 6.26. The Discharge tube.

whose form is characteristic for the pressure existing in the tube. The colour of the dlschprge is characteristic for the kind of gas exis­ ting in the tube. At pressures from 1 to 20 Torr. a spark (steamer) of discharge passes from one electrode to the other (Fig. 6.27). At about 1 Torr the spark widens to a glov discharge. As the pressure is still further decreased definite regions (striatlons) in the glow discharge _2 can be observed. When the pressure reaches about 10 Torr the number of collisions is not sufficient to maintain an easily visible discharge. The electrons, however, borabnrd the walls of the tube and fluorescence * _3 of the glass nay be observed. The fluorescence disappears at about 10 Torr, condition which is known os black-out. Spark OiKhtrg* e* charactoritlic colour

5tHr' torr + - 10"' lorr +

Fluorescence-*^

tig. 6.27. Appearance of the discharge at various pressures.

To iaprova Che correlation between the observation of the nature Of the glov discharge and the pressure, two aain techniques have been adopted. The first technique Is to Measure the applied potential differ­ ence across the discharge tube in terns of the length of the spark between ollsbad awtal spheres of a given size. The second technique is to inclu­ de a fluorescent screen in the discharge tube, and use the intensity of the luminescence of this scrten as the indicator of the pressure. 6,72. 1 tot-cathodeIonizatio n gauges

In May ionization gauget the residual gas existing in the gauge head is subjected to Ionizing radiation and some of the gas molecules become ibnlzed. Itot cathode Ionization nauRCR use the thermionic emission of a. cathode, the enitted electrons being accelerated by the'electrostatic field through the grid of radius r (Pig. 6.28) set at a positive potential V (*200 V) relative to the cathode. The anode of radius r is set at a negative potential V {» -20V) relative to the cathode. Hie grid Is made of fine wire, thus most of the electrons conning from the cathode miss the grid vires and continue towards the anode until they reach a point at which the electrical potential is the snme an that of the cathode. From this -387-

Eig. 6.28. Typical electron trajectory In a hot-cathode ionization gauge.

point (shadded area, Fig. 6.28) the electrons are turneii back to oscil­ late radially, through the grid until they finally strike a 'grid wire and are captured. The oscillating electrons will eventually collide with gas molecules, and ionization of the gas molecules may occur. The positive ions which are created in the annulus between grid and anode are driven to the anode, and produce an ion current. In order to pro­ duce ionization by impact with a molecule an electron must have a ki­ netic energy at least equal to the ionization potential of the pas

(12.6 V for water vapour and oxygen; 15 - 15.6 V for H2, Il„, Ar; and 24,6 V for lie). The probability of ionization e is defined as the fraction of electrons at a given energy producing an ionizing collision {the number of ions produced per electron) per centimeter of path and per Torr of gas pressure (Fig. 6,29).

Since at a pressure of 1 Torr and temperature of 273°K the mole­ cular density is n. » 3,54 x 10 cm- (cq. 2.18) the probability of Ionization e is

16 c - na o( = 3.54 x 10 n (6.31) -388-

where o. la the croaa section for an ionizinp, collision by an electron, Tha number of lona produced by an electron per cm of path 1B according to «q. 6.31 and 2.21:

+ n - n a± - %2f . ^ ot - ^ E P (6.32)

where n ia the molecular density at Pand T. For an electron stream of current i_ aatperea*, the positive ion current 1 (assunlng all the lone •re collected) la

i+-f cP.i, (6.33)

According to thla equation the Ion current is a function of the pressure, the temperature, the electron current, and the probability of Ionization (Fig. 6.29) thus of the electron energy, the path length mud tha kind of gas.

Fig* 6*29, Number of positive ions produced per electron, per cm. of path at 1 Torr and 0*C, as a function of the electron ener­ gy, 1. Acetylene; 2. Oxygen; 3. Nitrogen; 4. Argon; 5. Hy­ drogen; 6. Neon; 7. Helium.

* l'A- 6,24 x 10 electrons/sec. The sensitivity s_ of an ionization gauge Is defined by

i+ - s i_ P (6.34) thus the value of a could be calculated from eq, 6,33 as a function of e, which is expressed per cm, of path (Fig, 6,29)0 Since the average length of path of the electrons is not easily estimated for practical

tube geometriess and the energy of the electrons varies from the maximum value of the grid voltage V to zero, the practical way of sensitivity calibration of ionization gauges is against a McLeod gauge. Measurement -4 of the sensitivity s as a function of the pressure in the range 10 - 1 Torr shows that s increases with increasing pressure until a maximum is

reached and then decreasesB The maximum sensitivity for nitrogen is -1 -3 (10 - 20 Torr ) at about 2 x 10 Torr, while that for helium at about _2 2 x 10 Torr, due to the difference in the ionization probabilities (Fig» 6„29), The rise in sensitivity in the vecinity of 10~ Torx for nitrogen is caused by multiple ion production by each electron when the mean free path becomes small compared with the avarage electron pattu The decrease in sensitivity at higher pressures beyound the maximum is attributed to ion-electron or positive-negative ion recombi- -3 nations 0 Operation of the ionization gauge at pressures above 10 Torr greatly shortens the life of the cathode> thus the upper limit of operation is set at about this pressure. The lower limit of pressure which can be measured with a conventional ionization gauge is determined by the production of soft X-rays by the electrons. These X-rays posess sufficient energy to cause the photoemisslon of electrons from the anode. Electrically the emission of an electron by the anode is equivalent to the capture of a positive ion, leading to a total current in excess of that due to the positive ions„ The photocurrent appears to be indepen­ dent of pressure and is of the same order of nagnitude as the ion current at about 10" Torr,, thus at lower pressures this background will be

always, shown by the gaugeB

The common ionization gauget

An ionization gauge for the range 10" - 10" Tbrr (Fig, 6030) is similar to a triode valve. In the centre is a tungsten filament (ca- thod*) and surrounding It is a helix of nickel wire (grid)* The anode (collector), is a cylinder of nickel concentric with the grid and fila­ ment.

Mg« 6.30* Ionization gauge head, and control circuit*

The voltage* used in the gauge nuat be highly stabilised to prevent spurious variations In current, which would give apparent chan­ ges in pressure as is shown by the microsometer in the anode circuit (Fig. 6.30). In this connection it is particularly important to regu­ late the amission of electrons from the cathode, which is done by some font of feedback circuit.

The gauge racorda the presence of all gases and vapours (total pressure), but the sensitivity is different for the various gases (Fig, 6*31).

Initial operation of an ionization gauge results In the heating of the electrodes and the emission of large quantities of adsorbed gasss from the surfaces. Unless the gauge elements are heated vigu- rously to outfisss then, the reading will remain high as compared with the system pressure* The grids can usually be heated electrically, the anode can be heated by electron hombardanent by connecting the Pressure, tnrr x 10*

Fig. 6.31. Calibration curves for ionization gauge.

anode and grid together at the same positive potential. Finally, it is necessary to heat the glasB or metal envelope o£ the gauge tube.

After the gauge head has been outp.asaed, gas entering the tube is readily adsorbed especially on the tube walls. Chemical reactions induced by the hot filamdnt' produce further sorption. These processes are responsible for- the pumping action of gauges, which causes the pressure at the gauge to be lower than that in the sytem.

If diffusion pump vapour in present in the Bytern, these vapours react with the hot tungsten and change the electron emission, which produces an apparent change in pressure. It is therefore essential to employ a liquid nitrogen trap between the diffusion pump and the gauge. tf diffusion pump vapour is present in a system operating at very low pressures, a normal tubulated ionization gauge indicates lower pressures than the nude gauge. Thin phenomenon known as the Blears effect, lies tn the vastly different conductance of the small connecting tube for oil' vapour and permanent gases in conjunction with cracking of the oil -392-

molecules by th« gauge•

The Bayard-Alpert gauge

The design of Bayard-Alpert, reduced the low pressure limit to about 10 Torr. In this gauge (Fig. 6,32) the positions of the ca­ thode and anode are reversed. The gauge (Fig. 6,32) consit of a cylin­ drical grid structure with a fine wire Ion collector (anode) along Its axla and a cathode located just outside the grid structure to one aide. With this arrangement the X-ray falling on the anode is greatly reduced because of its smaller area.

The usual potentials are applied to the electrodes, about + 150V on the grid and about -45V on the ion collector (anode).

The Bayard-Alpert gauge has a sensitivity of 12 Torr for nltro- -4 gen and has a linear calibration curve over the pressure range 10 - 10

Torr.

The Laffcrey gauge

The low-pressure limit of the hot cathode ionization gauge has -14 been extended to 10 Torr with the arrangement designed by Lafferty in his hot cathode magnetron gauge (Fig. 6,33). This gauge consists of a cylindrical nanetron operated with a high magnetic field. The ion collector and shield era maintained at a negative potential relative to the cathode to prevent the escape of electrons. Electrons emitted by the lanthanum boride coated filament, spiral around the axial magne­ tic field In the region between the ion collector and shield. If the magnetic field is enough high, most of the electons fail to reach the anode, and sons of the electrons make many orbits around the cathode before being collected. Because of this Increased path length, the probability of the electrons to collide with and ionize gas molecules is greatly enhanced, and the sensitivity of the gauge is improved with no increase In X-ray photoemlsslon. At a magnetic field of 250 Oersted, the Ion current is enhanced 25000 times over what it would be without the field, and the electron emission current to the anode drops to 0.02

i. ION COLLECTOR

Pig. 6.32. Th« Bay»rd-Alpert gauge. CERAMIC INSULATOR

CERAMIC INSULATORS

fllg, 6.33, the Laffexty gauge.

£f Its zero Magnetic field value. The ratio of ion current to X-ray thotocurrent la thus increaoed 1.25 x 10 tines ty the application of the oagneclc field, making it possible to detect Ion current at ptes- hures as low as 10" Torr, For an emission current of 10~ k, the Hon current output of the gauge 1B about 0.1 A/Torr and la linear down, to 10*" Torr. This low emtr^ton current makes It possible to operate) the cathode below 700*C, which avoids the difficult/ in which the out- gassing limits the low pressure obtained. The sensitivity of this gauge may be increased and the X-ray phatocurrent reduced even further by the use of a shielded electron multiplier ion detector. In this way a pressure of 10 Torr should be detectable by counting indivi­ dual ions entering the electron multiplier.

The Klopfer gauge

This is an ionization gauge in which the ions are produced by a magnetically collimated electron beajn. Electrons are emitted from a thermionic cathode K (Fij». 6.34), collimated through a series of aper­ tures at various electrical potentials, traverse an isolated chamber

VK • 30 V

Amplifie

Fig. 6.34. The Klopfer gauge.

in which the ions are produced, and finally collected by an electron- trapping electrode T. Ions produced within the chamber between G. and negative relative ot the electron beam and the chanter walls. The mag­ netic field of about 1000 Oersteds is carefully aligned relative to the •cries of apertures through the electrodes G1 to G. so that all the electrons pass through the aperture and arc caught on the electrode T. The geometry of the gauge is such that X-ray photon* emitted by I cannot irradiate J directly, so that electron emission fton J Is minimized. the presence of the Magnetic field further reduces the X-ray effect by causing any electrons ealtted to novo in circular orblta. -2 -11 The response of thie gauge Is linear from about 10 Torr to lo

Tort.

£.73. Cold-cathode Ionization gatigea

The Penning gauge

The useful life of a hot-cathode ionization gauge is determined by that of the Incandescent cathode, which is very sensitive to chemical attack and boabardnent of positive Ions. The cold cathode gauge (known as Penning gauge, Philips gauge or PIG) eliminates this sensitivity.

Fig, 6.35, Penning gauge end its simplified control circuit -397-

Two parallel connected cathodes (fig. 6.35) are used and midway between them is placed the anode. The cathodes are metal plates while the anode is a loop of metal wire whose plane is parallel to that of the cathodes, A potential difference of about 2 kV is maintained between the anode and the cathodes. In addition a manetic field of the order of 500 Oersteds is applied at right angles to the place of the electrodes by a permanent magnet„

An electron emitted by che cathode is accelerated towards the anode by the electric field, but the action of the magnetic field causes its path to be in the form of a helix (Fig, 6.35). The elec­ tron generally passes through the plane of the anode loop until its path is reversed by the electric field due to the second cathode. The electron continues to oscillate in this manner about the plane of the anode loop. Due to the very long path of the electron the ionization probability is high even at low pressures. The positive ions created are captured by the cathodes, producing an ion current in the external circuit. The gauge is operated from a control unit consisting of a rectified a,c„ power supply. The voltage across the gauge head may be standardized by using the milliammeter as a voltmeter (switch at SV).

The range of the Penning gauge Is about 10~ - 10" Torr. The upper limit is set by the glow discharge which appears, and the lower limit by the smallness of the ion current* At low pressures the ini­ tiation of the ionization may be difficult, and in order to start the ionization process ic is necessary to produce a few electrons near one of the cathodes (gamma or beta source, auxiliary filament). It is -2 possible Co switch on the gauge at about 10 Torr and allow to remain on throughout the pump-down.

The inverted magnetron gauge

In this gauge -known also as the Hobson- Redhead gauge - the ca­ thode is surrounded by an auxiliary cathode (Fig, 6.36) outer shell. The auxiliary cathode actB as an electrostatic shield and protects the edge of the openings through the cathode from field concentrations, -398-

tba* avevewtlag field aalaalon. The cathode And auxiliary cathode nrr •ota grounded, but tha currant to tha cathode clona la taken at tlia HMWI of tha true poalttve Ion currant. The anode rod la typically •alatalaaJ at about 6 KV» and tha magnetic field intenalty at 2000 Oeraceda. The inverted Magnetron gau£2 i> effective in the range 10 -10 Tbrx.

9-fQkV^

Wig* 6#36, Inverted Magnetron gauge, the Redhead —matron gauge.

In thla gauge (Fig. 6.37) tlie rniodc conslate ot a cylinder,per­ forated to laprove gas flow. The cathode in nhaped Ilka a spool consis­ ting of an axial cylinder welded on to circular and dleka. The gauge la normally operated with a magnetic field of 1000 Oc and an anode- cathode potential difference of about 6 kV, -399-

Fig. 6.37. Redhead magnetron gauge. -A -9 The magnetron gauge tfl linear in the range 10 - 10 Torr and 1 7 -12 extends according to i • c. P dovn to 10 Torr. This gauge has a pumping speed of about 0.15 liter/Bee. 6.74* Gauges with radioactive sources

Any process which causes Ionization can, In principle, be used as a basis for an Ionization r.augc. X-rays, alpha particles, beta particles, and guma rays are all ionizing agents.

The Alphatron ^auge utilizes a snail source of alpha particles (e.q. a gold-radiun alloy).

The Alphatron (Fig. 6,36) consists of a aource holder and two grid structures,. The ionization curxetit is found to be substantially a linear function of the pressure from 10 to 40 Torr. The lowest llntit of thsaa gaogss la about 10 ~ Torr; but tnclr disadvantage is the necessity of shielding. h— man* -*h-— »»*.» \ \ \ "" ^C" h ^ \ it. ~l\\ln "S* — X. — k i\ *. A — \ "Sa — ^ ^ ^ ^ A *>. O%A\

**%*ONS\

MUM NO iNimn) Hyiiviwoi jo SUNH AAA

Tig. 6.38. Th« alphatron gauge and its calibration curves.

The alpha source can be replaced by a beta Bource, Gaugea using trltlun cuemlcally bound to titanium are also reported. These are ope­ rating In the range 1,0 - 10 Torr. -401-

6,8. Calibration of vacuum gauges*

6,81^ General

In the majority of cases vacuum gauges are used merely tc deter­ mine the order of magnitude of the pressure within the system. There are occasions „ however, when it is necessary to check the calibration provided by the manufacturer, or to have a more precise knowledge of the pressure* The most known methods of calibration are the McLeod gauge method, the expansion method, the flow method, and the dynamical method,

6„62} McLeod gauge method

Calibration may be effected against a McLeod gauge in the pressure

range 10 - 10~ Torr; The use of this method is limited to gases which obey Boyle's law up to the maximum pressure to which it is compressed in the operation of the McLeod gauge. The usual practice is to use a glass or metal chamber (Fig, 6*39) evacuated by a liquid nitrogen-Crap­ ped diffusion pump to which the McLeod gauge and the gauges to be cali­ brated are connected, each through a liquid-nitrogen-cooled trap. A needle valve is provided so that any chosen gas can he admitted to the system at s controlled rate to vary the pressure, _3 For calibrating thermal conductivity gauges in the range 1-10 Torr, a rather insensitive McLeod gauge is sufficient. For calibrating ionization gauges, the greatest McLeod gauge, is required, and even so -3 —6 the calibration is possible only in the range 10 - 10 Torr. 6,83, Expansion method

A Bmall known volume of gas at atmospheric (or similar pressure) is allowed to expand into an evacuated vessel of large known volume, and by applying Boyle's law the new pressure can be calculated accurately.

* For details see: J(,HB Leek; Pressure measurement in Vacuus Systems. Chapman-Hall, London, 1964, Liquid- •iliottn hop I luitroflfn frop |(««««J Uquid-mlrogM ond mtti-coaiti bsliln

fig. 6.39. Systea for calibrating vacuum gauges against a McLeod gauge.

The expansion process may be repeated to produce accurately known pres- -4 -6 iures between 10 and 10 Tore, against which gauges can be calibrated. The chief source of error In this method 1B the outgassing of the walla of the vessel, which has to be minimized by tlioruoglily baking the system before use.

6,84. Flow Method

across a known conductance C at a known (measured) throughput Q, by using eq. 3.26

q - C (Px - P2)

Th« lytcm used* for this measurement la shown shematlcally in Fig. 6,60, The conductance la In the form of an orifice or a short tube, and

* C.E. Morautsd, Vaeutm Synp. Trans. 1961 p.534. -4oa-

\ Vapour | I Pump |

Fig. 6,40. Gauge calibration sytem.

the value of C Is calculated From the dimensions (Sec. 3.3). The gas inlet 1B via a narrow bora tube from a large reservoir whprt the pres­ sure is in the range measurable by a simples nanometer. The throughput 0 is calculated by ntesuring the rate of decrease in pressure of the re­ servoir. The gauge to t>c calibrated is used to Indicate P.. Providing the rate of gas flow is snail the pressure r. changes only slowly with time, eo that during the time rtquired for a pressure observation, P.

>> tRan p c Tne fl may be considered constant. If P1 P2» i " Q' * 8 uge may be calibrated over a range of pressures by varying Q, or C.

6,85. Dynamical method

Thia method enables calibration to be made against a HcLeod gauge but at pressures beyound its normal range. The equipment Is Bimilar with that shown In Fig. 6,40, but with the addition of a fast closing valve in the pipe between the gas reservoir and the chamber* The pres­ sure P, in the chamber is Indicated by the gauge to be calibrated and elvo by a McLeod gauge (Fig. 6,39). A steady pressure P. as measured by the McLeod gauge la established and than the valve ia closed. The pressure in the chamber falls exponentially (Fig. 3,37) as. the gas ia -404-

pumped away through the conductance C, and the pressure F after a time t is given by

"it ~£t pt • pl ' e " pl e (6"35) where V is the volume of the chamber, and C is the conductances

The pumping speed S in the chamber is practically equal to the conductance C (eq. 3,28) if the conductance is very small compared to the pumping speed (S ) of the pump.

Thus the pressure in the chamber after a time t can be calculated, and compared with the reading of the gauge under calibration.

The main precaution to be taken in this method is to ensure that the amount of gas from outgassing of the walls is small compared with that pumped from the closed vessel (see Sec. 3.73).

6.9. Partial pressure measurement

6.91. General

Besides the measurement of total low pressurest the importance of analysing the residual gases in vacuum sytems is increasing as the attainable lowest pressure decreases. The Instruments measuring the partial pressure of the residual gases are basically ionization gauges in which the ions formed are resolved into a mass spectrum, the inten­ sity of each component being measured separately.

The mass spectrometer has proved a oust useful instrument in the -4 measurement of partial pressures below about 10 Torrt down to about 10~" Torr. There are several forms of the instrument but the general principle of operation is common: the gas molecules are firstly ionized, then accelerated and finally separated into groups according to their masses.

The means of ionization is fairly standard, the gaa molecules being bombarded by thermlonically produced electrons. The acceleration is done by electric fields, while the separation Is done by using mag­ netic deflection, resonance or time of flight techniques. 6.92. Magnetic deflection mass apectroaetera

In the magnetic deflection mass spectrometer the ptoceoa starts from the Ionization chamber which is a small box in which positive ions are formed by election impact ae In the conventional ionization gauge. The positive lone formed are drawn out of the ionization chamber through t. narrow allc by means of an electric field. The ions are de­ flected through 60 - ISO" (fig. 6.41) by a magnetic field normal to the

Accelerating '• and faamnq electrode

-Collector definmq iff •Ion collector

chamber

{bj Magnetic field i. plow of poper

Fig. 6.41, Magnetic deflection mass spectrometer, a) sector instrument; b) 180* deflection Instrument. direction of motion of the ions. For appropriate values of Che applied voltage and magnetic field all Ions of a given charg*—to-*aas ratio ze/n are refocuaaed at the point of the collector slit (Pig. 6.41), through which they pass to impinge on the ion collector and be recorded. The kinetic energy of the ions issuing from the ion source is -406-

(6.36)

where l-Mss of the ion; v-veloclty of the ion (ca/aec); e-electron chars* (*au); x-nuaber of electronic charges carried by the- Ion; V- •pplied voltage (Volts)J c-velocity of light (cn/sec).

The radius of curvaturt of the orbit for an ion In a nagnetlc field is »-frH <6-37> where B Is the magnetic flax density (Gauss)> Froa 6.36 and 6.37, it results ,10 In V 2 x 3 x 10 •10 « V M \ 4.8 x 10 •10 '5' (6.38)

Since the mass of an atom of unit atonic weight is 1,66 x 10-ill g, the -24 mass of an atom of atonic weight H is i - 1.66 x 10 K, thus eq. 6.38 can be written;

144 M V (6.39) B z

As an example, for a slightly charged atonic oxygen ion (E - 1( H - 16), and V - 1560 Volts, B - 3600 Gauss, R (M - 16) - 6.35 cnt,so that if the instrument has a 180' deflection (Fig. 6.41,b) the collector slit should be located 2R - 12,7 cm fron the slit of th* ion source. The radius of curvatura of the orbit of an atomic hydrogan Ion (z - 1; H - 1.008) will be Rg - 6.35 (Jg)* - U59 cm, thus th* collector slit for hydrogen should be located at 21^-3^18 cm from.the slit of the ion source. TJi* accelerating voltage required to record Che hydrogen

ion at 2K. - 6.35 en is V"H - 1560 T^jg- - 26700 V, which would be some­ what impractical. By providing one or two collector slits, and varying the voltage In • reasonable range the various parts of the mass ranges can be scanned, T?or aach arrangement of collector slit, by imposing the acce­ lerating voltage horizontally on an oscilloscope and the Ion current received through the collector silt vertically, a trace aa shown In Fig. 6.42 is obtained. Such a trace shove current peaks which are

60- 55 $0 43 AO 35 32 30 25

Fig. 6.42. Mass spectrum.

roughly proportional to the partial pressures of the gases of the various mass numbers.

It is obvious that the greatest resolution of Ions Is obtained by using a deflection of ISO". However this requires a relatively large Bvsten and a magnet of large pole-piece area since the field oust be uniform over the whole Ion path, More conpact and lighter aa^s spectrometers with deflections of 120*, 90*, and 60* wen built, but 6h« resolution of these instruments is not so high. The sensitivity of these instruments, when using a photomultlpller a* th« Ion detector p«mltB to measure partial pressures as low as 10" Torr at room temperature* and 10* Torr by taking the multiplier down to liquid nitrogen temperatures.

5,93. The trochoids! (or cycloldal) mass spectrometer

In these lnstrunents (Fig. 6.43) the positive ions are acted upon by crossed static electric and magnetic fields, in this situation the path of the Ions are trocholdal, the distance d between entry and col­ lector slit being given by

, 2n£H (met res > (6.40) where E Is the electric field (V/m), B the magnetic field (Wb/m ), e the ionic charge (C) and M the ion mass (kg) *

The instrument (Fig. 6.43) has two fixed slits, and by adjusting the electric field, ions of a given value of M/e which originate behind

[on source 0

00 Magnetic lleld

Fig. 6.43. Trocholdal mass spectrometer. one silt can be caused to pass through the other and fall an a collector. The mass spactrum can be scanned by varying the electric field over wide limits, keeping B constant. 6.94. The cmegatron

The omegatron ia a mass spectrometer using the principle o£ cy-

clotton resonancea The positive iona move perpendicular to a nagnetic field and are accelerated along helical paths of ever Increasing radii (Archimedes spiral) by a sinusoidally alternating electric field.

FILAMENT —

ELECTROMETER

Fig. 6,44. Drnegatron (Alpert and Buritz*)

This is somewhat similar to tlie cyclotron, where ions move in circular paths being accelerated with a sudden increase of rndtus twice per revolution. In the omegatron (Fig. 6,44) a narrow beam of electrons passes from the filament to the electron collector parallel to a mag- istic field B. Above and below the beam are the two plates which provide the r.f. field. Ions formed along the central axis by electron impact are accelerated by this field. If the resonant frequency oE the Ions In the magnetic field is the same as the field alternating frequen­ cy they will gain anergycontinuously and so move with ever Increasing radius, until they strike the collector. All ions not in resonance have no continuous build-up of energy and hence remain in the vicinity of the central axis.

* D, Alpart and R.S. Buritz, J. Appl. Phys. 25, 202, 1954. -410-

A alagly charged lea Moving la * direction perpendicular to a aelfora aegaetic (laid aovaa la a circular orbit of radlua X accor­ ding to eq. 6.37, with l-X. Thua tha period of rotation It

C - ^-5 - 2w ^ (aac) (6.41)

ao that tba rotational frequency la

1.53 x 103 | cyclee/eec (6.42)

nine* a - 4.8 x 10_iu eeu, c - 3 x 1010 ca/sec; and a - 1.66 x 10~24 M.

It can ba aaan that tha frequency of cha notion rcaalna cooatant (eq. 6.42) nn If tha kinetic energy, thua the radiua increaaee C«q, 6.37). If in tba plana of tha motion thare ia superiaposed an alter­

nating electric field of etrength E » E0 ein wt, then provided |w-2*£|« 2vf the partidaa folloe approximately aplral patha with a radius given by

The radlua of the path thua paaaaa through auccaaalve maxima and minima, except in the apeclal caae of "resonance" whan v - 2»f , vhea the radlua laereaaee indefinitely. Therefore the iooe having an H ao that tba frequency given by eq. 6.42 ia in reeonance'vlth that of frequency « of the applied field, will reach tha collector. lona vblcb are not In reaonance, oaclllata in radlua (aq. 6.43) hut never gat to the ion collector.

Coaaarclal oaegatrone are able to ahow a total praaaura of aax. 1 x 10 Torr, and detect partial praaauraa of tba order of 10~ Torr.

6.95. The Tarvltron

The Farvltron la a linear raaonauca aaaa apactroaatar, with a cylindrical eyaaetrlcal electrode ayataa with a catboda K (Fig. 6.45), an electron collector A and an ion collector S. Because of the geome­ try of the electrodaa and tha d.c. voltagaa applied, tha axial potential distribution la approxlaately a parabola, that la 4 • V - b , in which -«11-

V la the voltage applied betveen Che two end eleetrodea and cha central ting alactrcda. An Ion of chere.e-to-meea ratio a/n Injected Into nuch a

W A 3 JLJI r,r~,~""i ill "-!-•. -I

\^

Fig. 6.45. Schematic diagram of the electrodes and of the potential distribution of the Farvltron. field experiences an ax1*1 one illation of frequency f-^jivj (6.44) where L in the distance between the end electrodes at which the elec­ trical potential •• 0. If an alternating potential of frequency f tc superimposed upon the d.c. potential, an ion of «/• satisfying the above frequency relation will resonate and Rain sufficient energy to escape frra the potential pocket.

In the farvltron the ions are produced by accelerating a regulated currant of electrons froa a tungsten filament K (Fig* 6,45; 6.46) axially into the alcctroda the end of which is a wire sash. The electrons start froa a cathode potential of -100 V (Fig. 6.46), and will therefore pene­ trate the parabolic field to a depth of -100 V, producing positive ions by collieions with any molecules present. These Ions oscillate In the parabolic field, aoat of thsm not having sufficient enrgy to reach the -412-

cwp-thapmd electrode 5. However, when an r.f. voltage it applied to the electrode) on the left, ions of the e/a corresponding to the fre­ quency (eq. 6*44) escape to the collector S.

The Farvltroa ie e bakeable instrument; the sensitivity In appa- —8 rcntly Halted to partial pressures not less than about 10 Torr.

Fig. 6.46. Circuit diagram of the Fnrvltron.

6.96. The qoadrupolg

The quadrupole maun spectrometer docn not require it magnetic field, and in consequence In much \enn bulky thnn magnetic types. The Inntru- ment (Pis. 6.47) consIts of four cylindrical rods to which arc applied a combination of d.c. and ji.e. potential*. For a given applied frequency It csn be shown* that only lonn of a particular value of c/m pans thraur.h Che spectrometer to the collector. Ions of different e/m are collected by altering the a.c. frequency. The quadrupole Is able to detect partial -12 pressure of the order of 10 Torr*

A modiflestIon of this Instrument Is the mononole spectrometer. This consists of a single cylindrical rod and two plans electrodes, which act as reflectors giving three electrostatic images of the rod. thus com-

* W. Paul. at. si. Z. Phyn. 152. 143, 1958. -413-

Filamentf" t IlteSde

fig. 6**7. Quadripole »assspectro»eter.

Dieting the quadrupole arrangement as before.

6.97. Tim*-al-fllftht ««ga spectrometer*

The nost straight forward tlam-of-flight spectrometer consists of a pulsed Ion source and an Ion collector at the opposite and of an •vacuatad tuba. The iona are foreed by electron bombardment and acce~ laratad out of source towards the collector by atthat one. or * series of, alactric fields.

The abort pulse of ions traverses the tube of length X»; the transit tie* t.of lone of a particular si/e value, with velocity v la decided by

t - t/v (6.AS)

If the potential difference through which the loos ere accelerated -414-

Uitlally 1» V, than

Ml (6.46)

At tbe torn collector at the far «nd of the tube, ions will therefor* arrive at tiaee decided by •/*, the heavier ions requiring the longer dmmm of eravel, Tbne each source pulse results in a aaas spectrum wfaicb cao be displayed on an oscilloscope. TJjw-of-fllght Inatruewnt* ere uatially able to detect partial _9 pressure* of the order of 10 Torr. -415-

7. IlIGlt VACUUM TECHNOLOGY

7.1. Criteria for selection ot Materials 7.11. General For the construction of vaeuun systems or vacuum devices it Is conventional to use metals, glasses, ceranlcs aad soae rubbers and plastics. The materials chat becone part of the vacuum sycea, forcing the enclosure (vessels, pipes) oust have sufficient Mechanical strength to withstand the pressure difference, oust be enough lope me able co gases, oust have low vapour pressures, and good resistance to special working conditions (e.g. tenperature).

7.12. Mechanical strength Vacuum enclosures are nade up from cylindrical, plane and hemis­ pherical pares- All these parts tend to defom inwards ss a result cf t..o difference between external (atcospheric) and internal (:ero) pressures. Cylindrical parts tend to collapse easier if their length is greater than the crytical length L defined by

L - 1.11 D c fcr where 0 is the mean diameter and t Che wall thicknes (Fig. 7.1). Tabic 7.1. Usee the permissible value* of D/t calculated for cylinders longer than the mentioned critical length L . These values are satisfactory also for shorter cylinders, but for such cases even larger values of D/t ore satisfactory. Table 7.1. also list1;the values of D./t, (Fig. 7.1) for clamped circular plates, where a deflection & at the centre is permitted. For undamped circular end plates D./t. values are greater by a factor of approximately 1.2. The thickness t- of hemispherical ends have to be at least that resulting froa the ratios R/t (Fig. 7.1) mentioned in Table 7.1. -a •-0—f

Tin. 7.1. Kiln dlaenalons of cylindrical, plan* and haadapherical parts*

Table 7.1 PERMISSIBLE DIMENSION KATIOS FOR PARTS OF VACUUM SYSTEMS

Cylinders End p latas Material Healsphe- rical

0/t c Vi t,/. R/tj

Copper AC 20*C 84 10 52 15 600 Copper ac 500*C 58 8.5 --- Nickel at 20'C 200 11 7J 8 780 Nickel at 500*C 90 10.5 --- Aliolniu. 20'C 70 9 37 57 470 Alualnltai 500 *C 62 8.7 --- Stainless 20'C 105 11.6 89 3 830 Stainless steal 500*C 39 10.5 --- Glass (hard) 20*C 70 9 16 117 470 Heoprent 20*C 2.9 1.7 10 0.2 30 Teflon 20*C 12 3.8 14 9 - FVC (Tygog) 3.7 2.1 --- Psrspax -- 30 -- Mica -- 58 1} - -417-

7.13. Permeability to gases

The metallic, glass or rubber walls of vacuus vessels or pipes are more or less permeable (Sec. 4.22) to gases. The quantity of gas which permeates the walls can be really large as In the case of porous ceraiiics or castings or low as for the case of gas diffusion through "mm porous" walls. Porous, rough-textured, and loosely laminated materials are to be avoided. In the case of "non porous" Materials, their permeability to gases (Fig. 4.12 - 4.15) must be taken Into account in the evaluation of the gas load. (Sec. 8.2)

7.14. Vapour pressure and gas evolution

The materials used in vacuus systems should have a low vapour pressure at the maximum working temperature. Vapour pressure data are summarized in Sec. 4.13. Some aetals (Zn, Cd, Pb) have at 400 - 500*C vapour pressures exceeding the pressures required in high vacuum systeas and therefore these metals (or their alloys) cannot be used. For ultra­ high vacuum work (Fig. 1.1) the choice of metals Is only stainless steels, high nickel alloys and oxygen free high conductivity (OFHC) copper. The gas evolution from the metal surface should be low. To meet this requirement previously degassed materials are recommended, and the outgassing rate should be decreased (Fig. 4.32) by cleaning (Sec. 7.2) and baking.

7.15. Working conditions

The main influence of the working conditions on high vacuum systems la Che temperature Jn soma vacuum systems such influences as chemical corrosion, radiation damge, magnetic field, and others may be very Important In selecting the appropriate materials.

The outgassing for high and ultra-high vacuum (Fig. 4.30 - 4.32) requires baking of the system to temperatures In the range of 4OO-500*C. Small bakeable vacuum systems can be constructed of glass, but any greased joint should be excluded. The mcul bakeable systems are constructed using the various stainless steels. -418-

J.U. Hatal Hmb ana* plpaa Macala ara axeaoaivaly uaad aa building ucarlala tor vacuus alaata (staapa, conoaxlo* plpaa, valval or vacuiai chaabara). Tha aetata and alloya uaad vary from brasa and alualnlua to atalnlass ataal (Tibia 7.2).

Tafcla J.2.

UMUL H*mi o* MAIXXMIJ HM VACUUM VIKVUI hm

MatlCTMI 760 1 MO-1 iO-5-io** IS"*-10-' 10-MO-*

Iron, steel* food food good only after onlyMtiR- deftMlng lesiilwls

One kan, copper or 3lum4nitMit food good Kid bad bad

Rotted

Nidtd and alloy* fOOtl food good food food

Aluminium fiwd food only nricr degasing not iccom- fixflikti

Cilut, qu*t« good good $ood good onty thkV- with detaining willed

Ceramic* gcod gtHid only with vllfWHt*. only special coaling lypcs

Mka good good only after ilrong flOtTMOttt- dsgafiEfif frieHoad

*««*«• good good only bid bid dcfJHkfd

rtaHka" good only apedaF lypcf only Teflon, not rtcem< -419-

A-3 che chamber size increases, the weight of the chamber and the thickness of the wall increases rapidly (Table 7.1). Above saae criti­ cal size the baking of such a chamber becomes very difficult. A solu­ tion in such cases can be the double chamber. Here the outer chamber may remain cool and can be built from common metals (e.g. olid steel) with a wall thickness to withstand atmospheric pressure. The inner

chamber (e„g0 stainless steel) can be made quite thin since it supports practically no pressure difference, the space between the two chambers being evacuated by a separate pump to a"guard vacuum" level.

7„17„ Glass vessels and pipes

Glass is used as the envelope of many vacuum devices (lamps, electron tubes), as bell jars in small evaporation plants, as reaction vessels and connection pipes, and in the construction of some diffusion pumps and gauges (Table 7.2)„

Glass '- a noncrystalline material chat has no regular internal structure. It 1B rigid at ordinary temperatures and almost fluid at higher temperatures. It has no definite freezing point but becomes solid because its viscosity increases progressively to very large values„

Glass is a fragile material, and for this reason its main mecha­ nical characteristics have to be listed according to the factors gene­ rally leading to its breaking: mechanical stresses (tension, bending, impact), thermal stresses and composition changes (weathering, devi­ trification) ,

Glass fractures only as a result of tensile stresses and not due to shear or compression. The useful strength of glass is but a small fraction of its intrinsic strength because of imperfections (small cracks) usually existing on its surface. The useful tensile stress of 2 glass is about 0,7 Kg/inn »

When glass is suddenly cooled, tensile stresses are introduced in the cooled surfaces, and a compensating compression stress in the -420-

•ui of the glass. Sudden heating leads to surface compression and Internal tension. Since glass fails only in tension at the surface, sudden cooling is much tore dangerous than sudden heating.

If the glass is exposed to steady temperature differences between the two faces, thermal gradients are developed through the glass. These are even more dangerous to the integrity of the glass than sudden cooling.

The weathering and devitrification is discussed in Sec. 7,22; the thermal properties of glasses see Sec. 7.32.

7.18. Elastomer and plastic pipes

Elastomers and plastics are used in vacuum technology as pipes and gaskets. The elastomers used are classified in Table 7.3,

The field where elastomer and platic pipes can he used is limited (Table 7*2) to backing lines (Fig. 3,35) because of the gas evolution of these materials (Figs. 4.30; 4.32). The use of rubbers is also limited by their narrow temperature range, which extends to +80*C and -40*C. At higher temperatures or after long time rubbers present "ageing" effects in the form of hardening. At low temperatures rubbers become brittle. Silicone rubbers have a wider temperature range exten­ ding up to 180*C and for shorter heating periods even up to 2SD*C.

Because of their mechanical properties the wall thickness of elastomer pipes must be relatively large (Table 7.1) to prevent collap­ sing. Thinner walled elastomer pipes can be prevented from collapsing by a helical wire spring inserted in the pipe, which reduces the unsup­ ported length of the pipe to the spacing between turns.

The plastics used In vacuus technology are listed in Table 7,4. Acrylics have a relatively high outgassing rate and are not recommended for high vacuum, except for low temperatures or windows of small surface areas. These plastics are extensively used as glove boxes. Fluocarbons are suitable for use at operating temperatures ranging from -100*C to -421-

Table 7,3.

EUSKOMBU

"properties'

Elccir. Flame [mpcf Hoit Cold roetfel"merit. * — resistance Ifiy

Natural rubber Isoprenc

S.B.R, Buna S Siyrcne/butadiene

Butyl I.I.R. Isnprene/isotwtslwie Polyhutailyene

Thiokrf Orgnnic poly sulfide

Nilrfte, Pliil- AcsyfRiticrife/iiffts- prcnc, Hycar, Bunji N, Oil (itid Ferbunan pciro taint PoiS»r«thane Diisoeyit ita I c $><*(y- resist si nl tskr nrpolyelher

Neoprcrtc

f?yria(on potyMliytcne

Silicone, Htirt Silastic

hcxiiniwrnpropy- Icnc

* lit KHffwrf«»i wilh the nlhcr «isK«fticni,' is taax&ktit, G - &fd, P --- fair, I* -

+-300*C, and have « relatively law ouegassirtg rafee, thus they can be used in high vacuum syterns. ralyethylene has outgaasing rates near to that of fluocarboriB, but can be heated just up to fiO-100'C* Polystyrene haa alow outgassing rate but Is tough and brittle, and ±a not suitable for construction of large parts. Polyvinyl chloride Is used as transparent tubing in the backing line of high factum systems, where its outgaaaing rate is tolerable, yinylldene chlorine is satis­ factory for vacuum of the range 10" Torr. -422-

TabU 7.4.

FLAxnca U*D IN VACUUM TI CHNHHJK

Chmktt Common trade Orouf Renwrks* compotUion nrnnct

Aayffcs Priymcthyl met*- LIJCJIC, Pcrspcx TfBMptrcjil. wnlcr crllata Plexlglm fcslstini

Fluocarbom Polyleln- P.T.F.E.. Tenon, Chcmtcitty tncrl (Itiorelhyknc Fhlon hunt, cold rials- • l«nt

PT.T.C.E., Kel-r, As I'.T.I-.I!. Mrtnrcthjtfcne HoMafTon

Polyclhylcne Ptily Ihcnc, Alkuhcnc, Flcvtbk. wnlcr Itostalcn. Alnlhon. n-shlnnt, clicml- Plitxpak ail resistant

Polystyrene Styron, 1-iLiirw, Rnillatfnn rcsmUnt Polysiyrol, Slyro- fnnin

P.V.C. Polyvinyl chloride, Tygon, Kontmitl, tinil acctnlc-cliln- Vlnylilc, Anirnlnn rirtc vinyl capo- lymer

Vfovlfrionq Martda Satan, Vclnn . -• —. —

7,2. 'Clearilitg' techniques

7.21. Cleaning of aetals

Cleaning generally means the removal of undesirable materials laying on the surface. In vacuum technology, the cleaning must be regarded not only as the removal of the visible dirt from the surfaces» but Including the mbsequent removal of all the contaminants physically stuck on the surface (oil, grease, duat) or resulting from a chemical reaction (oxides* sulphides). The degree of cleanllnes* must ba higher, for higher vacuum* -423-

The oxides and other similar surface layers can be removed by mechanical and/or chemical methods, as abrasive blasting, wire brushing or pickling and etching.

The cleaning of oils and greases depends on their nature, i.e. if they are soap-forming or not. The soap-forming oils and greases are those of animal or vegetable origin; the mineral oils do not form soaps. The soap-forming oils and greases can be removed by transforming them by hydrolysis in fatty acids and by reacting these acids with alkaline solutions to obtain water soluble soaps. The mineral oils can be removed by dissolving thera in organic solvents, and, in parti­ cular cases only, they can be washed with alkaline solutions containing detergents. Since the nature of the contaminants is usually '"*'aiown, a reliable cleaning must consist of two succeslve steps: a) the degreasing with, organic solvents, followed by b) an alkaline degreasing.

The sequence of the cleaning operations begins generally with mechanical cleaning, followed by pickling, detergent cleaning and degreasing,

The mechanical cleaning methods are not specific for vacuum technology. For the purpose of cleaning from scale, rust, etc., abra­ sive blasting, or wire brushing is usually utilized.

Pickling is the chemical removal of oxides and other surface layers,leaving the cleaned part with a metallic appearance with a smooth or rough finish, depending on the concentration of the solution and the pickling time. Pickling solutions to be used for various metals are listed in Tabel 7,5,, After pickling the part should be always thoroughly rinsed and subsequently neutralized: in an alkaline bath, and dried with hot (oil-free) air.

Electrolytic etching and polishing is the anodic (or cathodlc) treatment of metal surfaces, in appropriate etching solutions (Table 7.6). -424-

Tftbla 7.5,

Piatt iKO Sowmiwi

Mtttt Pickling lolullnn (Alloy)

N*OH (16 jsoreeM «jl.) wlunted will* N«CI {If NK&» ; At HO *Ct IS-SO citing Mppeafi, Alhu Cu; in this cue lubtequent I sec, Subcoqu- p*ckHiiginHNOjOO-50%)«^«df*o«**'a»',to#) ent Immerakifl ( In HCI (10%) for shining surface. 1^10,(25%sol.)diluted S:l wlih HCICI.I6)

Electrolytic clcliing in ml, of 100 ft HsBOB in MWG ml 50-MMV. up In

•list, water plu\ 0.5 g NaJV>r 6DUV Utectrolylic rtcliine in 5~M% sol of chromic a eld 2!t-4«Vaboui Jit niln.

N.iOII (or KOH) 50-100 g/IO()Oml. dij,t. w»l*r il 20 ~C, electrolytic c idling

Cumianinn f UJSOt (IOX.wlJ.W-WC

250 ml. UNO, U.40) with Mitt ml. I l,SO, (I.B.l) mid 20 ml. IK'KI.lf.lm f.Wwrf.*l. wafer 5*K> tnl. I INf >j 465"„) ««!( SIX* ml. II.KO, <««!«*:.) «nd in ml. llCt I.t7"<„| ;,.t<| 5 pciirhim hlitik i MI up (OHO ml. UNO, with I0fm.nl. 11,-SO, niul ISKNIICI 11-5 mill 20 M carhitit Muck. To IK diluted I: I with (list, wait'r M hi' bffiwc use ID?£ Jrc (SO,), tn citric nciit >'1.l 1.0%) or in iicclic acid (0.1 0.5-J MCI iHl. tmtiKrvtm for 5 mitt fftlfowctt fry immersion

In sol. of I«>K (V.(>s, 7 ml, HjSO, U'i'nc.) in HHlOinl! dfst. wmcr

Sol.

20 $ Nl ljN(>3, 45 till. dim. water. Immersion Cor 3-4 Sol. 2 at mum Otin. Affer fmsttig itttftieriSftMf st Sol (2J: 65 »*). gla­ It-Hip,

cial acetic acid. .10 ml, HaPOj (1.7S), 5 ml. UNO, 0rij:lit dip 41.42). Immersion ahttirt f mm erriil gas cvttivcA evenly all over

lawtr Ciltiodicfitchiii* in 1 vol. 1ICI 07"i) wlilt I vol, II.SO,

II eNi 64/36 <

Metal (AlliivJ Pickling solution

50% Bill, nf HCI or 5 15% Ml. or HrSO<. Recommen- tli'd tt> udd hydrogen «vatu(Jon Inhibitor (e.g. Fcrro- ck'uiinl, sec tispe-"') 300 it (V.,0, witli S ml. HjSO, filled up with dlsL water Section 24.21 Iron- to I (KM ml. H-IOmln Chromium Atmilit etching in 335 ml. acetic acid with 240 ml. ivrvhloric «eid (HCIO,) in 100 ml. dwl. water. 6 V, Gilhtidcof graphite

\-ru UNO, (coiw) with 25% HCI

(I H.H. 13 pt>w HO(I.Ki), 1 phw Nt)3M ',1.40) in 75 pliwdisl. water, iit M> C. I pl'vv UNO, (hS"„) with I pbw utciic iidd'(50%) ""*" About 30 sec

7^ (.• |Mlt)-.SU( in KM ml. 11,50, (20%), iH 50-100"C 3-5 min Kovnr I vol. IK'I CIO"..) wiih ! vol. HN()a(IO%) at 70 "C. 2-5 min (FcNiCo) Siirnun required Ulerirolylic etching: 1"; NaCI in 1101 (10 15%) a.e. ju iZ V. l.fi A/tm!. I'^lLttrndc graphite or Kovar Ml- (40%)tir UC I is n%) 2 ml. ILSO, wilh 37.5 a CrfV,. 100 ml. HF, 10 ml. 11 NO, (cone) at W"C fvfolyhifvrtum

10 ^ NuOtl in 7.10 ml. dint. water with 250 ml. H,05 (.10-35%) at 40 C Elcilrolylic cubing in 20% KOH sol. d.c. 7.5 V. Car- b>

Bolting H,03 (3% sol.) Anndic etching in 250 a KOH and 0.25 a CuSO, in 1000 ml, dist. water -426-

Tabl* 7.6. i Ltcram.vnr Frtmiffwi

Cttrrcat Mctil fl*(h €t

40 ml. H,SO„ 40 ml. phm- ptiortc acid 20 ml. dtnt. water I*.' flil. perchloric acW, W5 ml, acetic add 50 ml. Uisl. 50-100 .10-50

200 ml. HNOa, dOO ml. mc- ttmnal 2.4-2.6 125-150 I

fiO'inl, orchopliosphoric acid, f 20H,5O, 2 diM. water 10 IB 900 ZlOrai. pcrehktrti; srfd, 7w> pil. acetic twld 22 50=ml. f J,S

Alkaline determent cleaning 1B perfonned either by immersion or as an electrocleaning process. The Immersion cleaning is used usually with hot (60-85°C) solutions. For ferrous metals and difficult cleaning operations stronger cleaners, containing sodium hydroxide (silicates or phosphates), soaps and wetting agents are used, in concentrations of 10-40 g of each component per liter of solution. Non-ferrous metals and especially aluminium, are cleaned with inhibited alkaline cleaning solutions, i»e» Bodiun silicates, phosphates, carbo­ nates with soap or synthetic organic detergents, at a concentration

of 15-45 g/l0 Eleetrocleaning in alkaline solutions, can be used with the metal to be cleaned as the cathode or as the anode, the tank being the second electrode. With anodic cleaning, oxygen is liberated on

the surface of the metal being cleaned, and the process requires 2 6-12 V d.c at 50-100 m A/cm . With cathodic cleaning hydrogen is

liberated on the cleaned surface, and the p-ocess requires 6-12 V d.c0 for a maximum of 50 m A/cm". For steels the anodic cleaning is recom­ mended but, for nonferous alloys the cathodic should be used. A recommended solution for electrocleaning consist of: 1 parts by weight caustic soda, 1 pbw trisodlum phosphate and 1 pbw soda ash, in water.

Solvent cleaning is done by using the solvent in a liquid or In a vapour state„ Liquid cleaning can use benzene, xylene (if the neces­ sary safety precautions are provided) or not flamable solvents (dichlo- rethylene C. CI. tL, carbon tetrachloride CC1., trlchlorethylene

C2C1 H, perchlorethylene C.C1,),

Vapour degreasing is much more effective than liquid solvent cleaning* The solvent is heated to boiling, the parts to be cleaned are hung in the chamber in the hot vapour, which condenses on the metal surfaces, dissolves the oil and grease, and flaws back to the solvent containers -428-

7.22. Claming of glass

In order to obtain Inside surfaces sufficiently clean for vacuum purposes, the glass or quarts vessels, pipes or other parts should be thoroughly cleaned, even if the glass is perfectly transparent. More careful cleaning is necessary if the glass has a rough appearance.

The glass may have this rough appearance due to weathering or to devitrification. Weathering is a result of the influence of the atmospheric vapours, and consists of the hydrolysis of the alkali sili­ cates, forming alkali hydroxides and colloidal silicic acid. The alkali hydroxides react with the carbon dioxide from the air, forming a film of alkali carbonates, with separation of silica. To cause minimum weathering the glass is best stored unwrapped or packed in plastics* When the weathering is not too advanced the matt surface can be cleaned by add washing.

The devitrification is a result of recrystallization. Due to this process the glass loses its transparency and becomes brittle. To avoid devitrification glasses oust not be cooled too slowly-. The appearance of the devitrified glass cannot be changed by cleaning its surface.

Glass vessels and pipes (new or weathered) can be washed by immersing them in hydrochloric acid solution (1-5 percent) for 3-10 sec, followed by a subsequent rinsing in water (40-50*C) and drying. Tap water if quickly dried leaves salts on the surfaces. To remove these salts a second rinsing in distilled (or demineralized) water is recom­ mended before drying. For the washing of soda-lime glasses sn acetic acid solution (3-5 percent) is recosnended instead of hydrochloric acid.

Chromic acid Is satisfactory for cleaning glass (subsequent to a washing in water) provided that care Is taken to make sure that the glass is free of mercury. If mercury is present a residue is precipitated which is difficult to remove. The usual cleaning solution known as chromic acid contains about 50 ml of saturated aqueous sodium dichromate In a litre of concentrated sulphuric acid. The chronic acid solution should be used only if It has its brown colour. If the colour is changed the solution is decomposed. -429-

A solution which is much more effective than the chromic acid solution, consist of 5 percent HF with 33 percent HNO_ in 60 percent water» The solution should be used cold.

Weathered glaBs can be washed with EF solution (40 percent in

volumes) by an immersion of 1-5 min0 The glass is superficially attacked by this solution, but it remains smooths A susequent washing with (distilled) water, and neutralizing in NaOH solution is absolutely necessary,, The final washing is made in distilled water (W*C) and alcoholo

Organic solvents are adequate to remove greases from the inner surfaces of glass parts. Silicone grease can be washed from glass surfaces by using dichlorethylene or kerosene with subsequent washing with a solution of lOg NaOH and 5g borax in 10Q ml distilled water or a solution of 10-15 ml KOH (50 percent) in 100 ml ethyl alcohol (maximum immersion 10 miu)„

xne drying of washed glass parts can be done oy hot air (free of oil)„ Acetone or alcohol may also be useful for drying glass parts.

Quartz parts have to be cleaned (before heating) with alcohol. Any material (alkaline, sweat) left on the surface of quartz, causes at high temperatures its reversion to the crystalline state, which show up as permanent marks on the surface=

7023. Cleaning of ceramics

Suitable cleaning of ceramic parts is obtained by firing the ceramic parts in air at 800-1000"^. Alternatively an alkaline cleaning solution can be used, followed by Immersion In dilute nitric acid (2-5 rain.). Chromic acid or other glass cleaning solutions are also satisfactory. -430-

?«24. Cleaning, of rubber Rubber* evolve a gnat Mount of gaaea (***• 4,30| 4.31), especial­ ly when they are ntw and untreated. In order to obtain Lower outga«ain«j rates the rubber must be cleaned with « solution of KOII (20 percent) et 70"C with subsequent washing with distilled water and drying with clean sir, and/or degassing in vacuus at 70*C fox 4-5 hours.

lubbers that are to be used in contact with mercury Bust be treated for at least one hour with NaOH (20 percent) solution at 70*C, After this treatment, the mercury will remain tmcontamiaated in contact with the rubber.

7.25. Balding The sost efficient method of reducing the outgassing rates of the parts of vacuum systems is their baking, the useful range of baking temperatures is 400-500*c (for metal or glass systems), and the efficiency is much reduced if only 150-200*C is used.

7.3. Sealicg techniques*

7.31. General, classifications

A vacuum system or even a vacuum chanter cannot be constructed as a single unit. One must use various components of various shapes and different materials and provide for the possibility to change the parts or to open and close the chambers, These various parts are joined together using various seals, which afford joining the parts but prevent leakage through the joint.

A common requirement and permanent problem of all the vacuum seals is thsir leak tightness. Any vacuum seel must be leak tight but must not necessarily be hermetic, A hermetic seal is designed to

* For a detailed treatment of the subject, refer to A. Roth* Vacuum Sealing Techniques, Pergamon Press, Oxford, 1966, -431-

permlt no detectable leak through It (on a sensitive leak detector such «s a helium mass spectrometer, see. Sec, 7,4), while a leak tight seal Is Just free of leaks according to a given specification.

Besides their function to prevent gas penetration, some vacuum seals must be capable of allowing the transmission of an electric current or of a motion, the transfer of material or the passage of radiation into or from the system.

The classification of seals used in vacuum technology can be based on the purpose of the seal, the requirements, the materials or the construction techniques used*

For the present short description we divide the seals in: permanent seals (welded and brazed metal joints, glass-to—glass, glass-

to-metals and ceramic-to-metal seals), demountable seals (wax and resins, ground^ liquid and gasket seals); electrical lead-throughs; seals for motion, transmission; seals for transfer of materials (cut­ offs, valves, vacuum locks).

7,32„ Permanent seals

Metal parts are joined permanently by welding or by brazing. Glass-to-glass is joined permanently by fusion; glass-to-metal, and ceramic-to-metal seals are constructed by using specific techniques.

Welded seals

Welding is the generic term to describe metal joining processes based on localized melting of the metal produced as a result of tempe­ rature and/or pressure (Fig, 7.2),

The non-pressure welding processes include the techniques of metal joining by the application of heat without the use of pressure. In these processes a mixture of molten metal is formed as a result of the local melting of the surfaces or edges. This liquid metal mixture (to which eventually a filler metal is also added) bridges the gap between th« components to be welded to each other. After -432-

thm source of welding hut has been removed, this liquid solidifies, thue "welding" tha parta together* Tha sources of haat used are; the flaae (ft** welding), the alectrlc arc or the electron bean.

Tha pressure welding processes are techniques for joining rectaI• using preeeure with or without heating the components. The reetstance welding heats the parts, but In eone ceiet cold welding is possible.

I wp I

IMPWWWWewem | IpiWOTprocnT]

Fig. 7*2* Velding Kathode for vacuum sealing.

In gee (torch) welding uelng acetylene or hydrogen flanea the torch cen be adjusted tD produce e reducing, neutral or oxidizing flaw. Neutral (or slightly reducing) flames are preferred In vacuum •aaling. The use of a reducing flame can be the cause of porous wolds due to the occlusion of hydrogen In the weld, a fact especially evident when copper Is welded. The use of oxidising fleets effects the strength of the veld introducing an oxide layer between the parts. The necessity to use fluxes (high ootp««fllnj; rate) Halts the use of torch welding for vmcwm sealing only to joints In very heavy copper or iron vessels. -433-

Age welding ±e based on the heat obtained from an electric arc forned between the work and an electrode or between two electrodes, From the many commercial arc welding systems those used for vacuum sealing are; the atomic hydrogen (areatorn) process, the carbon arc, the alr^omatic and the argon arc (heliarc) process„

In the arcatom process the beat is obtained from the a.c. arc (6Q-1QQ V; 20-60 A) between, two tungsten electrodes surrounded by

hydrogens The molecular hydrogen supplied through the electrode holder is dissociated in the arc to atomic hydrogen and recomblnes on

contact with the cooler metal, producing temperatures up to 4000*Ca The welds are homogeneous and clean;. This method is suitable for iron,

mild, steels, aluminium and chromium9 but is unsuitable for nickel and copper alloysc The hydrogen is soluble in molten nickel (stainless steels) and it escapes when the metal is setting, producing cracks and pores» Copper and copper alloys become brittle due to the hydrogen.

The carbon arc welding process uses a dac„ straight polarity arc (1-10 A; 100 V) between the carbon electrode and the work or between

two carbon electrodesa The process may be used to weld iron, nickel, aluminium or copper „

The alrcomatic welding uses f\ d0e„ current with consumable electrodes and a shielding of hydrogen or argon» The process can be used for aluminium and stainless steel.

In the inert gas arc welding (argonarcs heliarc) d*c,, or a«c« Is used between the work and a tungsten electrode« The arc works in

a shielded atmosphere of argon or helium6 Aluminium and its alloys are usually welded with a„c, (100 V 250-300 A), while d.c, (45-75 V; 15-200 A) is used for steels, stainless steals, nickel, copper, silver

and titaniuma Inert gas arc welding is the most commonly used, welding procedure for vacuum sealing for high and ultra-high vacuum.

The electron beam welding process is carried out in a vacuum chamber (5x10* Torr) within which a stream of electrons la accelerated -434- tftteugh « high potential end focused on tha work place* V*ry aceurafca and cleaa valuta «r« ©brained. Electron bee* welding can be used for atelnlaas steel, alueinlu* alloy a, tungsten, molycdanwe,, titanium an* tantalltsa. In resistance welding tha joining of tha piacee la produced by tfta heat obtained fro* tha raalstance aat up In tha Ratal part* by tha paaeage of high intensity currents, tha parte being pressed againat •ach other. Resletance welding la uaad aa apot welding or seam valdlng (TlK. 7.3).

(o) fc> le> <* *lg. 7.3, Resistance welding, a) spot weld, o) coil apot v»ld$ c) overlapping spot weld; d) seam weld. Butt welding Is the joining of two parts placed with their ends abutting each other, The fusion of the ends la accomplished by flash or up»«t. in flaah welding the parts are placed a lightly separated, tha alactric voltage applied produces an arc (flash) In the gap. At this nonetn tha parts are pressed againat each othar. Upset welding la similar to flash welding, except that the areas to be joined are pressed into contact and then heated by an electric currant which paaaaa across tha abutting surfaces. friction welding is a process Halted to welding tha ends of objects (Cabas, caps) one of which is rotated about Its axis (1500- 6000 rpai). Tha hast Is obtained due to tha friction between tha parts. Cold waldlng. Certain watala can be welded together to give a vacuusi tight seal, by applying sufficient praaaure. Tha pressures -435-

required for cold waldlng Are about 17-25 kg/m for alualnlua, 50-75

kg/an for copper and about 200 kgfrn2 for atalnltaa atasl. The eur- facea muafc be fraa of oxides and vail degreaaed. Tha fact chat the whole procedure t> cold, and practically no gaaaa ara released during tha sealing, make* this procadura adequate for tha seallng-off of evacuated metal tubes.

Heldability. indicates tha amount of precaution* necessary for successful welding. As a guide to the first choice of the welding method which can he used* Table 7,7 summarises the recommended (or dangerous) solutions. In the design and construction of welded seals for vacuo* tech­ nology the following points should bo observed:

1, The joints must be designed and welded with Cull, penetration avoiding trapped volumes in which contaminate may collect. Some recommended arrangements of welded joints for vacuum seals are shown In Fig. 7.A, together with the incorrect constructions which are to be avoided.

| A.Bj(ti 0. 1DB C TM Dl Zorrm l " 1 ...J 1 , ..n , "tdl II) rjVoeuum

i?)l * II fc 1 ,„J •*»n—r~i ft Voowm '"^ 13IP

Vacuum Vbcuunt't 111 l_J 1 (D nwieuwii -..-Jfteuum •^D »4>

Fig. 7.4. Wtlttad joints. •436-

Tafal* 7.7

SUtlH- ... ku PI . 'Nl MA Kc, •M _ c C _..... ». _!». "—.r_ . —. . -.. _-:„ - Be . .T._ _^. ,_'."-. . Draw R. -- ~- Co

Cr „7._. ,-_ .7 _ - CrI-c R, "c c —- CrNi .*? _... Hi R^ C ~ c Cu m __ H, <*>._ (SI J«_ c" - Fc R, "'.._ .. =_ I'cNI Rj C c C Kuvar __ .0) - <» W c~ c A.C Mo R, R> Ri em c f.IRS) Nl 3 R| R| M.R, | C C" PI *• ..*' Sluinless li.MU) nlccl m__

ft" -«7-

T«bl« 7.7. (Cont.)

of MHTMJ Vjt> ALtqvr

Brais All CrNI Cu H» Al

A^C.H M. B. R-. EOI

A

'C.H | T.ll,

T-iordi wckl. A--nrcniiim. t: ciirtunwrx;. II huli (nrgt«t)iin:.M- nln-tttnaitc. 1:. elec­ tron honm. I*- cold wclil. H rmftcincc weld, t wry ca«ry. 2 pind. 3 difficult.

(I) After cleaning with phmphnrlc acid (2> Only OF1IC copper 0) Llmltallnn Tor large purls (4) Use electrodes of CrNI-MccI (18 Cr, 8 Ml) <5) V*cuum>il|ht for many cycles up -438-

2. Whenever possible, singjlc-pass welds should be used. Double pees welds create trapped volumes, and sake impossible the leak detec­ tion.

3. Welds should be made from the vacuus side of the vessel.

4. If for strength reasons double weld is accessary, the inside weld should be the leak tight one. For leak detection, drilled and plugged holes have to be provided on the outside weld.

5. If structural welds are necessary Inside the vesselk they should be made dlscontinous to allow easy flow of gases from any pocket. These structural welds should not cross the sealing ones.

6. The welded assembly should be designed so that a maximum number of velds could be tested separately in the construction stage and corrected prior to salting the final assembly.

The ^T-'TMM'I leak rate (air) permitted in welded seals is about 10~ lusec/cm length of weld. If higher leak rates are found the weld should be ground off to the base metal and a new weld must be made. The supposition that applying another layer of weld over the original (leaking) one, will correct the situation, is wrong* Leaks are eeldam corrected this way, since stresses are likely to be set up, which cause a new crack.

Brazed seals

The joining process of two metal parts with a third one having a lower malting point is known generally as soldering. When the solder has a malting point lower than 400*C the process is known as soft solde­ ring; if the solder melts ablve 500*C the process Is known us hard soldering.

Brazing is defined a. the metal joining process is which molten filler metal is drawn by capillary attraction Into the space between the closely adjacent surfaces of the parts to be joined. The tempera­ tures required for brazing are above 500*C and they should be 50-200 degrees C lower than the malting point of the brazed parts. -439-

Brazing Is carried out by torch, furnace or* induction*

Torch brazing uses flames of oxyacetylen, oxyhydrogen, oxygen- butane, etc A neutral or reducing flame can be used, except for brazing of copper where an oxidizing flame is used to avoid emhrittle- mento Torch brazing needs fluxes which muse be carefully cleaned from the joint, tbe flux remaining on the vacuum side of the seal has a high outgaasing rate,

Fumance brazing consists in heating an. assembly o£ metal parts to be brazed in a furnace with a protective atmosphere (vacuum, neutral gas)„

Inductlor bracing utilizes a high frequency current (400-2000 kc) to heat the parts which ate placed in a specially fitted coil cf size pjai shape that match the assembly to be heated,

Brazing can also be achieved below the melting point of the brazing metal, by using tbe diffusion process- ID this process a thin layer of suitable metal (gold, silver, copper) is placed between the

partsD The layer may be formed by electro-deposition on. one ox both of the members, or may be interposed as a foil of about 0.01 mm thickneBBo The partB are assembled, pressed against each other, and heated to 4QO-700°C (about 15 min,).

The brazing materials for vacuum technology must have low vapour pressure, must be pure, have ability to wet and flow at brazing tempe­ ratures, and alloy with the joined metals. Based on these criteria the llBt of suitable elements to be used in brazing alloys for high vacuum seals is limited,, These brazing alloys are listed in Table 7.8. and the possibilities for using them with various netals are summarised

in Table 709,

In some cases of physical incompatibility, some brazing alloys are excluded from, use with particular metals. So, Kovar (re Hi Co) cannot be brased with silver, because the silver penetrates in the

Kovar and produces its splinterings Silver and gold (and copper) -440-

Tibia 7.8.

BBMINO MITAU AND ALLUYS'

LMdiifl £-.,„ Type of Cofflpotitinn

No. ,.C| SolHu. „„„,„ (p,^ b) ^^

1 jin 3IM p Rhenium 2 29% 29W p Tanutum 3 2497 2497 I? Niobium 4 2427 2427 p Ratlieiitem 5 2444 2444 F fritliuin 6 1965 196S r Rnouium ? 1950 1933 - Rhodium 140), Plutinum (60) > I8K l«52 V Zirconium 9 1770 1770 e Platinum 10 IW5 1645 - Au <3)-Pa(20)-Pl (73) II I5S0 1350 p Palladium 12 1452 1452 p Nickel 13 1423 1423 L NltS6H=0(M> H 1329 1320 e Mo (4S.3J-NI (33,3) 13 1330 1280 M (30J-NI (70) 16 1303 1263 - IM(I3>-A«<67I 17 (300 - Ni (5I>~Mo (49) II 1300 123--0 H Nl (4S)-Cu (Si) 19 1240 1190 Pd (8)-Au 192) 20 1233 I23B B- Nl (40)-l>il (60) 21 1232 1149 Mn (3)-Pd (33>-Aa (64) 22 1203 1150 H- Nl (2S)-Cu (73) 2) 1160 995 Pl(23)-A»(73> 24 1135 I0KO - Fo (3)-SI (IQK* (19)-NI (68) IS 1014 1084 -P Copper (OI'IIQ 26 1083 1083 Nl (3)-Cll(33> W (62) 21 106S 1000 Pd(IO)-A«(«l) 21 1063 1063 V Cold 2< 1060 1000 Al(5)-Cu(9S) 50 1050 1030 L N!(30)-Mo (7(1) 31 1033 1013 11 Au(3D)-Cll(70) 32 1030 975 Co (62)-Au (J5I-NI (3) JJ I02J 970 - Ol (971-31 (3) 34 1025 960 H- CO(93J-AB<3) 35 1020 Fo(33)-NI(J6)-P(ll) 36 lots 101--8 E Ni «U)-Mn «d) 37 loll 990 H Cu (OMli (37) -441-

Table 7.8. (Cont.)

Liqtlldtis Type No. SolJdus Coniposllmn CC) alloy' (parts by weight)

38 10|5 970 - Cu (77)-Au I20)-In 13) 39 IOI0 985 H Cu (60)-Au (40) 1005 996 40 H (3.5)-Si |5)-Cr (16)-N1 (72.5)- Fe (3) 993 976 _ B (.2.9)-Si (4.5>-Ni (9I)-Fe (1.6) 41 975 950 - Cu (50>-Au (50) 42 971 960 - Ag (85) Mil (15) 43 9(i2 -- Ni OJ-Ag (35)-W (62) 44 960 960 P Silver 45 950 950 L An (82) Ni (18)

46 •- U46 Cu (85) Sr, (B)-Ag (7) 47 920 904 Au (58)-Cu (40)-Ag (2) 48 910 779 Cu (60)-Ag 140)

49 900 860 Au (60)- Cu (37)-ln (3)

50 900 WO E C11 (76)-Ti (24) 51 900 Cu(50) Ni(l0)-Mn(40) - Cu(95) P(5) 5; 900 714 Au(75) Cu(20).Ag(5> 53 896 885 Au (80) Cu (20) 54' 889 889 L At) (62) Cul.)2)-Ni (6) 55 885 779 Cu (50) Ag (40)-Mn (10)

56 •~ 80!) - Ni(R9)-f(ll) 57 880 -- Ag(90) Cti 110) 58 870 771 H Au (60) Cu (20)-Ag (21)) 59 84S 835 - Ag(77) C11121)-Ni (2) 60 830 779 Au (50) Agl 10) -Cu 120) 61 821 794 - Ag (72)-Cu 128) 62 779 77') n Cu (93) 117) 63 770 714 Cu (87) 1" (7.5)-Ag (l.J) 64 721 640 Au (80) lu (20) .

65 630 •- 66 636 Ag(5n) Cu It5.5)-Cd 1181-Zn (16.5) 67 625 1563) - Al (95)Si (5) 68 570 1550) - Al(85)-(li<3)-SI(12) 69 550 Ag(60)

"PB pure metal; Fi ••' culectic: L -- lowest inciting pninl alloy: H • alluy having higher I Iqulduj temperature* If marc of t he nwiin mclDl eiiniponenl is milled. -442-

Tabl* 7.».

• - - SuinlCH TI la Nl MM Mf> Konr w •Ml

At - . " " •M43) An ' L ";~T "- i -~~ ~ -~ ~~~ k a -~ - " . ^ I""" "~~ —_ are "'1 "-- '•!'-•' - CrNi 44,21 - -" • ! ._-_. ~'l

54.45 62,45 45*1 54*1.32 44". 50 42,40 32 Cm 32*9 •4*2 39,62" «2~ 49

2S.31.34 22 ' 26 22 4x - - J7*9 37*9

FcNi -- " - -- -. " - lacoad 45 20*1 45 45 45 - 24,42 -

4542 21*9*4 Kuvar 21 45,2>, 26,a, 45(44) J4.II - - 3142*9 62" 62"

3-14,23 Mo 223 19*7 27*2 32 32,45 - - 45 32,45 Mend 29*2 -- 32 32 17,31 32 45*7 Ni 50 27 21,40,41 32 62

15,19 SUInk» _ 19*0 24,30,36 11.24 - 4IVI3 .... _. T» -- »•• 44" m •- .....

w .1-1* 23,45 -**3-

T«W« 7.9. (Cone.)

l*lK«llltlMM3(!* MlTM*"

PcNI At

54,62

25"

25"

29,33,39 31.39.48 34,46,47 54.63" 54.62

39,28.43 44.54

21.24 20.45

* Ntitnriefit refer to alloy number fnim T&Q10 7." *' Short h.f. healing; iillcrrwlive iccliiilquc: hm/c lie

*"' Healing In Ha, lor 3 min nl 920 "C -444-

brazing materials cannot b« used in plants where mercury vapour will b« preaent, except If they are protected by nickel electroplating,

To obtain a leak-tight brazed joint the following points should b* cbaerved:

1. Th* smallest quentites of bracing alloy ahould be usede With —all clearances and claan surfaces bettar joints are obtained than when applying big quantities of brazing alloy• 2. Wide or Irregular spaces between parts are to be avoided* 3. The overlap between the two parts brazed to each other mist be minimum 3 mm, in order to allow the capillary forces to suck-In the brazing alloy. 4. If metals with different thermal expansions are to be brazed, the assembly must be arranged to compress the brazing alloy during the cooling, i.e. the outer part must have the bigger thermal expansion.

5e The flow of the brazing alloy can be controlled by the construction of the joint, The clearance at the corners determines bow the brazing alloy will flow around these corners. Square corners

(Fig0 7.5„a) will give a good flow of the brazing alloy through all

the joint (Fig* 7e50b), the resulting joint being strong and leak

tightB Round corners stop the flow. If the first corner, from the side where the brazing alloy is applied, is round (Fig<> 7.5„c) the brazing alloy will not pass this corner (Fig, 7,5.d)» When only the second corner Is round (Figs 7 05,6?, the joint will be stronger and leak-

tight (Fig. 7,50f), A square edge pressed against a round corner

(Fig, 7.50g) would similarly stop the flow (Fig. 7„5.h). 6„ If the flow of the brazing alloy is to be avoided on a surface, the area must be coated with carbon or chromium. 7. Lap joints, or step joints are to be preferred in vacuum sealing applications (Fig. 7«.6)« -445-

™^Mn .sm^

ft ^j-

(«! r « T

Fig. 7,5, Control of brazing by means o£ the

•^—i

CI xzn P' C! Hi

Fig. 7.6. Poor and reliable designs of brazed Joints, -4A6-

Glaas-to-glass (quarts) seals

Although silica (S102) is the principal constituent of most glasses, the addition of other •citing agents and Modifiers gives to glasses a wide rang* of properties* A general classification divides glasses Into soft and hard ones (Table 7.10) corresponding to the temperature range in which the glass is soft enough to be worked.

Among the many points established to define the state of the glass the two most important are the Strain point and the Annealing point.

The strain point is defined as the teaperature at which the internal stress in the glass is substantially relieved In 4 hours, and "absolutely" relieved in 15 hours. The annealing point is the temperature at which the internal stress is substantially relieved u 5 in 15 min,. The strain point corresponds to a viscosity of 10 ° 13 poises, while the annealing point to 10 poises* Sometimes an 13 3 intermediate point: the transformation point (viscosity 10 ° poises) is given.

Soft glasses have their annealing point between 350-450*C,

while the annealing point of hard glasses is higher than 5Q0*Ct

Glass-to^^lasa sealing techniques are performed by the glass blower using manual or mechanized tools. The description of the operation of glass blowing exceed the scope of this section, never­ theless the following points are enough Important to be mentioned:

1. Two glasses can be sealed together if their thermal expansions do not differ more than about 10 percent. 2. The contact of the hot glass with conducting materials (metals) produces stresses. 3. The free cooling of a glass assembly is generally too fast to form a stress-free joint. The stresses may be released by a suitable

* Details see: A„ Kothi Vacuum Sealing Techniques, Pergamon Press, Oxford, 1966. ** Details: Barr W.E« and Anhorn V.Js Scientific and Industrial glass blowing, Instrumc Publ. Pittsburg, 1949. Ch AS.%S lMu» IN VACUUM TM-IINIOUF

Examples Consil- £«futit« Strain Anniailinjjj t, Tyiw M ou lifaw Pi'int point * * coclT.

SiO ; tM a Mte **$e *£ 413* 1 Jena I'bO (.Will 40-30; jllltaCS Ofkuii i -ciO l wd W2 92 415 1 Moiiabt>

Iron 133 llf " M\ 396 Sovircl" Lull! N(9I5U) ' 100 4Z5- [ Muosbr. 1 cCr I. 14 98 JfiO 430

slUvutu 20 .'5; Coj-ijx'WlutfClz' 31 JWi 433 | 871J utk:ill glasses Mnfl glass 01)10 91 3«J7 42B • Comina - Iff Hull glass 0120 39 ' i«r " 43.1 1 Coming I2.lii M »«"' 425* ! Osram <*:W2j 86 4)0 1 OtiUice l.tfiid MiT)'" fil 4K0* J.-.U m 96 ..; liCrC IV 95" 530 " BTIt Cut) lilllL* 95 ' 41.5 500 (ii-c 5. 12; . silteate >;» nlhtVi JfuJI> ODM) ~»y' 47fi 5I(J Comirig 13-20 M;igHLVia 105 *ny" 5D8" Osram tiWAKiWJ) B7~ 530 Chance

500 SiO.; Iron R L 114 114 _ GEO l.irnc C"2Z f(W . 585 BT» AUtniinn 340; lime A|iparato Olu*

sJJIuiip : »H S8 330* Osram 6-t2; glllMCS ; • alkali- The mitt meter B-23 16'" (!0 _4«5 537 Jena

510,: : Ut0 3 Amber Ms 1 73 • 400 SK0 GEC Alumitid 3.-8; twro AtfO,>-> llnw 3-10; siiicuic CaO^ tfhmtu 6-12; j Mi>, 1) 11, 47 363 594 Hussiin alkali * i 8-23 i — —. ; -448-

labia 7.10. (Cone.)

I- x u m p 1 c •. Coiwtl- Kxpun* lucnti Strain AnncMlini Tyt* sion Manu­ % OlOM point ftnlnt Grou j weigljl coelT. facturer" (10 7"C1

Ashe* ' Alumln_—o - fore but Mo. 1447'" .10 4111 529 Jena boro CaO~ time .1-12; zinc ZnO-3-7 glau alkali - 8 14 H SiO,; Kovar C.AQ AH 455 505 rifii Boro- • B-Oa*. 10 W KC.ll. W 1 .IK 540* 580 one a .Uicite Al.0,-0 DiintnW*} 111 .17 ,."*' .W7' Jena Therm in nclcr r ~ ~ 7520 61 530 566 Comin~ g

<1 Mo f».17h 48~ 550- " Osram NciilrohmftMo) 4B 505~» ' Baccnrui MO,: " FcNICo 756 48 *_." job* bsram f Alkali- ^-10: Mo. 11.11. 47 "500 " 590" oi-c bo ro- A!.Oa 3072«icriillc20) 46 " 55 B* Jena 1 •ilrcate -6; "Clear scnl 7050 4 ft 461" 496 Cumins •>tas«s •ilhnli 500 Mo. C 11 '45 " 575 BTil ft-8 (.Iran .1)2(1 •11 497 5.15 Cnrulng W glass C 9 .16 480" 525 BTH HysilGII 1 '.1.1 S13 556 Ch.1l KM Pyre* 7740 -1,1 515 535 Corning

SIO.; Kovar (iS .1 50 400 430 Chance Alumlno B.6.,r huro- 5-20; Mo. 11 26 X 4ft 725 .".iltatte ALU,™ too ore glasses ?~20; Suprcimx alkalis ft MI — .105M .1.1 - 7.18* Jcnn [.cud SIO,: W scul J62n .19 522» 0*ram horojiH- ».Oj « cato 15-16; Nonex 7720 .16 ..49 .4- SIB Corning gliSM!l_ PIIO:T,4-7

* Transformation point, with viscosity I013-3 poise i •• BTfl -The Hrilish Thomsun-lliiiiston Co. Ltd,, KtiRby, England. Chance —Chance flrothcrs Lid. Class Works, llimiinuhitiit, England. Corning — Coming Glass Worhs, Corning. N.V„ U.S.A. OEC - Osram-O.l.C. Glau Works, 1:11st Lone, Vemhley. Middli-sea, I: SovJrel — Sovircl Co., DaKneaux-sur-Lnrng, France. Oiram - Osram,! Berlin, Weil Gcrnmny. Moasbr, — Moosbmnncr Ofasfibrrk, Vienna 4, Anitriii. Jena - Jeniuer Glatwcrk Scholt a (icn., Muin*, Went Germany. Philfiu - Pfiftfw, Eindhoven, Holland. Baccarat — *j«i¥»WcrIcdeBnccarai. France. -449-

annealing, i.e. heating some degrees above the annealing point, holding it at this temperature for 5-10 min,, and cooling slowly (at l-3"C/min).

Glasses of widely different expansion coefficients can be joined by graded seals. These seals consist of a number of segments of glass, having progressively slightly different expansions; together the segments form a 2one of gradual transition between high and low expan­ sion.

Using graded seals, quarts can be sealed to hard or soft glasses.

Glass-to-metal seals

1o obtain a reliable, leak-tight glass-metal seal, the following requirements should be fulfilled:

1„ To achieve a good bond between the metal surface and the adjacent glass.

26 To base the seal either on the matching of the expansion characteristics of the metal and glass, or on the plasticity of the matalc 3. To control the cooling process in order to minimize the stresses in the seals

4. To choose the geometry (shape) of the seal so as to obtain minimum and not dangerously oriented stresses.

The bond_ in glass metal seals is based either on direct glass to metal adhesion or on an oxide-metal bond. In the direct glass to metal bond the metal surface adheres to the glass without any inter­ mediate layer. This kind of seal can be vacuum-tight but the bond is not strong. The seal is mechanically stronger if between the metal and the glass an oxide layer is formed, containing a graded series of oxide mixtures from the oxide of the metal to those forming the glass. The kind of oxides formed is determined by the composition of the metal

and of the glassp the kind of the atmosphere of the flame and the temperature during the sealing. Platinum can be sealed In glass only without oxides, since it does not form them; thus the platinum glass seal has a metallic appearance and a limited strength. Copper can -450-

give -very adherent sealsj If the oxide is Cu„0 and its thickness is the proper one. the correct colour of copper to glass seals is fron gold- yellow to purple; grey or black colour seal is not recommended. Nickel can adhere to glass with a metal or oxide bond (grey-green colour). Tungsten and molibdenum can give Metallic bonds, but the seals having an oxide layer have to be preferred. The colour of an adequate tungsten to glass seal la from golden yellow to brown if the glass contains sodium or potassium, blue if the glass contains lithium and grey-brown In lead glasses. The molybdenum seals are brows. Ghroaiuai farms an oxide bond and produces very strong seals which are dark green. The FeCr glass seals (Table 7.11) are brown-green,those with FeNiCo grey or blue, and FeNiCr seals are brown.

The metals and alloys used for glass-metal seals are listed in Table 7.11.

The same matched seals refer to those seals where an attespt is made to have partners of equal termal expansion. These seals'can be carried out with pairs of metals and glasses as shown in Tables 7.12 and 7.13.

Each such seal Includes a specific technology of cleaning, oxidi- zing, heating, cooling, etc. From the various sealing techniques we summarize here only that concerning the Kovar-glasa seals, which is one of the moat extensively used techniques in vacuum technology.

FeNiCo alloys (see Table 7.11) should be oxidized either before or during the sealing. To ensure freedom of gas bubbles in the seal it is useful to decarbonize the surface of these alloys by heating the parts in wet hydrogen (about 4 hr at 900*C or 1 hr at 1100*C), or better still in vacuum, since the remaining hydrogen can produce bubbles as well. The oxide layer is produced by heating the Kovar in air for about 1? mln. at 8QQ°C, 3 nin. at 900°C or 1 min. at 1000°C.

Details, see: J.H. Partridge: Glass-to-uetal seals, The 5oc. of Glass Technology, Elofield, Sheffield, 1949. Figure 7.7 shoWB the typical steps in completing a Kovar-glass seal. A sleeve of appropriate glass (Table 7.13) is slipped over the oxidized Kovar part (1 Fig. 7.7) and the glass is fused to build up a ring on the outside of the tube (2). The end of a glass tube is shaped as required and it is sealed to the glass ring (3). An alterna­ tive technique consists in sealing the ring so as to extend to the end of the Kovar tube (A.Fij>. 7.7) or even to cover the rounded edge of the Kovar tube (5 Fig. 7.7). It is recommended to annual the Kovar-glass seal by heating it to about itWC for 20 min. and decreasing the temperature at a rate of l'CVmin down to 450°C, followed by a cooling at 7-10"C/min. to room temperature.

Fig. 7.7. Stages in completing a Kovar-glass seal. During the glass-metal sealing process, the Kovar parts near the seal are always oxidized. They can be- cleaned by immersion (10-60 tnin) in a hot solution (6Q-80°C) of 50 g ferric ammonium sulphate, 3 3 125 cm sulphuric acid (1.84), 150 cm hydrochloric acid (1.16) made up with water to 1000 cm . The glass-metal seals known as "unmatched seals" are based either on the fact that the stresses developed in the glass are minimized by the elastic or plastic deformation of the metal or on the fact that the developed stresses are only compression. The first kind of seals are represented by those known as Housekeeper seals» and the second kind by the compression seals. -«2-

Tabl. 7.11.

METALS AND ALLOYS

Sped- frtpufiitan coeffi­ Compmhloii <%) B lie cient «

r (9.2 45 - 46 46 1 - - i 60 8.1 - 41 40 | - 21 18 S4 8.1 58 53 48

42 S8 8.2 52

42 5K 8.2 47 47 47 (from 0 29 - 17 54 8.S 56 52 I 4» (from 0 °C)

29 17 54 8.3 58 51 47 46

28 18 54 8.3 60 56 51 2" - 17.8 5.,2 8.3 59 57 52 28.7 17.3 (from 10 "Q 29.2 17.8 43-5344-52 45-51 31 19 57 54 53

'42 _ 52 53 63 43 61 60 65 2! 21 6H 65 63

46 74 73 75 28 - 23 80 77 74 48 --- - 52 S3 83 83

42 t - 72 83 101 -453-

Table 7.11. (Cont.)

urn (iiASS-MtTAi,SEALS

E!ectric«l Thermal resiilivity conduct ivity Manufuo (Ohm-cm. (cul/cm.scc. lurcr* 10*)

0.025

0.026

0.042

VM

AA 58 60-100 W 45

cs A -454-

*«M» Ml. ICemt.

|-K[U1uJtM tocfH- denf •-(()'(t/"C) Metal «w AN«y fmm Bt-CiorC) Ci> I ft 100 I 2001 JOB J 400 J 3001 MO*

trnwiOT) SI 1 ** »7 ' 51 79 ' 51 H

! » •J tl 1 21.4 »

» 1.7 »3 « n r.s R4

7J 72- 75 7.5 14 4« M 1.2 101 74 » «4- 77 -

X 7.« S) » » 4* »,» 106 75 10) 97 II

*A Allegheny LwJWtl Sfccf Corp. ilrwfccnrfctge, Pi..UJ.A,; Wt - PhlHp* i:Jnd- AQ.. Himav, Otnrmim CS - CWpemer Steel Corap. Rwdltw, ftt,, U.S,A>j I - Arietta ftyhwife KJectrfe Piod, Inc., fUj-uk-, New York, U.S.A.; Si - Slupikoff Ceramic % wet, KcuicTtriit, Amir!*; II - Slwlilwcrk Hnp:n, Ocrnw>ijri VM - Vacuum MetibGorg; -455-

T.bl« 7.11. (Cont.)

Electrical Titer ma Inflexion reslitivity comlucliv Mnnufuc- Point (°C> (Ohm-cm. (ral/cm. * lurer* -a

- 30 O.OM Curie Point 4>M) *C

423 SO 0.0J7

445 38

" 9.8 0.17 Curk Point 4S0 *C 470 41-47 0.032 M.I*. 1450 *C

43

0.051 M.1\ 1480 "C

0.04 M.I'. 1490 "C

65 0.029 Curie Poinl 571) "C

.140

480 51

60 0.057

63 0.029 Curie Point 640 *C

550 IS

70 SMI --

hovefl. Holland: D - Driver Harris Co. Harrison. N.J.. U.S.A.: V - Vakuumschcnclze (Tlmphy, Niivre. France: W - Henry Wiygin & Co. Lul., Ilirmlnthani, England; S - Ml*. Co. I.alrobe, PJL,U.S.A.;G -GeneralElcclric Co..U.S.A.; M •- Mctallwcrk, Plan- WD--Wilbur W. Driver Co. -456-

T*bl« 7.12.

SIIITOIA.WJH«OMW MIIAI StAis*

C : 02M. 0041, (7550), 7S70, 7560. UOSU, U0XO, 00(0 Ch : GW2 (GW1), I'Wll, TWL BTH :C 12.C 19, C 94 J : 16. 2962 O: JOIh GHC: X4.LI (L 15) P: DIAL-144 K : RS, H(. C: 00)0. 0120, 0080 FeNI BTH:C12 I50/J0 or J : 16 111 46/54) Ch: PWI>, PWL OKC: L 1 C: 0050.1:2962 III Ch : GW2, I'WD, PWL 0:352. XH*. 12.1ii IITH :C 12. (C 19. C94) GlfC: 1- I, K: K 5. R 6 "C : 0050. H0M». OOWt {9019, 9010) FeCr Ch:(.rW2((iW1), I'WU (74/26 or BTJf: C 11. (C 12) 80/20) Cil-C: L 14, X H. (1. 1) () : I23n

C ; 8H70. OOHl), 0(114. 0120. WHO. 0050 Ch:PWl). I'WL 111 II C 12 OHC:L J K :KO 12 C: 0050. 00S0 FcNiCcCr Ch: GW2, PWD <37/30/25/«> K : R 6

1 C:7290. 1990(1991) «F.C:R l6.IKCi20.NSOZ Iron BTH:C76.(C4I) J:42tO C:7295 Copper GEC: CSG 3

Tllanium BTH : C 77, C 78

••See Table 7.11 •*• For (he iltnlfiince of ihe abbreviations tec font notes of Table 7 • 10 -457-

Tabl« 7.13.

HAKOGl.ASftS |(M GLASS MIIALSMLS"

Gltiss*"

C: 3320, 7720, 7780 (7070,5420, 7741,7152. 7750, 7331.7050) Ch: GS I (lniasilV.

C : 71W>, 703a, 705O. 7042. 7510, HO). (7750. 1720. 7.1.11, 7055.7720) Ch:.Un, Mix. 906c, 632a Gl V: Ml. W 26} K: 51-26 l»:(Ki.iltal)

: : 7I1S2. 7040. HHOO. 7520. 7055. 7050. 7750. 7J40, 70WI, 1720 7h : (.S * I1T1I C40 1 : M-17. HM.l, SJ01, «4«2 > 7Sr.lv 'II Hi ilC I CN. SUM 124 C K (.Mi. K 705. 1 N I

J C : IK1S0. IHI5II FcNiCoCr ' Ch I'WI) K : K f.

I <;i:c MM. it 26 x.

, C: 7(H2 ! Hill . C40

I C:7UV>, 7720 J: 14I7Ch:CS4

"Sec Table 7.11 •** h'or Ihc significance ofiho abbreviations see (botanies of Table 7,10 •458-

Housekeapar seals can be made with copper, platlnun, atainlesn steel fend «olybd*nu«. Using on« or the other of these no talc, the seal can be Made In varloua shapes (Pig. 7.8)t (1) wire seal, (2) ribbon seal; (3) feather-edge seal, or (4) disc seal. In principle these seals can be sade wltb any glass having an expansion smaller than the metal part* Copper wire up to d - 0,05 tea (Fig. 7.8) and platinum wires up to d - 0.2 wm can be sealed In glass by this technique* For- copper ribJon* (2.Fig. 7.8) a - 4 tm; b - 0.1 ran Is appropriate. ) >CE^M 'I t h

Fig. 7,8. Housekeeper seals,

Molybdenum ribbons of a * l-3mm; b • 0,01-0,05 mm can be sealed even Into quartz,. Feather-edge senlB (3,Fig. 7,8) can be made vlth copper tubing of d • 10-100 mm, c * 0.07-0*1 mm (measured at a distance of 1 mm from the edge), a • 3-4.5 mm nnd a • 2-3". For platinum tubing n - 1-1,5*, while for stainless steel a » 1* is necessary. Copper discs up to t - 0.4 nai (4.Fig. 7.8) can be sealed at the end of glass tubes,

A compression acal consists of a metal ring (l.Flg, 7.9) surrounding a glass window (2), which may hnve metal rods (3) or pipes (&) sealed through it. In the window seal (Fig. 7,9.a,c) the expansion coefficient of the outside natal ring should be allways greater than that of the glass. In the fid* aeal (b,e) or pipe seal (d,f) the expansion of the metal ring (1) glass (2) and rod (or pipe) (3) must be in such a ratio OB to develop only compression stresses in the glass part, thus Oj - a* > a»» or -439-

Flg, 7.9, Compression sonln.

aj>a2-a_ are possible solutions. In imiltlpl* rod seals (7*9.e) the distance between rods or between the rods and the outer ring must be larger than the diameter of the rods,

Unfortunately the faulta_ occurring in Blass-glass

Ceramic-metal seals Ceramic-metal seals cnti be made =by using glass a* in interned!at* material, or by using the "sintered metal" technique. These t«cbnl

Other techniques aret the hydride processt tht carbide pr«c*aa, activa metal processes, etc. Details) A, Rothi Vacuum Sealing T«ehniqu*a» Parganmo PreM, Oxford, 1966, Table 7.14. Fault* In glass-glass or glaaa-wetal seal*.

Mliurc { Cn u»c

E*««lvc fircv.nrc tHiTctncc on the (INM will* Icollapiinn. bursting! fincL'ssivL- bctiilinp MI* the glnu tuhe Knocking wuh imnl materials Seraluhiue with liiird mnitrititj lrkHii'-i>»ii\ <>l ln.'tL-r»>gwu* train* or bubbles in lite J!l!t« 1 «tV df adhesion (IniuK

t'kLW>ivr iliiTmal |>riidfcnt* lixcevnw Hii'iniiil nhock

f'spiiiuicn ililTurcnlinl Itwlwccn iwn glu* IWJUIVJI pLiw and iniMal) l(iaik<|i»t(r .iiwiraliiiR Iniwruu shui-«.f lhc seal)

WiMl'irni'i'. ( »c\ ill ilK'.rlintt I^L-tiiilytn iirecH 1 lH.-inii.il -Hi,irt (llaini.-. M'fiilrnnt, wuiotirs)

Undinnnn ilmnufse

are uaed extensively In vncuum Healing, with ceramics such a* Steatites, FoHterltea, Alumina or llnrd I'orcalnlna (Table 7.15). The process of atntcnul ctiramlc-mctnl aenle constats (Tabic 7.16) basically In covering tlio ceramic part with a layer of molybdcmrn (or tungsten) powder with a slight addition of manganese (iton or titanium), and in sintering tli« layur at Iiifih temperature. After an eventual

coating with Hi or Cuk the ceramic part can be brazed to the metal part.

?«13* Sewl-pernanent and demountable seala

Vaovm Mils which have co -be opened irom tlne-to-tiM (atral- pennnentj oe often (demountable) can be made by using vaxta or adheatves, ground Joints, liquids or gaskets. -461-

Waxeri seals are used especially In unique or temporary applica­

tions * They can be used to join temporarily aetalt glass, quartz, ceramic (or plastic) parts or to seal temporarily pin-holes or leaky joints,, It is recommended to avoid their use in any long term vacuum

worfc0

Waxes (Table 7,17) are compounds, which when warmed are plastic but become rigid at room temperature, and this effect can be used in vacuum sealing. A good waxed seal is obtained only with clean (degreaBed) surfaces, assembled with a minimum amount of wax applied

at temperatures nigh enough for the particular vax0 Procedures consis­ ting of applying hot wax on colder surfaces, do not give reliable seals.

Adheaiyes are used for sealing pinholes or porous walls (sealing lacquers) or for proper sealing purposes (sealants), table 7,18 summarizes the characteristics of lacquers and sealants (irreversible

actbeslves)Q This table does not contain the epoxy resins which are summarized in Table 7.19,

The sealing lacquers (e„q, Glyptal) painted on the surfaces in order to seal pinholes» are drawn into the orifices and as the volatile

solvent is removed9 the residue plugs the orifice,

the epoxy adfteaiyes are available under various trade names from various suppliers , The choice of the adhesive to be used in a parti­ cular application depends on the shape of the seal, the thickness of the joint and the desired or possible heating (Table 7,19) for curing* Heat curing resins are preferred for sealing similar materials with small differences In their thermal expansion. For Che sealing of materials with very different thermal expansions or for heat sensitive materials it is hatter to use adhesivea cured at roc* temperature.

e.g. Araldite (Clba Ltd., Basle, Switzerland)

Epon (Shell CQEPC Mew York) Gen Epoxy (General Hills, Kankakee, U,S,A,) Torr Seal (Varian, Palo Alto, U.S. A J, -462-

Tabl* 7,15,

CMAMICT USED TOR

Ten. Com­ DCT- Munu •lie press. dln| fatlo- (10 , VO rcr" tlreng th (kl/mm1)

Aff'tnag 24J At (25 7011 -C) 112 7 60 14 Oicqucni.. M SM (20- KOOC) 100 4 85 II R.-salt 7 HI (20 IWXl Cj 90 5 90 15 Fmslcrile 332 HC* rruni 125 85 7 60 IJ UN 31154 (if (20 • 400 ' C> 105 7 60 14 l-ioqucntfie S set' (20 700 Tl III - 70 14

AKiimiKlWi Al 125 700 T) 86 7 63 14 Almnnox 1.18K9 IP (25 - 71W '<•) 7) 3 32 12 SK-alil SM (2(1 1000 <•) 90 5 88 1) Cw.it 11 (20 KHHITI US 5 95 IS .SJT (20 700 O 85 1 83 IJ

Al. 125 700 CI 7S M 9! 28 AP-HS (K5M if (25 - 1000 "("> 79 12 I4D' Stem.* A If' WJ SM (20 • 800 C> 85 15 1711 —

Almamtt *ffr3 (M?HJ M* (25 71WO 7? II III 2-5. AlsiimgM4 (%%) Al (25 • 700 "CI 79 18 280 54

Alsiiiwg 6S2 (9«;n) Al 125 7(H) 'O B0 18 294 44 AD-V9 fw;„j (I1 (25 UHKJ C| 92 220 'Di-giiuli A', 2.1 (99.5%) 11) 1 0 1000 'O 83 20 300 "

AI*iitMtf 475 Al. (25- 71MVO 4( 8 7(1 •13 Aimanojt .(569 II' (25 701)'CI 45 5 44 12 21-4 cr (25 MlOO'n 57 5! Zi f'orccllnn SM tSt-tWOft 52 -J SO II

ihmporcclltu SI* (20- 700-CI 38 3 42 8 Porcelain SIM' (20- 700-C) 39 3 30 8

'•• A'L — American 'Lav* -futix, 'Chaiuinnofa 5, Term., 'U.S.A.; FP -- FfcnchtoMA n - Datasm. tKnrlMfurt a.W- ticrmnny; SM - Sicniil-Mngmsia A.O.. Laiiiypcinltz, "HcrmsiitnT. TlnrrfnfM, 'Qctmuny; SP — SJflnilichcn Porzcllanmaniiraktur, Berlin. Ger- ForihotiitmcmKCittibbrcvlMlioivssct Tabid 7#10 Table 7.15. (Cont.)

SCAIH WITH Gl.AS* *\tl M*7AI.'

MATi'itlNn

temperature • I!"™? [ ( „| . . . ! C /L MI SW r OIOJ*' j Metal'1

tooo M-iii ' ii.mw ti.otv. ' 141.0 ' o.ntio '-•*:. r I'). 10110 ' U. Id. K 4, II ll

(.f < I. I; CnmiriK (HMO, j (HIHI); thrum .Wii, | ' l'Nl (42

Ntl S.Mi..W.|,l;jena:'«4. "'"•• If. Ill; 111 I Ml'; I *lcNiCo)

I Ni; 1-cNi.

o.ii.ir. tunti I

nor) 14-10 i 1.7 o.oi2 Mn-.unI plnvi; Coming j . J5S0 I.I ) 0,007 li.'il. 7050; Git H2h.\;l Molybdenum. 1.100 I ] j 0,(110 MM; HIM CM: j Knvor I ISIIO j .1.0 JL-IIJI 1072 !

Cormnf! 774(1, 772"; Jvnt 2'<0, H.i.lO; «l:(- W I: lllll (40, O; Chrnin .VI.IU, 38*11;

Cimiii. 'fronton, N.J., U.S.A.; CI' - Cnnrs huxcluin Comp,, Ci ildcn, Colnruilii, U.S.A.; Germany: HI [(tmenifiuMsolnlorm Ciinhlf, Jicfly<>lMriKin*M» Germany. II — HttehfK many: HC • Hackney &.Gi.tiil„ KimUiuil; <»C Cietieml Ceramic Carp., K«*$hiy,

•»S« Table 7.11 -464-

T*bl« 7.16.

Tech­ IVwik-r mixture Suipunsiun liquid Mining ! Coaling nique Moly- IM1 g Mo (200 IIHMI) f SO cm] amy lutein lc Dull rnang*. 40 K Mil (150 rm.%h( , 50 cnr* acetone t milting new pArtklc size 3-10 /' I 100 cl»3 pyroxylinc *** hr hinder (Oct I'ont 55J1)

40gMo . II) p nitrocellulose sol in ;imyl:icci(i(c 0.8 g !•<• C 511 cm* haiticr thinner tt ((hmfiyl) ttleolmJ 27 on' flhyJ accliifc (B5-8H"'„) 24 cm3 normal fiirthyf (tcefate (83 92%)

40 B Mo Kin c iiiirotL'Idtfosc sell I l.fi R I c (cnrln'tivl) linm 111 j: nitrocellulose purlitk- size .1 /i

TiMijesicij• p W I0g 1:cpnrliilc iiiiiii(.vllnto,<.'in tlliyliiiL'-j iron si/c I 'I /'

ml sfiL-cilird (sliclliii' 2% l sol in :iIciiluil) M.»ly- 200 g Mo <4IH> mt-jili) i.s cmJ iicclnno III II iiiiinpn- 40 R Mn (400 tncsli) rU'in'mclliylclliylkc- milling J W (! '« (JJ; reducciO lone W cm ciityl elder 100 l»r 2 K sili'it ncid powder 45 cm' nilhiccllulow 2 g -nlcium uxidc lacquer" (MM) lOOOscc) (200 nit sit) Activated I7C. g Mo (2IW m*lir Cull let! Mo-Mr 44 0 Mn (200 mesh) ns nnd 9 g (it-miiim liyiJ'iiir above tired hi two layers 200 g Mo (400 mesh) Hill urn* ncclone as Pointed or vatcd 40 g Mn (400 mcsli) .K>ci«JH«.-lJ>yJ t-lliyj ke­ above sprnyed Mo 10 | l-'c (II, reduced) tone 40 cm" ethyl ether Mn- 2 g ullicic ncid powder 40 cm* nitrocellulose 8 g alumina pnwder l:ia|iier*(fi(H1-1000scc> M me*ri> 20 cm loluune H g IIIiinltirrt hytitide * Nitrocellulose Incqucr: 40 g nitrocellulose, 165 enr1 (criticnc, 73 cm' ctbyj nlcohol. -465-

Table 7.16. (Cont,)

CtnAMic-MHTAL SEALS

Sinlering Hating

130O-14W) "C, in hyd­ Cu, or Ni or CVNi Bond sircnih rogen or dissoci­ tired at I (WO LC ated ammonia 15- 10 mm in reducing. ;:(rrKj-,[ihere

1251) C. 20 min in Sprayed with Ni MIS- . Itiiii M-.IW io Kovai. hyjronvn nitrogen j pension (•! ft [»;ir- ; |il;iUi( unti 1.5 mi! vin t n, hy mici

t-MW C, Ml Flllll *il- Mm, n>ue« nilr.'Livn

M II

1 Wll 1 IdO O/nVM | MM) 1 llM.1 I ( mm;,] |5 in mi.

C .„ n>J...^n Mu-i *..e r. >*

l:or high jluinina cei.iimes

J52VC Ni plaled lo lol.tl Ax A., emeelie. hr lYnu li­ iriickncs of 25 /< ;eil in cupper |im st re ngUi uhoiii

15110 °C .10 n in or Ni phucd 5H ft Good senl lo OPHt 1250 "C 45 rtt n in .slai'Led Cu discs wet hydrogen £l).2< mm thick I by luii/imi with Mi ('it \u iiiloy oi

AB O. vtucctic

60 env* ethyl acetntc. -46*-

TafcU 7.17.

Cmymitia* wmun ( Soh*r*tt ' lObfWfBlwrc PO

}ha»»t tt *»1 > 31«..| 25 Id-* J Acctoae, alcohol. l T«pc«>liW (I fb«) t (ftCtlKltf

I ~ *• I < rt*m (5 pirn) AO 75 J Acetone, alcohol, ' hcfewa* (I pfmi m-*ii hcucite. «hcr.

I j nn*)

: H«N« [I py-J Ike*- • 47 5.10" " ' MJAIIII? of tartan (25 "O ) iclrudtinrhie and . \ alcohol (I; I)

10-• | ChlurnfiMDt, »«- (25 C) i tunc

JJwUW. : iiMcU .mil IIH mill Ml Ml ! l|>hlui, inuiurv nl ! IIOI t'llmiiiitmi, elite r, ! js,Ijh%dro*r J4*.l» ; 1*5, ritii>l-phiii1iiic j amlc-ICtt U«Cll-Irw I »ci. «>lirii in wurit- -pi- | Un>| I »il*

Urd MM- SlielUv, vcflK« iur Ml JW Accionc. ak'nhul. firry MM pCf>l»*> . VC*«liJJi>B nno tviucnc, vhliiro* ' or t'litWC f«it J 23 form, ciher, x>kne

I* Khii. ' SfK(f« and Cfn'thv Accionc. JfcoJw), iifltky twootUifir W-150 hcrwnc, chtorr> winrM , form, titter, xylene

De KtwtHMky type eifiyf afcohof, •«• tone

Shcllic IMW, bawn •*67-

ttblm 7,17. (Cent.)

Attacfcwi | •nun-** I Afeypo- Slgialy tanfcr than plttddt*. Usiium Lews ifct ptsitidly by mid*-" sal. R ID sodium set

~;:,i K (slight Chantcicroik* maj hc'disngcd . by various mixluna colo-

floml adtLxtnn iw told «MMK

Vacuum-light hnncJ Ji>r rtiWhrr* livmcial or-gtm

A hy '. M'Htcraidy iiiujh retln, r>c*intc> pcnline. *>- UNO.' fralrt and sol. tw marc- ICIIC. WilltT, uiul Ut*l« for scaling, R> Iwat (JO rnntt a ill II.SC*, I hr al 90 -C. or ^ hr at 150 XT H to; l*rvt>Q)Ci tanJcr Jur in puly* , lit! ' mcruaiicn |

llciuh nr prone lo . low change*: •

feirofcuiTi.liir. Tough hut wry jlighlty pUnie. i penilnc oil, nriymcrltca n nxwn Icnvpcnt- water. lurt in • imwilh* ,

PofywcrfKi H IMMIMI Cfcinfes I off with dtrtmic »c0, I -46i-

Tabl* 7.17. (Co»e.)

Softening Mix. Vapour Sntrcitli I uft preutirt (In lipto 3 4 hr) dorrj tentperatur* (X)j

1.10-* Bcn/enc. hcnrinc, (90wet- (• 2JQ| cttlorutruiu. eitef,

»mt)W 4.10- ' turpentine. »yle«a 1103 wet- imgi

Sultd high mulcmlar »ei(Jil hjdrocar- I {drops) bom. (ioc imwRank* •WW

White wa­ Sa-rf!ac wtlh tcdm. t«i Petri ileum, heruenc. JH

Aptc/itn (iiaphilc ptaw: »r will n ft pairafin oil dKlil- 125 O 2.10 * [TO «

*irnilar li> Wn< Q

WK* «ra« In stick* 45 I 39 10-• [25 -CI 10- * (wtt) (tmm,

W«k wa* in MkU Wat W-IWJ (mrulum)

HlKk mil In «kl* ] US Wa*W (hunt) welting)

" SheHac tMcd rtww it brittle an* icmtt ta r«m ttalr-Hrw avefct. It If ttwreftr* win* A mfatiinf wfrich after mtKFtff c*ir to (wwed tote meulda flo POOR •tkkt] CQMIM* of: -*69-

I«ble 7.17. (Cant.)

Mtuckcd (A) Non-nolvcnti Resistant IR] Reference* {up to 60 hr) adds alkaline)

acetone, alco­ A by R Suitable for metal. gU»i waling.

hol water H,S04 Vi button rwhunt; nut brittle; Edward*, R to Available in 2 grades; Scpara- ' tcjheld. HO, lion of joint! by heating. COR UNO, (Maxtk P) find chro­ mic) '

r scaling ungrounded joint*

I Adhere* well to gfcm jnd metal [ I.r>hold

• lency or plasticine; Tern- FdS«aruS iling or hanking du- Shell nnc k;ii delJci ilotio n Kin

CT;R

Applimiion whcic il >* required Dt'i in tlow the wa\ in or round I dward* the joint. I or joint-, subjected ' Shell lo vibration, hut tin I heated JfiR

Safe for crick i in joint* tubjec- rdw*nh lett in vibration Shell JOIt

A by High vacuum work where Ihc [TuVarda H,SO« part* tend lo worm un in inc. Shell R in Ilriiitc 10 shock. DCC MCI. JOB HNO,

forced with other materials making fl more readily fumble, m-wt adhesitc and Wronger. afcctlac (50 pbw), wood crcouilc (3 pbw), ttirncnllncn? {2 phw) .ind ammonia U.SS 11 ph») -470-

Tftfal* 7.1*. ImuvtUMU Aotttuvfi Mu. oh Ma­ ura- Oumuriiita tun (oc mm •o I4UIHK of lush After Aryini ln»o- Sc-Sac potjmrn in joi- -20 tubk in oiii, pet­ 10 roleum, witer

Alkyd resin,form ­ Vapour pressure ed by conden­ I0-»lo« (-25 wciiincalso Al, sation of pfiia- -Q, 2x10-" and FfcxihUss; Cl>tHat tic anhydride. (23 "O, IS"* Diytrtf 1 hr Hi lacquer 100 IWO- Soluble room temp, To* in 4cett>ne, xylol tynwrirc. t -2 hr or benzine. Jnso- al I4Q ~C lubk* in mineral oik. alk^i hoi. wa­ ter

1— •— ffWrru- SoftiNe in ,mt'iriy * Hardens: I hr • un w»|. mi*td xxith har- lene chloride. HHfi O. 15 M detfNrr «MXM I methanol, wa­ lir 120 'O ItlittJeirt' <">- le* Itt.uiltihlc 100 (trttarih: snlvcniv Temtlc «re«|ti*i __ »kf/mm><2S*0 Euer of Jiyt J-4 aim before u*e 110 for 2 hr ai •»- IJ0 "C or X if* ! hrnl room tem­ perature. J Aitaero* foJymer of fwljr* Sel« by exclusion AppHed by tomb­ He per* •lycot dimclhi' of oxyaen (va* ing, nn the out- rrytafe twmi *We of pt'msti wild.

fofyvhti) seeiafc Vapout ptmutm After drying the (1 pbw) win- Ixl0"» torr roaletl lutbc* itonlniofuoHfO (23 "O h r»*ted for tiim pbw) OY PVA (n 30min>H50X •eatarrt Ketone 63 -471-

T«bl« 7.18. (Cone J

icro-

ilM • Jtl(C

J Thin liquid Liquid wlubk in I Bvnua mcul, gbu, I ukhluratiylcnc I ccruna* 4-12 [ Scu h, «ciu- hr (23 'Cy. 10 lioa of air. I rain UOQ 'LI | 5 mm (ISO O ;

NainfLii . Naiuut rubbci H rublur pf») '" ben/gl ' Applivtl un tough [ S nr dx.ing. then"

r al room lemp.; fir« fvniw a ral'dcr uJhmid e skin; coniptyictj cired'2-Jhr i tajw)

Epoxy adhesive* were found to have many physical and chemical

, roperticB useful in applications for high vacuum seals t they are very •table and have tolerable outgaasing rates.

In order to Make an epoxy resin seal tha procedure Includes: the nixing, the applying and the curing of the realn. Mixlny is needed since epoxy adhealves are usually supplied as a separate nain and a hardener. These have to be nixed iraedlately before use. The Mixed adhesive nay be applied on the (previously cleaned and dried) surfaces by painting, aprayin| or laawraion, Uaually it ia aufflclent to apply the adhesive on one of th* surfaces to ba bonded together* The adhealve should be applied on both surfaces if they arc very rough or if on* part should be inserted into the other (e,q, telescopic joint). Although curing can take place without pressure, the parte to be joined ahould neverthaleaa be fixed in euch a way that tha thickness of the resulting -472-

Tafcl* 7.13. i ! 1 1 h:S 1 3 1 H III I i 1 ill S i 1 ! 1P s fit : If s I c - « 1 111 iff;:; L*'_ I 1 j i* J p I «' ijl. j-l 22 i i •>« „ i ! B L R i I I • •» i 3R ) S - © £ « 1 . IP"2" - - j ' — — ... ^ | | an | i i | ss i 2 aa 2 1 P s "MI" j"T|si" i T *r~ 2 • 1 i i i > —| S '•"•"j ii 7 "a i*~ _.'J P. fii *'- II I 1 ii_ "I 1 33 : -3 3 i 1 i i 35 *" j ?. 33 1 22 I i 3 rl •iJ3 P 3 5S* ss i SS ! si s B8 hi s "s"~ § S B j s a s T8 T3 9 7a~ ••3 «c 1£ « g a a g 2 ; I 1 1 1 1 3 1 1 -471-

T«bl. 7.19. (Cant.)

m%

• P. 52

si *."3i Is s t I I P. iMJll

£S *8 c HI .8 E* -474-

•AINIV* Uyer ba that given In Table 7,19. Curing la done at the temperature end for the tine recommended for each adhealvt (Table 7.19). If heating la macnaeary thla can be done la an oven, with Infrared lmnpe, Induction heating, etc.

Plsmanthing of an apoxy joint can be done by Immersing the joint (for several days) in trlchlorethylene or In wan dinethyl-formanine. Sometimes the joint can be heated to 130-150*C and at this temperature the components can be pulled apart. Then the adhesive residue left on the surface* May be scraped off after softening by inversion for a few hours in dinethyl-formanine, nitrobenzene, phenol or cresol.

Silver chloride nay be used for vacuum seals that mist sustain higher temperatures than those to which wax or adhesive seal can resist. Silver chloride seale can be used generally up to 300"C, where its vapour pressure is only 10** Torr.

The technique of sealing with silver chloride consists in heating the assembled parts to about 500*C, and melting the sliver chloride by placing it on the hot joint, or by placing it on the cold assembly and heating it* On cooling the silver chloride expands, the design of the Joint nuet allow for this effect. Silver chloride can be used to aeal metals* glass, mica, etc.

id W) Fig, 7,10. Silver chloride window seals. -475-

Silver chloride is extensively used tor sealing of window, which la carried out by using a chin KUI part (Fig, 7.10.a), providing * channel (Fig. 7,10.b) around Che window,in a step (Fig. 7,l0.c) or on a tapered edge (Fig. 7*10.d) of die vails.

Ground glass (quartz) seals or lapped metal seals constat of two parts coanactad to each other on their ground or lapped surfaces. For vacuum-tight saala the surfaces have to be g^reaaed before placing then together. The ground surface may be plane, conical or spherical.

Plane ground joints are used in applications where the joint must be closed and opened without Moving the parts axlally, or where the diameter is too large to use conical or spherical joints. Metal lapped plane seals are used in allde valves of exhaust machines for pumping of electric lamps, electron tubes, etc. Such seals {KIT. 7.11) consist of two discs, one of which is stationary and the other rrtatin^. The holes of the stationary disc (the lower in Fig, 7.11) are cosncc'.r-, to the vessels (lamps, tubes) to be evacuated. When the holes in the two lapped parts correspond to each other, a vacuu»-tight connexion I-; established between vessel and pvmp. The joint is lubricated and •ealed by tha oil circulated In the concentric grooves on the plates (Fig. 7.11.D).

lb) ?ift. 7.U. Lapped seal. -476-

Conlcal ground seals consist of the inner «nd outer parts which fit together, having the same taper. The dimensions of a conical ground joint are expressed by the dlaaetera (d»D) at the small and

large endat and the length -J of the ground cone. The taper defined a* (D-d)/£ is 1/10 in the standard joints, and 1/5 in special joints.

Spherical or ball and socket ground Joints were developed to be used in applications where the alignment of the parts to be joined is difficult or where angular motion of the parts with respect to each other is required. Spherical joints are designated by the ratio between tbe diameter of the ball and the inside diameter of the tubing.

Ground seals can be used in vacuum technology only if they are properly selected., assembled and maintained. If small-bore ground

seals axe necessaryt spherical joints should be used, for medium else (3-100 mm diameter) conical joints may be used, and for large diameters flat joints are useful. For radial openings only flat joints are useful; if a small axial displacement is possible, spherical joints may be used. Tbe opening of standard conical joints require an axial displacement equal at least to the length of the joint.

Ground joints may be assembled in any position. It is recommended that the parts be clamped together even if it docs not appear imperative*

The maintenance of ground joints includes their greasing and cleaning. For the greasing of ground joints the proper grease should be used, corresponding to the range of temperatures in which the seal will work and the vapour pressure requirements of the system (Table 7,20), The grease is usually applied as strips on two diametrically oposcd sides along the taper of tbe inner part. It 1B recommended that grease be applied only on the portion which is an the atmospheric side of the joint (Fig. 7.12). By rotating the members on each other, the joint should become transparent* If after rotation, the surface of the joint presents any lines which do not disappear, cleaning and regressing is required. -477-

TabU 7.20. VACUUM OKBAMS

MM Vupour pressure ' Drupphisl *"' (melting) ITorr) v,ce j point

temp. , C1 at higher <"C> I _ _ temp, I IO-V25T) p dega *>cd 2 hr, Lcybold nl 90 'C

Graise PU I CGR

Ramsay grease I_eybold

only tempora­ Apie/.on L rily in well fitting joinK

Kdwards Apiczon M Shell

Apic/on N for tapered joints

Vncuiim grease 10- -»(23X) K .w ; r>5 Leybeltl i degased Gmixc I'D 2 (.10) ! CC;R

• -! Tor large joints Oniise PB .1 CGK (,10) | heavy Ltitiri.scnl CIMC'O JO ' 40 for ghis.Vmc1ul ViKusUill lip),I cr.NC" 50 Jotin grouse for rolary 1)1) 120 seal) Lcybolil

Vacusenl heavy CliNCO

Cclvaccrtc light CVC

soluble in Cclltxiettl chloroform lilachcr •«8-

Tabl* 7,20. (Cont.)

MM. Vapd jr pressure | Torr} I (meliinu DesignMiofi vice point • * -• "- -J Itemarki Supplier* icmp, RE25T «( higher I CCJ tan p.

U$\a m-p. lUwurtU ApteonT no - g reuse Shell

Crtvaccne medium - 130

Ctlto grease - 120 -clO-« Fischer

UllKtefl !» 210 WW low lithium soap Lcyriolu

SiKcoiK

A grease 200 io-' fO- (l7(rC) OCC Edwards SiHcom: high useful down vacuum grease 200 250 10-i m-*(i7o c) (« - 40 "C

HLg. 7.12. Conical Joint, a) befoie g»ea»ing{ b) tacomeaded (MMlngt f) ncrt tecoaaanted graaiing (dotwd part la greaaad). -479-

The washing-off of R«msay and Apiezon greases may be done with benzine, benzol or carbon tetrachloride. For cleaning of silicone greases see Sec. 7.22.

Liquid seals are joints in which the gap between the connected parts la sealed by a material in the liquid phase. If the liquid seal separates spaces having a pressure difference of an atmosphere, the sealing action may be based oni hydrostatic pressure, on a high Impedance to the flow of the liquid or on the surface tension of the sealing liquid. In order to minimize the height of. the liquid column needed for the seal, in some seals a guard vacuum is used (see Fig. 7.41) or the sealing liquid is frozen.

In seals where the sealing material is permanently liquid, mercury or oil is used. In frozen seals molten indium or Wood's metal can be used.

Oil seals are used especially where the lubrication is needed simultaneously with the sealing action. The most extensively used seal of this type is the cylinder seal in the rotary pumps (Fig. 7,13). For a reliable seal the clearance between rotor and etator nust be 2-3 u. If the two materials of the stator and rotor are not properly chose'fv the wear increases the clearance '-sntil tha oil cannot seal (Fig. 7.13,b).

lo) tbl

Fig, 7.13* Oil seal in rotary pumps. -*w-

a» eeallng action due to eurfaee teneion ia ahova In rig. 7,14, The Liquid which fIlia tha gap between tha, parca beaxa on ana aide tha

aeaoepharic ia given by (v-y (7.U << and *1 R2 •thia glvaa

<7.« i.a. a liquid natal with Y - 500 dyn/cm (Table 7.21) seals T,«l an > 10 dyn/cn . If the gap 1B less than D - 1000/106 - tO"3 cm. table 7.21.

'SlIIWAl-E VeMiuN ANP'OLKAHAMCr. •IfflilHWtt BHAI S suprocc Wiijtimtirotfle aronee'Wfnr Tamp.

Ciulliftni* W 7.IJ 14.7 «2 fin .1110 520 10.4 ;( Mc/wiry » 417 9.J 72 ieia >*> 420 1.4 64 'Blimiilh J00 170 1.4 36

Silver ublmlde KIM 114 2.9 IJ WlWr 21) n 1.4 il Organic MtiuMs 20 25-.10 0.5-0.4 1.1-4.1 -481-

Fig. 7.14. Surface tension seal.

7,34. Gasket seals * Sealing mechanism

Two flanges with a very good surface finish compressed against each other leave between them micron-size channels which constitute the leakage paths through such a seal. These leak paths are determined by the profiLes of the surfaces in contact.

We are used to the image of the surface profile in which the peaks have considerable slopes (Fig. 7,15»a,b) since these profiles are recorded by instruments which have a much larger vertical than horizontal magni­ fication. In order to obtain the image of the profile with the real slopes existing on the surface, the record should be taken at equal vertical and horizontal magnification (Fig. 7,15.d). On lapped surfaces the profile is the same in any direction, while concentrically machined surfaces present very different surface profiles in radial and tangen­ tial (circumferential) directions (Fig. 7,15. e-ti). It was established that on machined surfaces more than 90 percent of the peaks have slopes of i"-4". From these values, the typical, for* of the crass section of the leak paths foroed at the Interface-contact was determined to be a trLangie vita wA* (Fig. 7.16,b). The length of the typical Izak path Is practically equal to the width w of the seal, as It results from the difference between the outside and inside radii (r and r. Fig. 7.16.a) of the sealing annulus.

Details and rareranCM int A. Roth, The interface-contact vacuum sealing processes; J. Vacuus) Sci. lectin, 9_, 1972, Jan.-Febr, -*82-

Fig, 7.15. Surface profiles. -483-

Fig. 7.16, Dimensions ot a seal, a) the interface-contact annulus; b) a typical leak path; c) the surface contact (machined surface; d) single path on the surface; e) loaded Inter­ face-contact ,

According to eq, 3,118, the conductance (molecular flaw) of BTI individual leak path (Fig. 7,16.b) is expressed by

ci • 19-3 [a] (7.3) L Cos aj b

where C. la In liter/sec, A (cm), w (en) and K the correction factor

(Table 3.5). For haliun (M-4) at T - 29S*KV and a - 4* {thus K - 1.7 extrapolated according to Table 3.5 for the A/£ value corresponding to o - 4"), eq. 7.3. Is written A3 1000 • liter/sec (7.4) thtu equation ia plotted In Fig. 7,17 for the range of uaual aeal nldfche v, together with the throughput Q, - C., &P. (eq. 3.26), Since the peak-to-vallay height of the roughnea* of nachlned surfaces la In cha range 10" -10 en, and the Iowaat aanalcivity of haliuei leak detee- —10 3 tori (eat Sect, 7,4) la about 10 ata.cm /a«c, at Af • 1 an the leakage through one alngla leak path can atlll be detected (Fig, 7,17), -484-

fig. 7.17. Conductance and throughput of Individual leak paths (right stale) and of unloaded interface contact (left s«ale).

If the density of the leak paths on the surface 1B iftL, their nurter per unit length (L, cm, Fig. 7.16.a) la

"--IT- ir A machined surface Is usually completely covered by peaks and valleys, thus U/i. - 1 (Fig; 7.16.c). Hence the conductance o£ the unit length of an unloaded interface contact will he (for He, and a - «•) :

f-»Vffi liter/sec,en (7.5) this equation la also plotted on Fig. 7.17, and it show, that th. leak rate of one en length of seal made Just bringing into contact surfaces

having th* best achievable finish tk'ixlQ^cn) 1. .MUy detectable. In order to reach lower laak-ratea, the cross section of the leak path has to be reduced by applying a load onto the contacting -485-

surfaces. This decreases the height A of the paths (Fig. 7.16.c) to

Ax (Fig. 7.16„e). Since in this case ut/h-A /A, the conductance per unit length of seal is given by

By using the integrated result

eq. 7»6. is written '-^STTrTrTT*2^]3 "M which is an equation similar to eq» 3.177.A /A is the factor changing under the effect of the load, according to;

Ax -£R --£Lw-E ,, .. — » e • e (7„9) A where P is the pressure exerted by the tightening force F on the contact area Lw, and R is the sealing factor expressing the sealing ability of the material,, From eqs» 7,6 and 7„9 the sealing process (for He at 25°C) is expressed by

•> T _*» F

(7.10)

or the throughput (leak rate) for any gas is expressed by

Q - C . Ap - 2.4 ||j A2 £ e"3 w5" . Ap lit/sec (7,11) where Ap is the pressure difference of the gaa across the seal.

Equation 7.10 (or 7.11) can be written in the normalized form

C £- 34 A2 e"3* (7.12) by using the concept of the tightening index K-P/R - F/L w.R. -486-

Iquscloa* 7.10; 7.11, 7.12 «n nrmutiil on Fij. 7.18. Tha value of tha saalini faster l la obtained on thie graph u thi I • (/• , I value which produces a decrease of eha conductance by a factor a* « 0.05. The vain* of C j^uhlch results for F • 0, represents tha iaitlal surface rouehneae, according to C ~ » 3* A2.

Fig. 7.18. Plot of Che equation of the mealing proceaa in various ays teas of scales.

Figure 7.18. (aq. 7.10) shows that tha Influence of the various factors on tha aaaling process is i - The curves for seals using hard gaeket aatarials (2 R) will have lower elopes than those for soft asterlals (0.5 S) i.e. for the aaa* decrease of th* leak rate soft aacerlals require leas sealing force (Flga. 7,18, 7.19) -487-

- The slope of the curve Is also Influenced by the width v; the curves of narrow seals <0,5 wj o.l v) have higher slopes, i.e. for the same decrease of the leek rate narrow seals require less sealing force„

- The width w of the seal also Influences the position of the curve; curves of narrow seals (0.1 w) are shifted towards higher C/L values D

- The initial roughness A of the surface determines the position of the curve relative to the C/L scale, If the surface is smoother (0.1 A) the curve is shifted towards lower C/L values; a change in the roughness by one order of magnitude shifts the curve by two orders of magnitude of C/L,

- The straight lines (Fig. 7.18) represent cases in which R and w are constant during the sealing process. If the material hardens and/or the width increases during the sealing process (increasing wR) the curves will be bent towards higher C/L values (Fig. 7,IS).

Figure 7,19. shows a comparison of the sealing factors of various metals. The positions of the curves show the ranges of the initial surface roughness; the roughness (A) scale refers to zero load (P-0). 2 While copper has a sealing factor R from 500 kg/en (soft) to 3300 kg/an2 (very hard), and lead R = 70 kg/ca2; Teflon has 2 2 R - 150 kg/cm and neoprene R • 5-30 kg/cm . Experimentally obtained sealing curves altf&ys include a part which corresponds to eq. 7.10 (or 7.11), but also present various deviations from this equation due to superimposed effects- and imperfec­ tions. These appear both on the tightening curve obtained by increasing the load and on the releasing curve obtained when F is reduced (Fig. 7.20). These two curves together were termed a force cycle. -468-

1 Helium 2S*C u TwokilcrlKci »-«i

w*

«" h «•- #•

wo

POtgton*J

Fig, 7.19. SeaXing curves of various raetals.

Fig* ?,20, Stages of the sealing process. -489-

The tightening curve (Fig. 7-20) nay contain three different sealing stages; accomodation, normal sealing and local sealing. The accommodation stage appears at the beginning of the tightening process where the force ia mainly used to overcome the waviness of the surfaces, to crush foreign particles or to push local protrusions into the softer sealing surface„ This is the stage in which the interface -contact is gradually established along the entire circumference. The accomodation stage extends usually up to a tightening index K * 0,5-0.6.

Upon further tightening, the process corresponds to eq. 7.10; this is the normal sealing stage (Fig. 7.20; 7.21). In this stage the leak paths existing at the interface-contact are gradually throttled by interpenecration or flattening, If the smooth sealing part is the

harder one, the sealing occurs by flattening (Fig* 7„21„a)> while if the smooth sealing surface is softer than the rough one, sealing is a

result of interpenetration (Figa 7,21„b)c In both cases eq, 7.10 is valid, but the value of the sealing factor R is different; sealing factors for flattening are smaller than for interpenetration; Figure

7D21 showB steps of the normal sealing process. Early steps can be represented (Fig, 7,21,0 by real slopes, hut advanced steps (Fig. 7.21„d) must be represented by using an expanded vertical scale (the slopes becoming schematic).

Upon, further increasing of F, the effect of local sealing (Fig. 7„20) may appear. This stage is due to the conductance of local deeper grooves which is initially (line c-d, Fig« 7,20} much less than that of the whole seal (line a-b) and decreases very slowly with the applied force, .Thus the sum of the two conductances appears to be equal to the normal sealing at smaller loads and to the local sealing, when-the force is higher than that corresponding to the intersection of the lines a-b and c-d (Fig, 7,20). Local grooves can be radial scratches, eccen­ trically mated concentric machining marks, or grooves of helical machi­ ning. -490-

Fig, 7,21* The norns! sealing process.

The releailn^ curve (Fig, 7.20) reflects the kind of deformation which took place during the tightening curve. If this deformation was entirely elastic the releasing curve coincides exactly with the tightening curve. If the defonation during tightening was entirely plastic then during tha r«l#a*« tha conductance remains constant, i.e, the releasing curve ie horizontal line, For partly elastic deformation the releasing curve is located between the two curves mentioned. Even when the releasing curva is a horizontal, it continues only up to forces.corres­ ponding to tha Unit between the accommodation and 'the normal sealing stage, where tha conductance begins to rise, since in this region the elastic defoisuttion of the wavineas begins to separte again the parts in contact (Fig. 7.20).

The bottom of the leak paths Is round (Fig. 7,22)t rather than notched (as it was ehenatlcally represented in Fig, 7.16}» When the oposite sealing surface approaches the rounded surface of the pntli, .1 sudden closure of the path occurs. Supposing that tlie curvature begins from a height A (Fig. 7.22}, the conductance Is given by :

fA'l 2* 0.35 \-r-l 1 + .2 .-3* (7.13) 0.012 U±

which is plotted In Fig, 7.22. It can be seen that the tightening index Kappa at which the sudden drop appears In not a function of the kind of material but of the shape of the surface roughness (value of A /A), Since the tightening index depends on the sealing factor it, the force at which the drop appears is different at various materials i even it the shape (A /A) is the same. The value of K at which the conductance drops to zero, results from eq. 7.13 as being (for U=0);

ln (7.IO lira 7X37X7

&>,*»» 10 B llrTU * -i •5 w 2 ^ Si 1 L=t0cm R=500kgta"'

S 11S*

10*

16' »"'

Kf

Id*

KH» 1000 3000 Ktg> *•!•& 0 1M 2» JQaF/Ukg/em)

Fig. 7.22. Sudden drop in conductance due to rounded bottom of the

leak paths

Fig. 7.23. Conductance of seal with increasing width -492-

fuQction which is also plotted on Fig. 7.22. This phenomena bends suddenly ch* sealing curves towards low values of the conductance (Tig. 7.22).

the scaling curve is bent towards higher values of C if the width of the seal increases during scaling. Such phenomena are almost inevitable in seals, since the increase of w Is either due to plastic deformation of the softer sealing, surface » or to elastic straightening of the cylindrical contacting surfaces (Fig.7.23). lite effective contact width is given in this later case by

w' -l( (F/U* (7.15) where f( depends on the elasticity moduli of the two materials in contact, and the radii of curvature of the surfaces. From eqs. T£0 and 7.15 it results that in this case the sealing is described by

« K R (7.16) K (7/L)h which is plotted in Fig. 7.23 in terns of (F/LK as well as of F/L. It can be seen that c' - f(F/L) is a curve, which has a much higher slope than the line of the normal sealing C for the nominal width w, while

Cf (7/h)H « f(F/L)'Ji8 a straight line.

Equation 7.11. shows that in order to obtain a leak rate as small as possible or to use a tightening force as small as possible, the various possibilities used separately or combined arei

- To use gasket materials with a low sealing factor R; this is done practically by using neoprene, Viton OT indium as the gasket material. - To have a seal width w aa small as possible; this is done by using 0-rings (Fig. 7.24)shear seals (Fig.7.37) or knife edges (Fig. 7.38-7,39). - To have surfaces as smooth as possible (low A value); this is done by using polishing, lapping, etc. - To have Ap as low as possible; this is done by using a guard vacuum (Fig. 7.41). -493-

O-rlng seals

An Q-ting seal Is a demountable joint which uses a gaskec with circular cross section, the 0-ring, made of an elastomer or a metal Is compressed between the sealing parta. If the main compression force is exerted axially the seal is known as flange seal; if the force works

radially the connection is a shaft seala When the joint closes, the sealing parts touch the cross section of the 0-ring in two, three or more points (Fig. 7.24), hut in all these cases the contact surface is just a thin band, thus the seal is based on a small width (eq. 7.11, 7,16).

The various shapes shown in Fig. 7.24 are obtained by: 1) placing the 0-ring In a groove machined la one of the sealing parts; 2) closing the 0-ring between the sealing parts and keeping it in place with spacers; 3) compressing the 0-ring in a conical seal; A) keeping the 0-ring on a step and compressing it between the flanges; or 5) compres­ sing the 0-ring between flat flanges„

In groove seals one of the sealing parts (flange, shaft) has a machined recess designed to receive the 0-ring. Basically a groove seal may he designed either for constant deflection of the 0-ring or for constant load on the 0-ring. The constant deflection aealj (known also as seals with limited compression) are designed so that when the required compression ratio is reached the sealing parts meet in a metal- to-metal contact enclosing the 0-ring in the groove, the limited compres­ sion seals are preferred for elastomer gaskets, while the designs with ualinited compression are used with Teflon and neral gaskets.

The cross Beetion of the grooves can be rectangular, trapezoidal or rounded. The dimensions of th.e rectangular groove are determined

(F1B, 7.25) by:

A ,' B - k . tf d2/4 (7.17)

B/d - Kr i 1 i - i r I 1 •fc ! ! 1 *•' A, 1 1 1 r.8r 1

< 1 V 1 ft ? 1 • 1 1 I 1 1 l r

a 1 's 1 l 1

j" u y iy 1 ".H 1 5 I > ^ 1

-9- -6-

Fig. 7.24. O-rlng sea's. •495-

whare k li i factor determining the dead volume, and K 1* the compres­ sion ratio required for the gusbec material. With the usual value of K -0,72 (for rubbar gaskets having a S/jore hardnesa of 40-60) and a dead volume of 5 percent (k-1.05) It results that the diaenBlons of tlie

Fig. 7.25. Rectangular grooves for elastomer 0-rings.

groove are BHJ.72 d; and A-1.15 d (Fig, 7*25.a). The groove should have a radiua of 0,15 d & R < 0,22 d. For better retention of the 0-

rlng when the aeal is not closed (vertical flange)t the width of the groove can ha made equal to the cross sectional diameter of the 0-rinp, (Fig. 7.25.fa), tn this case C-21>0.32 d. The sides of the grooves can be bevelled at 45* (Fig. 7.25.c).

(o) (b)'

Fig. 7,26. 0-ring aeflls-unlinited compression, a) closed groove: b) demountable groove; c) pipe edge seal. -496-

If the rectangular cross section groove is intended to be useu with nets! O-rings (eft indium) the groove should be made with its width equal to the cross sectional diameter of the O-ririg (Fig. 7.26.a) and having a depth 15-30 percent greater than the width. In order to overcome the difficulty in removing the 0-ring auch seals can he cons­ tructed with demountable parts (Fig. 7.26.b,c).

Sliding or rotating shafts (Fig. 7,27) may be sealed with 0--inga. The groove may be aachined on the shaft (Fig. 7,27.a) or in the cylinder bore (Fig. 7.27.b). The dimensions are listed in Table 7.22.

toi ia

Fig. 7.27. 0-ring shaft seals.

Three types of trapezium grooves are used to receive 0-rings: the closed trapezium or dovetail groove (Fig. 7,28.a), the open trape­ zius groove (Fig. 7.28.b) and the trapezium groove with parallel side vails (Fig. 7.2S.c), The dovetail groove is difficult to machine and always hes some trapped volume in the seal, but has a very good retention of the gasket. The dimensions should be (Fig. 7.28.a) C/d-O.75-0.80; A/d* 0.9.

To avoid the difficult machining needed for the dovetail groove the trapezium groove with parallel side (Fig. 7.28.c) can be used. The dimension of this groove are given by: -497-

Table 7.22. Dimensions of 0-rlng shaft seals Cm).

0-rlng cross sectional diameter

B-A C R G 2 nominal minimum h

In max inula m 1.71 2.80 1.92 1.78 0.070 0.13 0.5 0.1 1.85 2.90 2.02 2.55 4.10 2.76 2.62 0.103 0.13 0.5 0.1 2.69 A.25 2.93

3.43 5.60 3.70 3.53 0.139 0.15 0.8 0.2 3.63 5.80 3.90

5.21 8.50 5.50 5i33 0.Z10 0.18 0.8 0.2 5.45 8.70 5.80

6.85 11.2 6.90 7.0 0.275 0.20 0.8 0.25 7.15 ' 11.5 7.25

Fig. 7.28. Trapezium grooves 0.02S (ram)

(/.IH)

Ik

specified for a given type of O-ring. tectstifiulsr flange.? nay lie equipped witli O-rlns gaskets by cutting the groove of one of the cross sections used for clmcular O-rinya (Fig. 7.25, 7.28) but following the rectangular outline of tlie flange. Tti« straight parts of tJjc groove should be connected by circular bends vhosc radius R is not less than that specified Ln Fig. 7.29.

W

Fig* 7.29. MlnlMUM radius R of bend as a function of the cross section disMter d of the O-rlng. -499-

The typical spacer seal is shown in Fig. 7,30. The spacer seals use a retaining ring 3. Fig. 7,30.a which haa a V or rounded V groove machined on its circumference. The retaining ring is closed between identical flanges 1,3 by the action of the external clasping rjng 4 Fig. 7.30.b (made of two half circles)*

£1 •

Fig, 7.30. Spacer seals with retaining ring.

The spacer seal can also be used without specially machined retaining rings (Fig. 7,31), The 0-ring can be enclosed between two concentric cylindrical rings (1,2 Fig. 7.31.a) and the ends of the pipes (3,4), In the specific seal shown* the components are pulled together-by the clamping flanges 5 pulling against the external spring steel retaining rings 6. Such seals can also be based on a single spacer (Fig* 7.31.b).

A pipe 1 (Fig. 7.32) can be jointed to a flange 2 of a vacuum vessel or pipe, by surrounding it with an 0-ring, and compressing the 0-ring towards the pipe and the flange with the conlcally shaped sealing ring 3. The conical surface should form an angle of 45* with the surface of the pipe 1 and the flange 2. In this case the optimum dimension will be A * 1.32 d. If A is smaller a seal with metal-to- metal contact is not possible (Figure 7.33.a); if A Is larger, the compression ratio Is Insufficient for a tight seal (Fig. 7.33.b) and the seal will enclose large trapped volumes. -500-

K m\v/, 1 !

Fig, 7.31. 0-ring aeala with spacers.

' W^'

Fit. 7.32. Principle of conical 0-rlng seals.

7//>.\U-

Flg. 7.33. Conical seals a) A too snail; b) A too large. Tha arrangement shown in Fi

Figure 7,36 showa the Wheeler sealL a conical bakeable seal with metal (copper) O-rlnga. This seal Is constructed for diameter$up to 18 Inches, where A - 21 £ in; B - 20 in and C - 18 in. The seal uses OFHC copper wire gasket which Is captured between the flanges (Fig. 7.34.a). The sealing surfaces of the flanges are at 20* to the horizontal and the wire Is supported by a vertical face. To ensure the location of the

I tS-l J

WIHE

Fig. 7.34. Wheeler seal. gasket against this vertical face, the gasket Is made somewhat under­ sized and la slightly stretched as it is snapped Into place. The sealing force is applied directly over the gasket by means of clamps.

The 0-ring can be placed In the step_ machined in one of the sealing parts and is compressed against the other (Pig, 7,35.a) flange or against a second step machined on the other sealing part (Fig. 7.3J>,b), -502-

(t>) tb)

Fig. 7,35. Seep seals with 0-ringa.

If metal (gold) gaskets are used, and the stepa are constructed with appropriate clearances (Fig, 7.36) very reliable bakeable seals ace obtained * The location of the gasket In the joint should permit

Pig. 7.36, Comer seal. the crushing of the gasttet in the shape shown In pig, 7,36,b, the ratio H/d being Q.5-0,4, Ihe radial clearance should be T - 0.06 A to T - 0,12 d, and the surface finishes of the step about 16 jiln r.n.s^

Assembly and maintenance, of P^ft^jg, sea^g. "R»e main suggestions f*»r the correct handling regard; the 0-rlng, the sealing surfaces and the tljjhtanlnp process.

ffle proper size of 0-ring should be uaed«li/hcn using 0-rings with

ynallerr croBSpBactional diameters than that required, the needed compres­ sion ratio cannot be readied and the seal may leek lassdiately or after -503-

a short time, With an 0-ring having a larger cross-sectional diameter than designed, the seal cannot be brought to metal-metal contact or the 0-ring is sheared in the seal. The resulting seal can be leak- tights but the alignment of the sealing parts is difficult and a larger surface area of the 0-ring is exposed to the evacuated space. It is imposible to use G-rings with larger diameters (circumfe­ rence) than that of the groove, since the rubber is not compressible.

Ihe opposite solution, i.ea the use of 0-rings with a smaller diameter than that required, stretched over the inside diameter of the groove, step or spacer is the most common mistake made in 0-ring seals. Such a solution is equivalent to the use of an 0-ring with a smaller cross- sectional diameter than that provided for the seal but has an. additional danger. Due to the stretching, all the small Irregularities on the surface of the 0-ring are enlarged to real channels which make the seal leaky,, Very often the 0-ring is damaged during the mounting operation itself by the cutting action of the edges over which It is stretched, especially when these edges are not radiused (e.g. end of a glass pipe not fired previously).

To make a tight seal the surface of the 0-ring must be free of dust or any particles which would prevent the direct contact between the 0-ring and the sealing surfaces-

0-rings may be grea:-.:clt The grease may seal for th*. moment small scratches on '.he surface of the flange or of the 0-ring itself, but it cannot be recommended as a sealing material. Greasing is particularly not recommended in seals where the gasket is not enclosed in a groove and may slide laterally. 0-rings cannot be reused if they present any permanent deformation. Ihe seal is restricted to a very small area of the parts* the surface finish of these areas is a basic criterion for a reliable seal- Small machining marks remaining on the sealing surfaces are incomparably less dangerous if they are parallel.to the 0-ring. Radial scratches» -504-

•ven if thay arc very small are dangerous («e Vig, 7»Z0) for the leak rata obtained.

The tightening should be done using the required torques. It la no tint in tightening with heavy tools, seals constructed for hand tightening. One of the «ost laportant facts in assenbllng 0-rlng sealB la to prevent and to avoid drag (rotation) on the 0-ring during tightening. This la done either by keeping die two sealing parts without rotation (bolt seals) during the tightening or by the use of compression washers Interposed between the 0-ting and the (rotating) tightening coupling.

Shear seals

Shear seals or step seals are based on the cheering of the gasket* They aay be designed either with a clearance between the opposite shearing steps (Fig, 7,37.a) or with overlapping steps (Fig, 7.37,b) The gasket used should be 1 mm thick OFtJC copper, hydrogen annealed at 950*C after cutting.

Fig, 7,37, Shear seals. -505-

Knife edf;e seals

The maBt extensively used technique to concentrata the tlt-htviihtr, force onto a small width is to provide n "knife cdce" on onu

Fig. 7.38. Knife edges.

The height of the knife edge is h - 0.7-2 mm, the gasket thickness Is usually T » 1-1.5 mm and the depth of bite b - 0.2-0.4 mm. A widely used application of the knife-edge principle is the Conflat seal manufactured by Varlrni for ultra-high vacuum systems, The sealing ridge 1B made {Fig. 7.39) with a vertical edge (normal to the gasket) end with the second nidR of the ridge inclined at 70' (the angle la not critical). Fig. 7.39, Conflat seal. guard yacuui In the eeals

According co eq, 7.11. the leak race of gasket aeala can be reduced* by reducing the pressure difference across the seal* i*e> using the technique knows as guard vacuus) or differential pumping. This technique consists of providing an enclosed evacuated space Intermediate between the vacuum system (gasket) and the atmosphere. The guard vacuus le generally bult as a double gasket system, but also double chambers are constructed (see Sec. 7.16),

—1,1—

W Fig. 7.40, Single gasket,double gasket and guard vacuum seal {principle). A. gasket seal (Fig. 7,40,a) with a conductance C, or a leak rate Q»C(p -p) sealB against the external pressure (usuplly atmospheric) p * If a pumping speed S ia used, the lowest pressure -which can be reached ia given by

S - P - C

C. (Figs 7(40Db), the limiting pressure obtained with the same pumping speed S will be C . Cl P' - % (7,20) (C + V s + c C l

or with C - c l

c s C7.21) P' - P - 2S + C o 28 H

iBeft always p>p* >p/2„ thus the gain by this double gasket is not of great importance*

If a guard vacuum is used

*"" rrs PI " «-=• • "rs p~ - »•• 1 .-• p (7-22>

Considering po»760 Torr, and pj-10* Torr (easily obtained with a rotary pump), the pressure inBlde the vessel will be

p" - 1„3 x 10'4 p -508-

thue a considerable gain la obtained,

Th« guard vacuum may be used between concentric sualu (Hg* 7.41,

a»b,c,), A puwp-out connexion Is provided In one of tlio flanges, iuivLnt>, one or more connexions to the space between the two gaskets, i'or ume of assembly, die double 0-rlng can be made as a HingJe moulding (Fig. 7.41.d). Seals with guard vacuum can also be made by uulng concentric knife edges, or double edged copper goakets (Ft*;. 7,61,e).

Fig. 7*41* Guard vacuum In gasket: seals.

7.35. Electrical lead-throughs

Xo transmit an electlc current Into a vacuum system or envelope lead-through* should be used. These are electrically insulated from the envelope ( and other leads) and joined to the system by vacuum tight seals. The selection of the materials forming the electrical lead-through and its vacuum seal Is determined by the service require­ ments , i.e. the voltage, current, frequency, temperature, etc. According to the requirements, permanent or demountable electrical leod~throughs nay be used. Permanent lead-throughs are based on glass-metal or ceramic- metal seals (Sec. 7.32), According to their shapes, permanent lead- througha can be: rod oelae, stem seals, pin seals» ribbon seals, disc or cup seals.

In rod seals, metal rods (W, Mo, FeNlCo) are sealed in a glass part having a specially designed shape (Fig. 7,42.a) to avoid the build-up of stresses in the glass. The rod seals are usually used as double (multiple) lead-chroughs on extensions (Fig. 7.42.b), on parts protruding Into the vacuum vessel (Fig. 7.42.c) or crossing the wall of the By stem (Fig. 7.42.d). The lead should carry only current <= which do not heat It sensibly. I.e. Less than Che values in Fig. 7.43,

Fig. 7.42. Rod aealB.

Metal rods sealed in ceramics are used as assemblies connected to metal flanges which are then joined to vacuum vessels by bracing, elastomer or metal gasket seals.

The "ajtems" are lead-throughs obtained by flattening a glass tube over the lead-wires. Such stems used in electric lamps are bused on lead-glasses and copper-clad (or Dumet) wire. The copper- cladded Fe Ni (58/42) wire has an axial expansion which matches that of the lead-glass, while the much higher axial expansion difference ti%* 7.43. Admissible current In wire (rod) seals. CC-copper-elad wire; K-Kovar; V-Vocon; F18-Fe Cr (18 % Cr) ; F30- Fe Cr (30Z Cr).

la compensated by the elasticity of the thin copper layer. The stem lead-through* m*y have various shapes (Fig. 7,44) and number of wires!

A pin seal Is a Blast: disc In which metal pins are sealed per­ pendicular to the faces of the disc. These seals are also celled horizontal sterna due to the position in which they are placed in the electron tubes, constituting their bottom. Pin aeals can he In the fom of a disc (Fig* 7.45.a) or a cup (Fig. 7.4S.b) according to the requirements of the subsequent sealing Into the vacuum device*

Ribbon seals are represented especially by molybdenum ribbons staled in quarts as electrical lead-throughs of quarts leaps.

Electrical lead-tit roughs carrying heavy currents are constructed -311-

t*> m (91 w

Fig. 7.44, Stem seals of various shapes,

Fig, 7.45. Pin seals. so as to separate Che seal from the electrical lead. This la done by Connecting the lead through a petal disc or cap which is sealed (on its circumference) to the glass (orceramic). The electric lead 1B brazed (Fig. 7.46,a) to the two faces of the disc, or it la machined as one piece

{«) (hi (O (d>

fig* 7.46, Lead-throughs with disc seals.

Far various applications, dewountable lead-throufthq ace required. In «uch cases, the electrical lead-through may be either demountable by separating the electrical lead from the insulating part and this part fro* the vacuus vessel! or being able to separate the electrical lead-through as a whole fro» the vacuus) vessel. In both cases the sealing and insulation, or just the sealing Is tione either by wax (re*in) seals (Sec. 7.33) or by gasket seals (Sec. 7.34). Such lead- throughs are supplied by uost of vacuum equipment manufacturers.

7.36, Motion transMisaion

Hie transmission of motion from the outside into the vacuum chamber i» « vary frequent requirement In the various vacuum technologiest cloamg systems of valves, motion of shutters, targets, dicing of crucibles. The techniques ueed to transmit notion into the evacuated •p«e ace: *) the tilting of the vacuum device[ b) the bending of elastic pip*e; C) th« defamation of bellows; d) the deformation of diaphragms; e) the relative motion of ground seals; f) the notion of gasket seals; g) the usa of Magnetic fields; h) the use of heat transfer or electric current,

ttmifced translationel or rotational notion say be transmitted using s shaft (1 Fig. 7.67,a) sealed by a rubber •leave 2, slipped -513-

Fig. 7.47. Transmission of motion by elastic pipes.

over (Fig. 7.47.a) or joined by wax (Fig. 7.47.b) to the parts. Rubber tubing may be used to obtain unlimited rotary notion by using a crank (1 Fig. 7.47.c), the handle of which is enclosed in the rubber tubing 2 in such a way that the rubber flexes on the crank handle. Hie end of the crank rotates freely in the stopcock 3 which seals the end of the rubber. For tilting the shaft, a large diameter tube (1 Fig. 7.47,d) and a flange 2 are joined by a rubber tube 3, supported from the inside by a helical wire spring 4.

Bellows are pipes having their walls bent to form consecutive rings. Tills corrugated shape allows limited axial compression or bending. The bellows are made from rubber. Teflon, metals, and in special cases from glass. Metal bellows allow an axial compression of about 20 percent of their net length, and are extensively used In all-metal valves. Bellows can also be used to transmit tilting motion (Fig. 7.48) or rotary motion (Fig, 7.49).

Rotary motion can also be transmitted fay using diaphragms (Fig. 7.50). The driving shaft 1, in alignment with the driven shaft 2, rotates this later by means of a wobble shaft 3, sealed through the centre of the diaphragm 4. Fig. 7.48. Bellows used for Fig. 7,49, Rotation transmission tilting motion. using bellows.

Fig. 7.50. Rotary notion through a diaphragm. -515-

Ground and lapped seals may be used to transmit motion into vacuum chambers, if Che vapour pressure of the oil or grease used with these seals may be tolerated in the system. The lapped part (1 Fig. 7.51.a) la sealed to the system, while the shaft A la connected to the second lapped part 2, The sealing and lubrication la done by the oil flowing through the circular channel 3. Lapped seals are able to transmit rotation up to 3-4 r.p.nu Conical ground seals {Fig. 7.51-b) or spherical ones (Fig. 7.51.c) are alao frequently used to transmit slow rotation.

(

Sliding and/or rotary motion can be transmitted in principle through shaft scale using one 0-rlng (Fig. 7,52,•) but in current

practice seals using two O-rings are preferred (Fig. 7,52,bsc) since they ensure better centering of the shaft. It la easier to machine the groove on the shaft (Fig. 7.52,a) but when the shaft must have a long stroke the groove should be made in the wall of the port (Pig, -516-

7„52.b,c), A sliding or rotary shaft seal with a double 0-ring can fe# constructed without grooves, by placing a cy-2.iRdri.caO. spacer (Fig* 7.52.d) between the 0-rlngs and compressing tha assembly with a dosing ring.

Fig. 7.52. Sliding and rotary shaft seals.

^* Up aeala known as tfilaon seals are based on the sealing actios of a rubber aheet towards a shaft crossing it through a hole, cut with * diameter considerably smaller than that of the rod. The periphery of the rubber aheet is held tightly by a circular metal part (4.Fig. 7,53), and the rubber elose to the shaft is distorted and bant out from the plane of the sheet, its Up sealing agalnBt the •haft. Xhc sliding shaft does not necessarily have to be particularly ftraight* buc its surface must be smooth. The gaoket is usually 1,4- 1,6 a* thick and aede of a rubber of Shore hardness 50-60. The hole should ba 0.65-O.B of the shaft diameter, thus (Fig. 7.53) d - 0.65 D~

0.8 D; d2 - 3 D - 4 D £or small I) values, and d, » 2 D for large 0,

Wilson seals can bt used with shafts 1.5 m up Co very large

Ca.g'. 70 m) diametersp but at larger sizes (over about 20 mm) it nuat bt anaured that the pressure difference vlll not force Che shaft into tha chancer. -517-

Fig, 7,53, Wilson seal 1) shaft; 2) base place; 3) rubber washer; 4) metal ring; 5) lock nut.

4a) tbl Ul

Fig. 7.54,, Wilson seals with double gasket, a) with parallel gaskets and guard vacuum; b) with opposite gaskets; c) Wilson seal on a glass plate; 1) glass, 2) Neoprene, 3) mica.

Wilson seals are very often constructed using a double gasket the space between the gaskets being evacuated (Fig. 7.5A,a) or filled with grease (Fig. 7.54.b>.

Shaft seals for the transmission of rotary and/or translational motion are manufactured by the various finw constructing vacuus equipment. -518-

Magnetic fields nay be used to transmit notion inside the evacuated space by placing a magnet ( or electromagnet) outside the chamber and transmitting the field through the wall, which should be of non-magnetic Material, The electromagnet nay be placed in the chamber as well and energized from the outside, but this solution presents the need of electrical lead-throughs and degassing or break-down difficulties,,

Vacuum tight, bakeable magnetic feed-throughs for rotary motion (up to 700 rpm) are available from vacuum equipment manufacturers

(eag, Varlan),

7,37= Material transfer into vacuum systems

In every vacuum application, gases (liquids) or solids are to be transferred from the vacuum system to the surrounding space, from outside into the vacuum system or from one part of the system to another. The seals used in connexion with material transfer must allow a port of adequate dimensions to be open during the transfer, while still keeping the rest of the vacuum system vacuum tight. The port itself should be vacuum tight when closed„

For the transfer of gases from one part of the sytem to another (exhaust) it is always desirable to have means of shutting-off various parts: cut-offs can be used when the pressure difference is not too large; stopcocks or valves for larger pressure differences.

When gases are to be introduced Into the vacuum system, stopcocks or valves are used for large throughputs and controlled leaks for small and very small throughputs,

to achieve the transfer of solids (specimens, photo plates, etc) into or from the evacuated chanber vacuum locks are used,

A cut-off is a device in which a liquid surface is used to separate two parts of a vacuum system. A cut-off consists of a system for raising the liquid and a closing system. The raising systenB are those described in connection with the McLeod gauge (Figs, 6.8-6,10) which always includes a cut-off. The closing action may be achieved by the liquid itself or by floats moved on the level of the liquid. The liquid may close the connection between two parts of the system, either by entering into a Y-shaped part of the connection (Fig. 7.55) or by sealing on a porous (sintered) glass surface (Fig. 7.56).

Fig, 7.55. Cut-offs with liquid seal; a) U-shape; b) V-shape; c) with separating wall; d) concentric pipes; e) for variable flow.

Fig. 7.56. Cut-offs with sintered glass; a) with single glass frit (unlimited raising of the liquid); b) with double glass frit (limited raising). -520-

Cut-off* using floats (Fig.7.57) have the advantage that when cloud the vapours of u»e liquid are excluded from the evacuated space. They also permit positive closing.

(a) (*)

Fig. 7.57. Cut-offs with float; a) cylindrical 1) guiding bulge; 2} closing taper; 3) mercury; h) guided mushroom float.

Ths liquid used Is usually mercury, but oil or molten metals {Indium, gallium) are also used,

GIUB. atopepeks consist of a plug and a body. The plug can be polid (Fig-. 7.58.a) or hollow (Fig, 7.58.b). Sometimes hollow plugs are provided with en inside connecting pipe (Fig. 7,58.c). In order to guarantee the leak tightness of the stopcrck* the design with the body cloaed over the plug (Fig, 7.58.b) ia recommended. This arrangement

111 (hi tc) Fig. 7.SB, The parts of the stopcock; 1) solid plug; 2) hollow plug; 3) open shell; A) closed shell; 5,6) handles; 7} outlets -521-

avoida Che atmospheric pressure on the small end of the plug and thus ensures that the plug will be pressed Into the shell, greatly reducing the leak rate of the stopcock.

The main characteristic in defining a stopcock is the nunber pf ways (connexion possibilities) which is not necessarily equal to the number of outlets,, Thus the stopcocks in Fig. 7.59,3-d have one

way and two outlets, those in Fig„ 7059,e-g have two ways and three

outlets; those in Fig, 7059„h-i have three ways and three outlets, and

that in Fig„ 70590j four ways and four outlets*

The usual glass topcocks must be lubricated using vacuum grease

(Table 7820)B The gas evolution from the greased stopcock can be decreased by degassing the grease before use or after being applied onto the stopcock or by maintaining the stopcock at lower temperatures.

In vacuum systems where gases ar-3 handled which dissolve the grease, greaseless stopcocks must be used. These are glass stopcocks

lubricated with graphite and sealed with mercurys or stopcocks having the plug of Teflon,, A valve consists of; the body„ the bonnet and the stem (Fig. 7,60) The function of a valve is to adjust or stop the flow and is achieved

by the closing system0 This svstem needs motion inside the valve (ope­ ration system) and the moving parts should be sealed (sealing system) . From the conmbination of these systems (Fig» 7,60) a great number of valve designs is possible , the number of types constructed being really very, large„ The most frequently used are; the diaphragm valves.

plate valves and gate valvesB using elastomer or metal gaskets, bellows seal£ and mechanical or magnetic operation.

In diaphragm valves a rubber mambrane (1 Fig. 7,61) is forced against the seat 2„ by the external shaft 3 operated from the head 4, The diaphragm functions here as both the closing and the sealing system

For a discussion of all the types shown in Fig. 7,60, see A„ Roth;

Vacuum Sealing TechniquesB Pergamon PresB0 Oxford, 1966, -522-

t.l u> lei «>

It) 111 ID IK)

Jlll (i>

ri|. 7.59. ShapM of glui stopcoclu.

Hf. 7.60. CUnifintion of valvu. -523-

Flg, 7,61. Diaphragm valve. of Che valve. These valves have a good conductance but the large area of exposed elastoner leads to outgaasing properties Which make the valve useful only in backing vacuun lines.

If the valve is to be used on the high vacuvn side it nuac be of high conductance (large opening) and the amount of elastomer exposed aust be kept to a minimum to reduce the gas load from this source. This is done by using one of the various gasket seals (Fig. 7.62). The gasket may be of Neoprene, Viton, Teflon or a metal, and stay havs rectangular, trapezium, or circular cross sections. It may be placed on the closing disc or on the seat. The sealing aystea of the valve may be an O-ring shaft seal, a Wilson aeal or metal bellows. Bellows •••lad valves havs leak rates leas than 1x10" luaec, i.e. by a factor of 100 lass thsn 0-ring (or Wilson) sealed onss. Bellows seals are absolutely necessary In bake able (all-metal) valves.

All-metal valves era baaed on dosing ayetame consisting of edge seals. In these constructions a silver part la compressed against

a harder Csional) part (Fig. 7.63«a)t an Al ting la closed In a conical joint (Fig. 7.63,b) or a copper poppet (Fig, 7.63.c) la pressed Into a atalnleaa atall cutter aeat. Cross sections of bakaable ultra-high vacuus valves are shown In Figs. 7.64. and 7.65. The valve In Fig. 7.64 used a bellows which la aealed at the upper part by a gold gasket acal. The closing ays tea is baaed on a copper nose againat a stainless steel aaat. -524-

Pig. 7,62. Closing syeeemfl of vacuum valve* using gasket eeala, a> plate valve; b) flap valve; e) plug valve; d> cone valve; <) gate valve; f) plunger valve; g) butterfly valve; h) ball valve.

fig. 7.63. Closing ayateas of all-aatal valval. -525-

Tha valve shown in Fig. 7.63 uaao a stainless stsel knife edge •••1 against * copper disc. Tha various valval should fit the kind of process achieved In cli

Fig* 7,64, Ultra-high vacuum valve. A) copper nose; Bl bellows; C) guide; 0) ball bearing; E) gold wire seel; r) driving eachanism.

ft mmtn

Fig. 7.65. Ultra-high vacuum valve. -526-

are known *» isolation valves. Valvea that have alao othar purpoaca or chat afaould correspond to special requirements ara known aai seal- off valves, throttling valvaa; air admittance valves; baffle valves. Seal-off valves ara uaad to cloae an evacuated vessel, the sealing system being taken away, leaving the veaael sealed by the dosing system of the valve. Throttling valves are used to adjust the rate of flow. Baffle valves are so designed that the disc of the valve remains in line with the port, thus acting as a baffle. These later valves are installed above the diffusion pimps(Fig. 3.35).

Controlled leaks are used to introduce metered quantities of gases in the evacuated vacuum systems. The basic features used to construct such leaks are listed in Table 7,23, the throughput ranges being those mentioned in Table 7.24.

To achieve the transfer of solids into or £rom the vacuum chamber, vacuum locks are used. These consist of a space (chamber) ad J cent to th* vacmai system, and which can be opened either to the atomosphere or to thm vacuum system. The lock ia first opened to the atmosphere (being sealed towards the va- *um chamber) and the object Is Introduced into the lock* The port to the atmosphere is closed and the vacuum lock is evacuated (through its own pumping line). The port to the vacuum chamber is opened and the object is transferred from the lock into the vacuum chamber, and the port to the vacuum chamber is closed. Vacuum locks may ba based on sliding rods, rotating plugs or chambers with double port.

Vacuum locks using sliding rods (Pig, 7,66,a) consist of a rod 1, which may ba pushed into the vacuum chamber 2 through a aeal 3. The rod carrlca the objects to be transferred in a pit Cor pits) 4 of appropriate shape. A more sophisticated construction (Fig. 7.66.b) uses a long rod made of three sections; the two end sections 1, are carriages and the middle section 2, it a blank rod. The sample is mounted inside a carriage, and the rod is moved through the channel 3. Table 7.23. G«< Ittkm

Bajtc feature of the I l teak Chit JC iciiitlo

Pinhole through thin soiled gin*, tjuaru of siajul I

Oni'itc | at UK end uf pipe*

j in jfcm waJb; variable by iwbjjjij iftc tubs Crack or change of ircrCury Icrrti

Capillary j cuiiiuni tak

Flattened lutp | variable t»y bending, iwisiing, sque«injs

! ceramic, ;/as* fn( or ntetaJ; xariabfe by t^ian- I Porous plug k ging ihc mercury tcttl j

:otin.-pinc pipe*. ci>ne anJ seal, wa-Jicr*. j Annular Impedance O nopt ;

N«(f(c valve

lnn^ttwlmat cipamton hum ml Tum|K*rature IK> I radi-il c*pan*ion I •emi'crautre or (he gas

' for liclmm, liyJrojcn. and 0*5jen

I hubUlinp. I vibr.uctl cup rolling pil* titir.ilcd nccvllc -528-

TabJ* 7.24. Throughput rugc* of gw leak*.

Lowes Throughput rant* pres­ Uik Remarks (luicc) sure wrr)

Orifice tfwpwflfcpla* liMim dfcc 6* 10 '-30 S-tS ft diameter

1.2|i bone taper Orifice on ihr end 39 **/cm of tapcrci capil­ lary 24 /< bore ttper 1.6 30 (i/cm

Crack on capillary varied by (wilting 2xtO-*-IJ the (obe

Slii oa fJas» luhe 0-y».iW io-» varied by mercury level

Circular era** *ec- 0.2 to 10 i> di». Imn Ciptlla'y io-*--io-* - 1 cm ion* see

Fix 11 etmt copper varied r»y rolling Ittt*.' Jy IO-»-) the lube we -•

PonxM nxu) plus S*IO-*-l<>-» Compfcvtcd *t viri- out fiudt

forow* nu(enal viricd fay the kvel of thf mercury

Porou* ceramic rod lxlO-*-IO |0-'- 10-•

Permeation (hrtrufh KJItcofic rub>wr ihec! -: IxlO-1 - Tabic 7.24. (Cont.)

1Low«M

i (tlRCCl »W ! " __ 'JlWTl 1_

O-ring irth with j | vaiiahto tomprc-*- J j j sior* | ;- 1> 10"*

Conical, ipiing v.a»hcr*cal ••(0-*-?xt»~a

itHIvtcotstl sew. Sicel hall uri seal 7

Ndrtifc vaivo. t .-10-'

Pin valve _. *•» H bcllo*! *sale»t

Needle **!** » 10-

H-W.bte j)).n»t») valve » T * 10 "

PUtinom wire «<- (uniting in ybv. * • 10-'"25

luojtMen rod e*-

Its* «cd twdy ( • tO-'- 9.10-

thro(t*hpa( vtiwtf j «. » lt$Ull of | ffnlsd t»pill*r> ! 4--10 -530-

Fig. 7.66. Vacuus locks using « sliding rod, a) with simple pumping; b) with cascade pumping.

three pumping lines (P., F„f P.) assure the pressure gradient (guard vacuum) from Che atmosphere to the VACUUM chamber 4. At the two ends of the rod, double 0-rlng seals 5, were provided.

In vacuum .locks using plugs (Fig, 7,67) th* plug receives the object from one side, and by its rotation allows the transfer of the object to the other side. In the technique shown in Fig. 7.67,a the object is pushed into the plug 4 through the opening 1, By the rotation of the plug Its opening reaches the evacuation outlet 2, and by further rotation the object is transferred to the chamber 3, In an alternative construction (Fig. 7.67»b) the plug 1, may be withdrawn free* the shell, since a second rotating shell 3 is Included In the lock. After inserting the object into the pit 2, the plug is placed egaln in the shell, and by rotation the pit la brought to the evacuation outlet U and then to chamber outlet S.

The principle of a double port chamber us-d at a vacuus lock In electron microscopes Is shown in Fig, 7,68, Here the probe 1, Is introduced In the chsmbsr 4 between the ports P- and P*, and then

by opening the port P,f into the chsmher 2, to) <&)

Fig. 7.67. Vacuum locks, usinj- pluj.-s. a) pluR uitli pit; h) plu£ with pit and doublt shell.

(si ib) (c,

Fig. 7.68. Vacuum lock with double port, a) the object In Its bolder; b) the object tn tlio closed vacuum lock] c) tbe object transferred Into tlie vacuum chamber*

7.A, leak detectIon

7.41. Leak rate and detection

An ideal vacuum chamber should ranintnln forever the vacuum (pressure) reached «t the moment of Its separation from the nuiaps. Any real chamber proeents n rise in pressure after being isolntcd from the pumping ayatem, 'Ihc pressure rlac Is produced by the Ran which penetrates through lenU, that which evolve* fro« the walls (ottteaaalsg} and that entering by permeation. Toe leak rata Cor real leak rate) is the quantity of gas (In g.V, Qftita^ flowing par unit timm into the ayateau Obviously a perfectly tight vacuus eyateai or chaaber la one having a zero real leak ran, but to achlrve this ie aa impossible aa it' la to reach aero preaaure. The leak rate ia expreased in throughput uaita (Torr. lit/«*c, lueec, etc) Indirectly the leak rata of a given ayste* or chaster ia sometimes expressed aa the pressure rise in a given time, and Cor a specific voiuae^or as the time required for a given quantify of gas to flow into the systeat. Table 7.25 =OMpares these specifications. labia 7*25. Leak rate specification*.

Trmrfor Lwfcniie. I micron iimc) in f litre iem^STP Equivalent Qp&tiat • vektrae t'HcflHlK gauinltow

' (-'- . I/*/** HeatejuJir slit with t cm width, 0,1 mm height itid I «n depth . »-' . & piitAi* RecfantuJur alit with J cm width, 30 » height and J cm depth |0«i 36/i/hr 1.66 min Zlhr Cipillaiy 1 cm lone«m) 7 /i dii W-» l.tff/hr lb£ft\\tt 1.7 days Capijtary I cm long 4 /i dlt

-~4 - t0 l.fl,./ttay 2.77 lir 17 days Owllliry I cm long 1.S ft dla

iO-t OM jiftoy 27.7 hr 1.4 yr ^ CaplUiry I cm low ttj fi dli It"» . 31 Wyr ll.6d*ya 24 yr i Cipllliry 1 cm long 0.4 f* dii lo-» Si'tyt II6(tnyD 240yr Ciplllwy 1 cm long 0.2/tdm -533-

ID approaching the problem of a supposed leak greater than the admissible value. It ia necessary to determine first If such a leak actually exists. Plotting of pressure vs. tine curves (Fig. 7.69) will assist in determining the actual leak rate. First the system Is evacuated to a atable minimum pressure. To Minimize the effect of tht vapours present ID the system It is useful to employ liquid nitrogen traps, When no further inprovement In the pressure is evident during further punplng, the pumps.are valved off from the system. The recorded pressure rise vs. tine curvfc during a long period of Isolation shows IE the rise of pressure is produced only by real leaks (straight line 1 Pig, 7,69), only by outgaseing (curve tends to a limiting maximum value as 2 Fig, 7.69) or by a combined effect of these two (curve 3 Pig. 7.69). The recording should be continued until the shape of the curve becomes evident.

Fig. 7.69. Pressure-time curves.

In order Co obtain higher sensitivities and to locate the exact position of the leaks a series of leak testing methods were developed, using various test |ases (or liquids). In principle in all these beats the gaa (or liquid) is spread over the outside of the part and •the presence of the gas In tested inside, or the vacuum system is filled with the test gas and its presence detected outside, table 7.26. lists noat of these methods. LEAK DCTTCTMN Mrntooa

Minlnnun d*. Method Test •"• Test principle FUttUW rant*. UcttbktittMk iMMTlt* (IDMC) i

Flame wtwring, | *tr. nitrogen i The su ttreim is hetrd by its Using or U seto­ upto3*tan 40 SftBtbtlimer ff thp watwntf. of * ftume unlet room re*

EleetricdachaigeJ acetone, tnetha- Ctrfour chance of the discharge 10 ! Ml. CO.. -

Wet OUUHfc prassyriad The inside fs HHed wii6 liquid; observe the points (Involves weiiinj surface liquid ' •vb.eac outside will be wet i the inside of the ' vtttdudfub- tequcnt ckaaiiv

*Sr, aitrojen ' The leak is indicated by a bubble appearing when up to 3 attn j the «u—pressurized inside—on pass to the " AUiside

Bubbles on to*p! air, nitrogen upto3ata» . CWloxto-*) *Hiw-,pfaminatn- Mlaed'Sini

Ammonia fume* j ammonia, CO,, ] Ammonia inside; delected outside wJuVCOt or op to 3 »tra i 0.04

- SO, HO or CO! nap, SO( inside; detected cvtiide | with ammonia Ammonia sen­ Ammonia inside; Whtrt leaks to out tide produces sitive paper I black spots on wet ammonia sensitive paper rolled over tbc parts

Single Ft'nni Trie-test gas changes the thermal conduction i gauge side the jaupe 1 tolXlO"' torr

A pair of gauges, one sensitive to both >ir ind test gas-the other connected via itrapiiKDii- 1 to 10"* torr (ive only to air

Charcoal Piranj hydrogen Cooled charcoal trap is inserted in the gauge line 1 to 10-< torr to reduce the effect of pressure fluctuations j

Sfntfc Ionization i ***<>*» Test fas changes gauge reading **uge ' CO,

PIlTcrcniial . C"0* l ionization ! Pair of ionization gauge* arranged as in- din*. yaupa butani Pirani test

Ihe gauge tPirani, ionization) separatee from vi- I at by Pirani or i 1 Care not to poison I cuum system by a Pd barrier, which when hoi li J lonliation • I llie gauge by I permeable to Hi only | impurities -536-

Xabl« 7.26. (Coot.) 111ifffllllf JK.a'S l

1 Iff".,

! £ I iji

•B!.li ;<§£ J{. M.i |81 -537-

7,42, Leakage measurement

The leakage measurement gives the value of the total leakage,

of the whole system being measured. For this purpose the test gas should enclose the whole contact surface and the detector should measure the test gas concentration on the opposite side.

Leakage measurement techniques fall into two major classes: static and dynamic.

In static testing the device to be tested is pressurized with the test gas and placed in the test chamber. The concentration of test gas in this chamber is then monitored as a function of time; since the concentration increases during the testing period, this method is also called "accumulation testing". The measurement is carried out and interpreted according to

Q - V g- (7.23) where Q is the leak rate, V the volume of the collecting system, P the test gas pressure in the collectlng^volume, and t Che accumulation time.

In dynamic testing, the system or envelope being tested is continously evacuated. The test gas flowing into the pump passes through a detector section where its concentration is measured. The leakage measurement can be done (Fig» 7,70) by continously pressurising the system to be tested end placing it in an envelope connected to the leak detector or by introducing the tracer gas into the envelope and connecting the system to the detector {Fig. 7.70).

The magnitude of the response In the dynamic method Is' dependent on the sensitivity of the detector to the test gaa used and the speed at which the gas Is removed by the pump. Thus the leak rate Q is given by

Q-S.K0a-S.P (7.24)

For details refer to: J.W„ Marrj Leakage Testing Handbook, Rep. NASA CR-952/1968, -5W-

:•• ••_._-• ::_• ;. =1 leO©) r U¥ELOm - DCtEHH films'.'*

ta pBtssimaeo wren none

&**

MEISUMIKO twctow mm

Fig, 7.70» Dynamic leakage iseasurcment modes.

where S is the pumping speed, F the pressuret £_£ is the pressure reeding* K is the amplification factor of the detector (gauge) end a_che detector current. When test gee it introduced into * systen through leefce at rate Q, the rete of pressure build-up is

' dp o (7.25) dt" V where V is the voliwe of the -tystea* However if eieulteneously the test gee is contlnoualy pumped out, the balance Isi , Q - PS <7,26) dt By integrating ( and for Che cose of teat gas pressure equal to zero Initially) i

*-! a. (7.27)

V , ,, PS v • j In (1 - ^ ) (7.26) At long exposuretlme, the exponential term In eq. 7.27. approaches -zero, and the aquation reduces the equilibrium value given by eq. 7.24, The time required to reach equilibrium la shown in Fig. 7.71 for three examples. In case of curve 1 (V-10 C> S- 1 £/aac; V/S- 10 ace) the detector requires iO second* to reach 1-e" -0,63 of the ultimate signal, and the equilibrium signal is attained in approximately oca minute. In case 2 (V- 100 l\ S-l l/sec; V/S-100 sec), in 10 seconds the signal • -0 1 reaches (1-e * 0.095) only about 10 percent of the ultimate value, and about 10 minutes era required to reach the ultimate signal.

INITIATE I cusi/E • - 1 t / -" CI«V f 1. JrafUl V0UMC *»litm / y CURVE e AVC /srsrw VOUME • IOO »m \ " / RMWUtt 9KC0 • 1 llto'Mc MVC 3. SnTEM VQUK • I0O blMi / \ ° W* (pwGSPcn • JWHt/i t ;.. \ , \\ 1 \ / [ \ / ,.., CINIVt 3 \ \' N *%. i "*»^. —.^. vv no too 300 400'

Sig. 7.71, System, response time. By Increasing the pumping speed as in case 3 (V-100 4; S»5 ft/sec; V/S-20 sec) the time to reach the ultimate signal la about the sane as In case 1, but the value of this signal is lowered to 20 percent of its value in case 1,

Mien removing the test gas from the systemt

OF PS (7.29) " V

p t - »1 " ( (7.30)

? "*« t (7.31) 'o * P and F being the pressures at time t and time zero.

Equation 7.31. is plotted in the right side of Fig. 7.71, for the three cases considered. It may be seen that in cases 1 and 3 the clean-up Is done rapidly, while in case 2 the pump-out takes considerable time.

The response time and the clean up tine are determined by the time constant of the sytem (X • V/S), thus the detector response and detector cleaning can he expressed as values valid for any system (Table 7.27).

The response time Is usually understood as the tins required to reach 1-e" -0,63 of the equilibrium signal, while the clean-up time is the time required to reduce (by pumping) the signal to e" "0,37 of its initial value. To increase the sensitivity it is often suggested that the leak

detector he connected between the diffusion punp and the fore pumpr thus "on the forepustp side" Instead of "on high vacuum aide" (Fig* 7,72). table 7.27. Detector response.

DETECTOt RESPONSE TIKES

friction of Fraction of Tl*» Conaltst Ultiuto SLEUI SttrtlnK Signal 0.001 O.O009 0.999 0.002 0 0019 0.998 0.003 0029 D.99T 0,004 a 0039 0.996 0.004 0 0049 0.B35 o.ooa 0 0039 0.894 0,007 0 O0d9 0.993 0.001 0 0079 0.991 0.01 0 O099 O.990 0.03 0 0168 U.9M0 0.03 0 0-970 0.D4 0 0392 0.901 o.oa 0 04 SB 0.951 O.Ofl 0 0383 0.942 0.07 0 0976 0.831 0.01 0 0769 G.B23, 0.08 o 0661 0.B14 0.1 0 093 o.aoa 0 181 D.819 o.0.a3 0 0.741 0.4 0 as329s 0.670 o.a 393 o.eor o.a 0 4M 0.349 0.7 a 503 0.496 0.1 0 550 0.449 0,9 0 393 0.407 I o 0321 O.M7» 3 0 8647 0.13&3 3 o 9302 0.0498 4 9816 0.0113 fi 0 9932 0.0087 t 0 99) & O.OOIS 1 0 9991 0.009* 1 0 9997 0.0003 0 9999 0.0001 . 1•0 1 0000

The gain In aanaitlvlty depends on Che factor which Halts the ultimate sensitivity of the teat. In a clean aye tea, the ultimate sensitivity aay be Halted by the partial preaaure sensitivity of the detector* la thle event, the pressure amplification obtained on the forepreaaure aide will result in a sensitivity gain. In a contaminated aystea or when searching for extremely email leaks in the presence of large leaks, the sensitivity is frequently limited by the resolution of the detector in distinguishing test gee partial pressure from the background. In this event, forepresaure leak -542-

Fl|i 7.72, Alternate sites for leek detector. detection results In amplification of both test signal «nd background, and linlass selective pumping nesns are employed t«*S> Pd barrier for

i: the leak detector is connected on tht forepuap side of the diffusion pump, the forepump removes gas from the system by a batch process. The pressure of its inlet side tends to fluctuate, and these fluctuations would contribute to the noise level of the detecting elevent which Is directly connected with the mechanical pump, By placing the detecting element between two diffusion pumpi, this effect is greatly reduced* An alternative approach' Is to piece e baileet tank (Fig. 7,72) or e throttling valve between the pumps and the detector. -543-

7.43. leak location

Theara ara two dlaclnct technlquea of locating leaka: the use of a tracer probe and the use of a detector probe (Fig. 7.73). ID the tracer probe technique a stream of teat gas la spread cm the suspected area, and the gas penetrating Into the ayatem Is pumped via the detector* In the detector probe technique the teat gas Is filled Into the aysten under test and a detector probe (sniffer) connected to the leak detector Is passed over the suspected area to receive the test gas escaping through the leaks.

frig, 7,73. Leak location techniques. -544-

The tracer probe technique is usually more sensitive than the detector probe technique, nevertheless this later should be used If a) the systea has to be tested la a pressurized condition, b) the test gas Is one which may be readily be absorbed on the leak surfaces; c) the detector has a sensing element that.nay be operated under «tMD«pberlc pressure, The tracer probe technique should be used if a) the detector sensing head has to be evacuated for use; b) leak location is performed after dyoaadc leakage measurement.

7.44. Sealed unit testing

Sealed units are tested by back-pressurizing. This technique consists of thtee stages: a) The application to the external surface of the test speclnen of test gas at a high pressure; the gas is flowing in through the leaks b) The period between the release of the external test gas pressure and the leak test- c) The leak test.

If it is considered that the flow through the leaks is molecular, the leak rate described by

C t A E' 1 - i (7.32) where Q is the measured gas leakage, C. is the conductance of the leaks for the test gas; P is the test gas pressurizing pressure; t_, the pressurizing time; V internal free volume of the system; t_ residence tine at one ataosphere after pressurizing.

Figure 7,74, shows a plot of Q vs« C, for typical values of P-,

V. t£ and tR.

For very small leaks (less than 10~ Torr. lit/secjit is possible to reduce eq, 7.32. to a simple form; COHDUCriMCE, CA , TOff* t-/<

Fig. 7*74, Back~pre5surlzing-computed values.

i J-.*-.— • *c (7.33)

The smallest detectable leak C, depends upon the pinimmn leak rate Q, which gives a signal on the detector appreciably above the

background* Equation 7*33, defines the value o£ the product P„ t£

detectable eigne! Q_ln* Tlie plot Fig. ?.75 is based on 0 - 2.5 x 10~ Torr. lit/sec. The background level of the detector results taalnly from the presence of test gas which has been absorbed on the surface of the system during pressurizing. The anoung of tills adsorption and of the subsequent desorbtlon during the leak test, is dependent on the gee used, the materiel and finish of the surface. -5*6-

i«i 1 1 1 1 O.t 1 (9 100 1000

PRODUCT Pclt#«lB*H

Ti$. 7.75. P tE values for detecting CA»

As « first step in any back-prcaauriatng test, the signal Eton desorbfed test gas eliould be measured experimentally, using a apectoam of the material which will be used. The effect of significant absorption background can he reduced by heating the specimen.

Figure 7.74, also shows, a part of the curve calculated for laMlnar (viscous) flow, according tot ',-*. \f 1+ ^t'P - P -2 -V^ &

S.A. BOB! mil C.A, H«nn, Vacuum 15_ (7), 347, 1965, where P is the atmospheric pressure, and other notations are as eq. 7.32.

7,45. Sensitive leak, detection methods Halogen leak detector

This detector stakes use of a red-hot (-900'C) platinum filament which emits positive ions. The presence of small traces of halogen vapours (Cl, F, Br, I) increases the emission of positive Ions markedly. It Is this increase in emission that is measured to Indicate the presence of a leak,. The detector consists (Fig. 7.76) of a platinum cylinder mounted on a ceramic-clad heating element placed centrally within an outer cylinder. The heated inner cylinder is made positive (100-500 V) relative to the outer cylinder, and the ion current is read on a mlcroammeter. The halogen detector is most effectively used as a leak detector by placing it inside the vacuum system and probing the ayaten with a fine jet of Freon 12 or other halogen containing gas.

Fig. 7.76. Schematic diagram of halogen leak detector.

One feature of halogen leak detector which can cause difficulty la the relative Lang "aemory" of the detector once It has been exposed to a surge "of halogen gaa. To reduce the memory period, ';he detector head has to be purged with a gas free of halogens, the sensitivity of the halogen leak detector (Table 7.26) is appropriate for detecting medium leaks. -548-

Detectors using vacuum, gauges

These procedures are based on the fact that most vacuum gauges (ionization, thermal conductivity) have a pressure response dependent on gas composition. If the composition of gas in a system changes, the reading on the gauge (detector) reflects this change. Leak, location therefore consists of spraying a test gas on the suspected leak and observing any response of the gauge to the test gas that enters the system through the leak.,

The procedure la very popular for leak location on vacuun systems, because a gauge is usually present on the system. Its major limitation is that it is most directly applicable to the major leaks in the system. It would be very difficult (if not impossible) to locate a leak 1/100 the size of the total system leakage. The procedure is dependent on a constant pressure in the system. If the system pressure varies for reasons unrelated to testing, leak location using this procedure is impossible. The sensitivity of detection using single Plrani or ionization gauges (Table 7.26) Is realtively low.

The principles of operation of thiB techniques are the following f considering a system being tested as shown in Fig. 7,77. If a leak is present, the system pressure will be P.

Q - P2 . C (7,35) where C is the conductance of the pipes leading to the pumps.

If the flow through the leak is viscous the leakage will be inversely proportional (eq. 3.52) to the viscosity of the gas n, thus

a a where C_ is the conductance of the leak, P, is the pressure outside the

system (one atmosphere) and n is the viscosity of air; *\»P2. IE the concentration of the test gas is x atmospheres (x

1 • (7.37) uhile that of sir will be

(7.38) a The total resulting pressure LBt

»««***,- ci»ia fcTj* ^ while Che preeaure difference due to the presence of the teat g*a la <£ro« eq, 7.39 and 7,36):

(7.Ml) If the sensitivity of the gauge Co sir is K. and to the test gas the change of gauge reading AG will be

0.41)

From equation 7.41. it results that the maximim sensitivity is obtained when the test includes:

a) Complete coverage of the leak by the teat gas (x*l); b) High sensitivity of the gauge to the test gas (large K) e) Low value of viscosity (n ) of the rest gas, d) A small value of C . Since the conductance is inversely proportional to the square root of the molecular weight (eq. 3.92), the test gas should have a high molecular weight*

If che leaks are small (IQ~ ate, cc/aec) the flow is moleculart and the leakage 0 is inversely proportional to the square root of the

molecular weight

Since there are a variety of factors involved in choosing a proper gas and gauge combination it is often easier to determine the sensitivity factor as:

i m pressure caused by tracer gas on the leak pressure on system with air on leak which determines the minimum detectable leak:

AF c oa ' » ^in • , • (7'42> where AP, is the smallest measurable air pressure variation. Some experimental values of 4 are listed in Table 7.28. •551-

Table 7.28. Leak testing substitution factors $.

Test gas Hot cathode Piranl gauge ionization gauge

Hutane 10 1 Diethyl ether 5 0.7 Carbon dioxide 1 0.3 Carbon tetrachloride 1 0,05 Benzine 0.3 0.1 Hydrogen -0,4 0.4 Coal gas 0.25 0.25

The gases are preferred to the liquids (diethyl ether, carbon tetrachloride and benzine) as these liquids may block the leak or may enter the vacuum system in considerable concentration through a large leak. Hence the most suitable gaa is butane (Table 7,28) which has low viscosity,, high molecular weight, good thermal conductivity and high ionization probability,, For an ionization gauge» the second best choice is carbon dioxide, while for a Piraoi gauge it is hydrogen (Table 7.28), Carbon dioxide has the practical advantage, compared with butane and hydrogen, of being non-inflammable.

If in a certain leak detection plant the conductance (pumping speed) for air is 5 lit/sec, the air pressure fluctuation is 2x10 Torr, and an ionization gauge is used with butane ("t"10), the minimum detec­

table leak is (eq0 7,42);

The leak detection with vacuum gauges can use: a) a single gauge; b) a single gauge with a barrier that admits only the test B*S» C) too Identical vacuum gauges in a differential mounting.

Single gauge leak detection. The response of the Pirani gauge head to * cast gaa is caused by the Change In thermal conductivity (see Sec, 2.73; and Sec, 6.6) compared with that of air. Since the -552-

pressure Indication of « Pirani gauge is the out-of-balance current of • wheatstoac bridge it is moat sensitive to pressure changes when the out-of-balanc* current is aero. For this reason the control unit ia * modification (Fig. 7.76} of that used in low pressure measurement, in which the balancing aim resistance can be readily varied. For maximum sensitivity a mare sensitive meter Is used, and the head Is protected from temperature changes by thermal insulation or by use of a compensator head.

Fig. 7.78. Control circuit for Piratii leak detector.

Both hot and cold cathode ion igaugea may be used for leak detection (Table 7.26). Their response to a test gas is due to the change in ionization potential.

Barrier leak detection. A method to overcome the difficulty due

to pressure fluctuations produced by outgessingt is to arrange between the gauge and the vacuum system a barrier which passes only the test A;M. Two techniques of achieving this both use hydrogen «e the test gas i the caatcoal»Pirani method and the palladium barrier atcthod. The charcoal-Pirani technique depends on the fact that degassed, activated charcoal cooled with liquid nitrogen will adsorb (Fig. 4.24) readily atmospheric gases, but will adsorb much less hydrogen. A cooled charcoal trap is therefore inserted in front of the Pirani gauge.

The palladium-barrier detector is based on the fact that in the cold state palladium is impermeable to all gases, but when red hot it is highly permeable (Fig. 4.12) to hydrogen. The detector head Is a hot cathode ion gauge (Fig. 7.79) which is evacuated, sealed-off and gettered. In operation the palladium barrier is heated by electron boiabarament, the electrons being produced theroally and accelerated towards the positively charged palladium. Hydrogen test gas which enters the ays tern via leaks, diffuses through the hot palladium and ie

Calhode (—100V) . , Ion collector \ (--150V) '^ovar lube

Fig. 7.79. Palladium barrier leak detector head. ionized by collision with the electron stream. Th2 resulting positive ions are collected fay an electrode which la at a negative potential with respect to the cathode. The current flowing in the collector circuit Is amplified as in the hot cathode xonlzation gauge.

If. water or hydrocarbon vapours reach the hot palladiua, there is a high probability that they will be dissociated to produce hydrogen. -554-

Thia will cause an. erroneous Indication of J leak, or * high background current. To avoid this effect it ii essential to use a liquid nitrogen cold trap between the detector and the vapour pumps. At the completion of leak detection the hydrogen 1* pumped out of the head until the ion current indication is a minimum.

Differential leak detection uses two similar gauges connected in an electrical circuit so that the difference between their readings Is registered. Both these gauges (Pirani, ionization) record the residual permanent gas pressure in the system, but one of thee has a trap which is selective for the test gas. Liquid nitrogen trap can be used with hydrogen as the test gas, while calcium hydroxide trap at roost temperature is suggested for CO as a test gas.

Mass spectrometer leak detectors

In principle any of the types of mass spectrometer described in Sec. 6.9. together with any probe gaa may be used for leak detection, since the device can be adjusted to respond only to that gas,. Although mass spectrometer leak detectors tuned to argon Cor hydrogen) are also constructed, the widely used gas-is helium. for the following reasons: a) Helium's molecular weight is low thus it gives a high leak rate through small leaks; b) Helium occurs in the atmosphere at an extent -4 of only 3x10 percent per volume (Table 1.3); c) There is little possibility that an Ion from another gas will give an Indication that can be mistaken for helium {-except deuterium).

A helium mass spectrometer leak detection system (Fig. 7.80) consists of: s) A vacuum pumping system for pumping the spectrometer tube and associated lines; b) a cold trap for pumping condensable vapours; c) an appropriate test inlet (vacuum coupling) for connecting the vessel to be tested or the calibrated leak; d) the mass spectrometer; e) vacuum gauges controlling that Che filament: of the mass spectrometer be not "on" at too high pressures; f) a pumping system to evecuste the vessel under testa Fig, 7,80, Helium mass spcctometer leak detector.

The mass spectrometer helium leak detector can be used for dynamic leakage measurement (Sec, 7.42), for leak location (Sec.7.43) or for sealed unit testing (Sec. 7.64). Its sensitivity is very high (Table 7.26).

Ion pump aa leak detector

Ion pumps can be successfully used as leak detectors in the range of ultra-high vacuum. For a given type of gas the current drawn by the pump Is proportional to the throughput. Thus for gas of type x

I - Q - S P (7.43) where I Is Che current drawn for a given throughput 0 » P the resulting partial pressure, and S the pumping speed of the pump.

If a leak exists, through which the leak rate of gaa 1 is Q,f the ion-pump current is

•ii-fn^+IrVj/i (7.«) -556-

vhere (^ reprennta the outgamlng load. It »t tl»e t-0, gu of type 1 Is replaced by gu of type 2, thin if Mr a tine t tb» chenge in the Ion puap current is

t A1(t).I(t)-I.[^F]2Cf2a.e"~ ).

(7.45)

since presumably Q the internal outgasslng. remains constant. After a sufficient time the exponential factors approach zero, and the fractional change in the current is

AT (I/P)9 S.Q, (7.46)

±ne data for this parameters and the observed values for AI/I are given in Table 7.29.

Table 7.29.

HrSLATJVK VAl.l'KS Or VlTJIl'IXtt Kl'KEUS, -LliAK /{ATK«, AWJJ If/' 1 UsKl» IN iHSTKHMISISn TIIK t'HAWIK IN (JKTTKR-KM* I'lIMP CURRENT J)UJS Tii SfUKTirfTHis or UXK UAH FOK A.NUVHKK*

IVOIJU jjiut uinjwi'h SJSt QJQi A///t A ).2fi 0 834 f).80 4 0-fi H-

VHa -0.S The value of I/SF for air waa measured and found to be about 20. Using the data fron Table 7.29, and this parameter for air, from eq. 7.46 it results that by substituting helium for air at a leak in the system, the change in current drawn by the ion puap is il » 10 Q . <&I is in amperes; Q . is in Torr. lit/sec). Since Air air , * electrometer circuits can measure currents of 10" A, the minimum —13 detectable leak will be theoretically about 10 lorr. Ht/see. Practical values are in the range 10*" - 10~ Tarr. lit./sec.

-559-

8. VACUUM SYSTEMS

8.1, Basic criteria of design

A vacuum system is the assembly D£ the components used to produce

(Sec. 5), to measure (Sec* 6) and maintain (Sec. 7) the vacuum In a vacuum chamber,, The gauges, pimps and other parts of the system are connected together and to the vacuum chamber, by means of pipes which allow the flow of gases (Sec. 3) from one part to the other. All these parts contribute to the gas load of the system (Sec. 4), as well as to the transport phenomena (Sec, 2) occuring in the system.

The basic criteria which define a vacuum system are:

a) The size and number of the vacuum chambers;

b) The required operating pressure;

c) The gas load.

Vacuum chambers include any evacuated container as: the envelope of electric lamps or electron tubes, the bell jar of small vacuum coating plants, the chambers of large'coating plants, vacuus furnaces, vacuum distillation and drying plants, spectrometers, particle accele­ rators, or space simulation chambers. The sizes of the chambers listed, extend from a few millimeters (miniature lamps) to more than 20 metres (large space-simulation chambers) . The number of the chambers connected to the same pumping system is usually one (rarely two), except the systemB for pumping lamps and electron tubes where 20-50 "vacuum chambers" are connected simultaneously on the same vacuum system.

A A very long vacuum chamber is that of the ISR in CERN-Geneva. see E. Fischer,* Two kilometers at 10~10 Torr- The CEBN Intersecting Proton Storage Rings; Trans, 5th Internat. Vacuum Congress, Boston, 1971. -560-

The operating nretflura depends on the application (Table 1.4). Holding, lifting, transport and forming applications require pressure of 10--00 Torr*; vacuum drying or, degassing 5-100 Torr, vacuua melting -1 "2 -4 -5 10 -10 Torr; vacuua coating, and melting 10 -10 Torr; tube manu­ facturing 10 -10"* Torr; and surface studies 10* -10"* Torr. These requirements situates the vacuua systems in the category of: low-medium vacuus; high vacuun, or ultra-high vacuum (Fig„ 1.1), and define the

kind of pumps (SecD 5,1), gauges (Sec. 6.1) and materials (Table 7„2) to be used* Designing the system consits of deciding upon the features of the vacuum charter, evaluating the gee load, and arranging the appro­ priate combination of pumps, gauges, seals, etc, in order to reach the conditions required for the particular application,

8.2. Evaluation of the gas load

The gaa load is a result of the process (e.g. vacuum drying, degassing, etc.) or a "by-product" of the vacuum chamber, and the components (materials) used. In any case the gas load is the sum (eq. 3.240) of the residual gas remaining from the Initial atmosphere, the vapour pressure of the materials present in the chamber, and the leakage, outgasslng and permeation.

The pumpdown from the initial (atmospheric) pressure to the working pressure of the system is discussed in Sec, 3.72-3.75, The vapour pressure of materials can be the factor limiting the ultimate pressure of the system if the precautions required for high vacuum (Sec, 7.14) are not respected. Usual high and ultra-high vacuum systems, or syterns which are tested "empty*1, have their ultimate pressure determined by the leakage, the outgassing and the permeation. All these factors depend on the kind of construction (Sec. 7) and the materials used (Sec. 4), and this dependence is very often complicated or not exactly known* Nevertheless the gas load and appropriate pumping requirements can he evaluated, by making the proper simplifying assumptions.

The ultimate pressure P in the chamber can be expressed (eq. 3.264) by

""•».'-«•ik • where Q is the tocai gas load. V Is the volume of the chamber, S is the pumping speed in the chamber (eq, 3,28), AP/at is the rate of change in pressure (pressure tisej, and t*V/S is the time constant (eq. 3.271) of the chamber (system).

If the total gas load consists of three main components; Q due

to leakage „ Qn due to outgassing and Q_ due to permeation

Q - Q^ + QD + Qp (8,2) the ultimate pressure can also be expressed as the sum of partial

pressures due to leakage (P.), outgaasfng (P ) and permeation (K) t

(8.3) ru • \ + % + ?P

(according to eq. 8.1)!

(B,4) >i-h-ffih<

'••><-$.*

P -^T-M T (B,6) T ** v Mp thus the gas load and pumping requirement can be evaluated separately for leakage, outgsBsiog and permeation. Thfc nomograms shown in Pigs.

8,lj 802j 8,3, were constructed by using «qe. 8.4j 8,5j 8,6. respec­ tively. The nomograms consider two values of the time constant t - 1 sec (alow pumping) and t - o.l sec (East pumping). The gas load &/V (or Qn/V; Qp/V) la expressed in terms of the specific gas quantity entering into or evolving in a unit voluw of the chamber ————__»N (Torr.liter/sec.m ), as veil as in tena of the rate of pressure rise AP/At (lorr/hr) which would result from that gas load. Number* in italics on tha nomograms 8,1-8,3 indicate the relevant literature reference, as listed at the end of this section.

Leakage (Nomogram I) The gas load Q due to leakage is expressed by

where q ie the leak rate per unit length of aeal (e.g. Torr liter/sec.cm) and L is the length of the seal (cm). From eqs. 8.4 and 8,7 It results that

Usually the shape of the vacuum chamber nay he considered to be either a cube of side a ot a cylinder of diameter <1 and length f. The seal length-to-volume ratio L/V is a maximum in the cose of a cube having seal* (e.g. welding) on ell its sides, and in this case

a a

The ratio L/V is a minimum in the case of a cylinder having a seal at one of its ends

• - |j (B.10)

The nomogram in Flg„ 8,1. considers theie cases of maximum (eq„ 8,9) .and minimum (eq„ 8.10) seal length-to-volume ratios. For a. complicated shape, which cannot he considered as a cube or cylinder. -563-

the actual value of the ratio (L/V) ^ must be evaluated, and the act * resulting equivalent value (d.J) - 4 / CVV> (8.10.a) eq act can be used on the nomograms

Combining eq. 8 with eq6 9, 10 (or 10,a), the specific gas load or the ultimate pressure can be expressed as a function of the size

of the chamber0

The leak rate per unit length q can he considered as a charac­ teristic value in the case of permanent seels (e.q. welds) and as a first approximation in the case of demountable (gasket) seals. In the case of gasket seals the real specific value is expressed (Sec. 7,34) by

JTorr. liter mm (sealing width)I \ J sec * cm (seal length) j

which 1B given (eq. 7e12) by the expression

«L " h • "i - >0E "3r/R •A " • <\ " V (8-U>

where h is the geallng width (see Fig. 7.16; 7.32) p is a factor

specific to the gas (e,g0 for He at room temperature p • 34t) -— * —j ;

see eq0 7o10); P/R is the tightening index (see eq. 7,12; Fig. 7,19) in which F is the compression stress exerted on the sealing surface of the gasket (kg/cm 2) and R 1B the "sealing factor (see eq. 7,11, and Fig. 7,18) refining the sealing ability of the material. A is the peak-to-valley height of the surface roughness (see Figs. 7.15; 7.16). P.-P is the pressure difference across the seal (Torr).

Efficient sealing is obtained up to a tightening index P/R ** 3. At higher values considerable deformations of the bulk of the gasket occurs, which reduces further efficiency. Thus the nomogram (Fig. 8.1) 1B constructed based on P/R • 3, P, - P • 760 Torr. p • 79 --—• * —-*> * 1 o * o sec (3,3 (for air), and two sealing surfaces (on both sides of the gasket), -564-

l.e. (according to eq. 8,11):

2 2 qL - h q^ - 79 x 2 x e" x 760 A - 14.5 A (8.12)

The nomogram {Fig. 6M) is based on eqs„ 8B8-8e12» The factors involved are grouped in the nomogram under the sections; sealing effort, seal width, surface finish, leak rate, gas load, ultimate pressure and system size. The stales of the nomogram form three alignment charts; t i

1) scales P, Ph, and h; 2) scales h, qL and qL; 3) scales qL, QL/V and system size,,

The example marked e,f>BOOn> on Fig. 8„1H can be expressed as follows: A cylindrical chamber of diameter d = 10 cm and length 1 - 30 a (d.l - 300 cm ; point e) is evacuated by fast pumping {T - 0.1 sec). The desired ultimate pressure due to leakage Is P, • 2 x 10*" Torr (point f). The horizontal through point f shows that in order to -2 achieve this, the gas load due to leakage must be lesB than 2 x 10 Torr liter/sec, a (point g, right scale). Cut off from the pump this _ ? Chamber will have a rate of pressure rise (due to leakage) of 7 x 10 Torr/h (point g, left scale)«

The leak rate required per unit length of seal Is given by the t t intersection of line e-g with scale q,, i.e. q, <

sec cm (point b)0 This leak rate can be achieved by elastomer seals, t

at shown by the ranges narked parallel to scale q,B Since a bakeable seal Is wanted, a wire seal is chosen,,and the width of the sealing

surface is assumed to be h * 0,4 mm (point k)0 Line h-fc Intersects scale q, at q, • 6 x 10" Torr„ liter* mm/sec, cm (point 1), sad the horizontal j through point 1 shows that a surface finish of 20 fin

RMS is required (eDg, honed surface),, Gold (point m) is chosen as the wire material. The intersection of line k-m with scale Ph deter­ mines the sealing effort Ph • 46 kg/cm (point 1)» This seal could be closed just by the atmospheric pressure if

2 2%- i. it . A (P.h) thus if di 4 x 48 • 192 cm PB?

Solt Rubbers Hard

I i W'i "•" iwn'" PUg/on?)

,1 ? °™ ^ d(cm)(atmdosure] l • |'"rtiMi|'"i"i,*M_ .' 1 Ph (kg/an) Knife edge S ^ ^ l Wire_ T7^ O-riPg Flat

ff " • *•?•••? V •'

Superfihish tow , + _,^jju^jggj^__..\— .J^T. -l^T-c ,._«- A<(jnH.yS> n~ A(on> ITySr q (tSLM- ESS 1 iLn-^f H sec cm 1 -Elastomer ^S , fjorr. cm/

OL OI OI OJ OI Oi — O 3 S fVessirerise _LLC_if_ J) *. di i & — . K) «* (SHM • c Sn 3i Si 3A °» ** ^ V3"1 S- 3i S* 3i i (HH«) ^

Fig, 8.1 a* shorn by point t on scale d (atmospheric closure). Sine* the seal

baa a HI a—tag of only 10 em, it must ba tightened mechanically by a

force of iO.T x 48 = 1500 kg.

For rcaaona of clarity, the above example follow* the scales in the order they appear on the nomogram. Obviously the nomogram can oc osed by beginning with any of the scales, and proceeding towards any other scale.

Gas load, pressure rise and ultimate pressure obtained front nomogram I (leakage) have to he added to the respective values obtained for outgasslng (nomogram £2) and permeation (nomogram III) in order to obtain the total value of these factors. putgMslng (tomogram II)

% - qD . « (8.13) where q* la the specific outgasslng rate (e.g, Torr.liter/sec,cm - Fig. 4.30), a la the area of the outgasalng surface (cm ).

item eg, 8.5 and 8,13 It results

As a basis for the evaluation, the outgasslng surface is considered to ba equal KO the Inside surface of the chamber, and the shape of the chamber is esstawd to ba similar either to a cube of side a. or a cylin­ der of dtaneter d_ (and length i). The surf ace-to-volume ratio in the cas« of a closed cube is

whlla for an open-ended cylinder The nomogram (Fig,. 8.2) for the evaluation of the gas load due to outgassing Is based on eqs. 8.14 - 8.16, and on the range o£ q_ of various materials and treatments (Fig. 4.32). The factors involved are grouped In the nomogram under the sections: outgasstng rate, gas load, ultimate pressure, and system size.

The values of q_ for untreated, degreased and polished states were taken for 4-8 h of pumping. For baked metals the high q_ end of the range corresponds to baking at about 300"C for 24 h, the middle of the range to baking at 400°C for up to 100 h, while the lowest values for stainless steel also include an additional subsequent baking at 1000#C for 3 h„

The example marked e,f.,.k on Fig. 8.2, represents the case of a cylindrical chamber d • 10 cm (point e) which has to be pumped by slow pumping (T - 1 sec), to an ultimate pressure (due to out gassing) P_ - 2 x 10~ Torr (point f)» The horizontal through point f, inter­

sects scale Qn/V at point g, thus in order to achieve the required -4 P-, the gaB load due to outgasBing must be less than Q_/V - 2 x 10 Torr. liter/sec,m (point g, right scale). Cut-off from the pump the chamber will show a rate of pressure rise (due to outgassing) of -4 7 x 10 Torr/h (point g. Left scale). The specific outgassing rate required 1B given by the intersection of line e-g with scale q_, i.e.

% " 4 x 10 Torr liter/sec. cm (point h). This means that the entire (100%) inside surface of the vacuum chamber may be constituted o£ any of the materials In any of the treatment states, which are intersected by the horizontal through point h (e.g. polished Al) or which lie below this line.

for evaluating the distance at which the q_ value has to be shifted «) •3 t» •c^ fco-H^i'»'ti-«i a s iutab,. !H 111 imiE52SZaSS3 PolyomMes

1 » liiilllllllilllllllllBW^WMW^I VHon imiia BtwSSiKswiSa PTFE * •.- is is:.. K ts -.-.a iiiiiiiiiiiimiaMwmv^wA^vsga steels

J & . VMdW////Wt ////^;\\'K*t Aluminum miiiiHi£s m>s Es '•* '£; ** «* *• ^

«TimimMiiiiiiiii?MiiTiimiBiwoi)^^

ir *to!i«ii M ©, S, 5, i5,io, IB 1 o, o. o, 5. n Morr.lH I . I* . A . A . )*, l£ |J* |.J-! . I* J_'"_J^J ' MSEW

Q. O. 5, S, 3. o\ S, 3, 3, „ 5 5L 5 S^ Pressor* risa $ '•' 'fi' 6 CI' '*,' < 'M»CI' Ci' CI' '. •• I. ,' I, ,• Ci' I ("»,/W I a, o, B, „. „, ^ 5, i. S, 3, S, S i\i:~v=|T.£I»« + Ci FS H & to i i I ) I II ll M ll ii II II ii II i U, *-". -M-Vt J-"i %J, *-»• tJA U. <-«> < ° "^.gf.jr,,,r,K A°"a" 1 %!%P"l" .'±"'ilinin i tarral«m 5 \ + \ \ 6 a (cm) closed » I ' lult'ii'i' v ' inliyi'i1 i g' S'"i X

*lg. 8.2. if the outgassing surface is smaller or larger than the inside surface of the chamber. By shifting point h, according to this correction scale, we find that 10% of the surface can also be constituted of materials (treatments) at or below the level marked i (e.g. baked Viton) and 1Z of the surface area can also be of materials at or below the level j (e.g. degreased epoxy). If the outgassing surface is not only that of the chamber, but, as usual, includes parts built inside the chamber, point h has to be shifted at a level corresponding to the ratio between total surface and that of =he chamber. The example shown on the nomogram (level k) corresponds to the case when the total outgassinf surface is 5 times (500Z) that of the chamber. At the level k, baked Al or Cu, or polished stainless steel is required for the entire surface. In this case the distance between level k and i, measured on scale Zs, shows that only 235 of the surface can be a material (treatment) of level i (e.g. baked Viton).

Permeation (Monogram III)

Qp - 1p {• (8.17)

8 where qQ is the permeability (e.g. • • • * • • ' • ^jp), is the permeable area (cm ) and h the wall thickness (mm),

From e

Nomogram III (Fig. 8.3) for the evaluation of the gas load due

to permeation is based on eqs, 8t15 - 8,18 and on the q_ range of various materials for various gases (see also Figs. 4.12, 4.13). As a basis for the construction of the nomogram it is assumed that permeation takes place on the entire surface of the chaaber (lQOZs). -570-

Th* mppropri*t« fraction of permeable surface 1* taken by shifting

tut value according to scale Xu% The abape of the chamber is asstsMd to be either a cube of side toi a cylinder of diameter cl (and length 1). The surface-to-volume ratio a/V is according to eqa. 8.15, 8.16. Nomogram III Is constituted of two alignment

charts: 1) scales h,^_ and qp/h, and 2) scales qp/h, Q /V and system size. Parallel to scale h the ranges of wall thicknesses required to withstand one atmosphere pressure difference are marked for snail diameter (a - 1 cm) and medium else (a - 10 en) windows of various materials. Parallel to the c_ scale the ranges of permeation rates

of typical materials for various gases> for a pressure difference of 1 Atm are plotted within various temperature ranges. The permeation through nonmetals is proportional with the pressure difference (see Sec. 4.22}, while for metals it is proportional with the square root of the pressure. In order to evaluate the permeation from the surrounding atmosphere, scales taking Into account the appropriate corrections due to the abundance of various gases are plotted for metals and nonmetals.

The example shown of Fig. 8.3, represents the case of a allies window of thickness h - 0.1 BOD (point 10 indluded in a vacuum chamber. Tha q_ value of SiO, for He Is q • 5 x 10~9 To" Ut " 2L (point f> Jr * c sac cm' at AP - 1 atm. Since the partial pressure of He in the atmosphere is only A x 10~ Torr (abundance scale), point % must be lowered on scale q„ by the distance shown by the scale "abundance" for noavetals,

to point n. tine k-nt intersects scale qp/h at point n. The area of the window is only 10% of that of the chamber, thus point n must be lowered by the distance shown by scale Ss (for 10%) i.e. to point o. The vacuum chamber is similar to a cube of side a - 5 cm (point r); the line o-r intersects scale Q /V at point p, which shows that the -8 3 gas load resulting from permeation is about 4 x 10 Torr lit/sec. m . Point p shorn on the scale "pressure rise" a value of 10** Torr/h, for the system cut-off from the pump. By fast pumping (T * 0.1 sec), an T I '< ' I ' • ><} i ' jj Mmm) tf* |8lS|f*» v«S£=-'i-'r£- i 100-20 O'C /i •Rubbers «;#—-> Roan

•C02 fwifi «/ -H;a0

— _^_ •a'r-J /• •" " •" w "* >cv l crttl%i^Es^ta ^^'-^ U I il." 1 Li JIUXU- Non mElolsl ..„„,„,„ ' fif SPlfcfbl*- ]"«' ftg^fl/> " ^v^^^M?^ :•?•,• .?»?(g&) "58S«iS*

I 'fcV'&fi'^8- o, o. o, 5, 5, M. fer ...M^'i-?M^^mcr^vvv^. ir-feisHa- -'V \

a(cnOtfosed.<

' 'dtan] 0 ^ 1,'e.i.

Pig. 8.3.

sii:; .io ..oj-U}.i3.iE:il hna JooJbaiJI ,=T3iIntaaf. .,'..:! {point q) can be achieved. The gas load values obtained on nomogram lit refer only to permeation, the total gas loud is obtained by the sua of the values obtained in X, II and III,

Fliplng requirements

The gas load and Che ultimata pressure define the required pumping speed at the vacuus clumber (eq. 8,1). The required punping speed and the conductance of the connecting parts {Sec. 3} define the pumping speed of the pump, which together with the ultimate pressure define the kind of pump (Sec. 5) to be used,

In.Lthe choice of the pumps the following practical rules may be useful: a) The diffusion pump represents a. small amount of the cost of the system, thus the selection of the largest pimp which fits the chamber is not an extravagance, b) If the phstp is doubled in diameter the pumping speed Increases about four tines, thus the ultimate pressure decreases to a quarter. all this onlv if the conductance between gimp and chamber is large compared to the pumping speed. c) Cryogenic pumps have pushing speeds ap least ons pfder o£ magnitude greater than diffusion or ion pumps, thus they arc app*ap*iejto for hantuing high gas loads at low pressures.

8.3. Vacuum chambers

2h» design of vacuum chambers include its physical design and ° .-.- • ;-u-!t' jv.fi physical aesxfn ana its functional desing, the physical design refers aainly to the mechanical strength of the chamber. For snail chambers, the thickness of the veil must correspond to the requirements listed ia fable 'fyi .

For details on design of vacuum chamber*, and description of the construction of Vacuum furnaces, rifftr^oj'l^A^SfiiMarS.^ana^ook of Sigh Vacuum ^ibterl^V•?*^h^^r^toit, |?fr -573-

Large chambers hae to be designes with stiffeners according to the rules specified for pressure vessels.

The functional design refers to Che vacuum process which has to be carried out in the chamber. The functional requirements usually dictate the choice of the material, depending on the required final pressure, temperature, heat transfer, corrosion, magnetic properties, etc

8,4, Pumping combinations

The simplest vacuum system consists of a chamber (volume) evacuated by a mechanical (rotary) pump. Such a system also includes (Fig. 8„4) atleast a vacuum gauge and an air admission valve (vent). The vacuum gauge generally used in such a system is mechanical gauge

(Sec a02),a McLeod (Sec 6,35) or an Alphatron (Sec 6.74). A single stage mechanical pump (Sec 5.24) is sufficient for pressures -1 -2 down to 10 Torr, while a two-stage pump is needed for 10 Torr. This simple system has two contamination problems: the pump oil may contaminate the chamber, and the chamber out gassing products may contaminate the pump. In order to eliminate the first problem a trap has to be included between pump and chamber. When water vapour is the predominant gas, a gas-ballast pump should be used (Sec 5.24), Diffusion pump systems

The simplest form of diffusion pumping system consists of a

diffusion pump (Sec, 5D33), its backing rotary pump, the chamber being evacuated, the interconnections» a vacuum gauge and a venting valve. In order to allow venting of the chamber, while the diffusion pump Is hot the system also includes a valve (baffle valve) between diffusion

pump and chamber (Figa 8.5), and a by-pass system through which the rotary pump can be used either for roughing the chamber volume:, or for backing the diffusion pimp. For large systems two rotary pumps

are required (Fig., 806), one of them is used for roughing and is sized -57*-

•(7)— -^•vw

-Mffiftanicai Pump '

Fig, 8.4. Simplest vacuum system.

Fig* B*5, Diffusion punping ay-stm with a *lngl* rotary puap. on Che basis of the required puuipdown tine (Sec, 3.72)» the second on« 1B sized only for backing the diffusion pump.

Mtdionical Pumpi -Diflmlon Purnc

Fig. 8,6. Large diffusion pumping system.

Generally a high vacuum gauge is required on the chamber. The addition of another high-vacuum gauge between the pump and the valve la advantageous when the system performance Is being ohecked, and when leak testing Jr troubleshooting the system. These gauges are .usually ipnisatjLon gauges (hot or cold cathodes - Section 6,7). The gauge located between the diffusion pump and the mechanical pump is e. Pi rani gauge.

Vacuum systems such as shown in Figs. 8.5{ d.6, are capable of maintaining pressures between 10~ - 10~ Tore, depending on the size of the pump and chamber and the gas load. The principal limitations of the uLtiamte pressure are the water vapour from outgassing and the pump oil from the diffusion pump. To overcome these prohlens a liquid nitrogen trap (Sec* 5*65) can be used between the diffusion pump and the vacuum chamber. -576-

Systema using molecular pumps (Sec. 5.28) or ion pumps (See. 5.4) but evacuated by a rotary pump often experience oil contamination problems due to the rotary pump. The common practice of pumping by the rotary pomp to a few microns prior to opening the diffusion pump is probably the principal cause oC hydrocarbons in vacuum systems (see Rules for Operating Vacuum Systems; in tfcia section).

The high vacuus (baffle) valve has an important effect on -7 -8 systems operating to the 10 - 10 Torr range. The valve sets as a restriction for the pumping and as a gas source. The high vacuum valve increases the ultiamte pressure in this pressure range at least by one order of magnitude. Thus its use must be well justified by the need of rapid operating cycling to atmosphere.

Backed vacuum systems

To obtain low ultimate pressures the entire vacuum system may be cooled at low temperatures to reduce the outgassing of all the

sourcest or the entire vacuum system may be subjected to baking* Usually the cooling is possible only on parts of the system (see Sec, 5.6), but baking is possible on the entire system if properly constructed. It must be noted that the liquid nitrogen trap and the upper regions of the diffusion pump are part of the high-vacuum system* and these parte should be included in the baiting process.

8.5. Bales for operating ;y.acu«a aye-teas

, When starting a mechanical pump, insure first that the rotor moves in the correct direction, and that the level of the oil la correct. 2, Always vent a mechanical pump to atmospheric pressure when the power is turned off. The presence of a residual vacuum in the pump frequently will cause oil to be drawn into the casing or the system. -577-

3, Do not permit a mechanical pump no exhaust a high-vacuina system below a pressure . of a few hundred microns unices the pump Is separated from the high-vacuum chamber with a trap stopping the puotp oil vapours from entering the chanter, At lover pressures, the flow is molecular, thus the oil vapour expands towards the chamber, yielding an excessively high vapour pressure hydrocarbon contamination, which may require many hours to remove.

4a Do not run a mechanical pump at excessively high pressures for continuous periods. The motors are usually not sized for such runs. The pump will eject oil together with the gas,

5, A diffusion pump should be ccoled to a safe intermediate temperature before it is vented to atmosphere, Venting at a too high temperature results in oxidation of the pump fluid and an excessive carryover of the fluid into the mechanical pump.

6» Check that the cooling water supply for the diffusion pump is turned on prior to the heating, Diffusion pumps should be provided with a thermal protection device to turn the diffusion pump heating off in the event of loss or failure of the couling water supply.

7„ ID liquid-nitrogen trapped systems, the trap should b'i cool enough to condense the diffusion pump oil prior to turning on the

diffusion pump0 Maximum backset reaming

8„ Do not vent a liquid nitrogen trap to air while cold. Remove all liquid water condensate from the reservoir of a liquid nitrogen trap prior to use to avoid water freeaup in the trap. On the initial pumpdown, it is advisable to partially fill the liquid nitrogen trap, so that the principal vacuum system condensables can be trapped on the lower portion of the trap. After high vacuum has been reached, the trap should he filled to a higher level. A liquid nitrogen trap which has elastomer seals at the top should not be overfilled, as the freezing of the gasket may cause leakage. -578-

9. Vbm tb* cheater is vented Co atarasphezic pressure for short period of ti»a. It la advisable to use * dry, inert gu such aa Ritresss. to sdniedse the •oistuxe adsorption on the eurEaces o| tha vectaae systaeu Venting should alwaya be on tha cbaaber aid*, never vast tha syteet froai tha forellne of the diffusion pu«p. 20.Ionization gauges

REFERENCES MARKED ON NOMOGRAMS FIG. 6.1 - 8.3.

LEAKAGE (NOMOGRAM I)

1. ARM&ND, G., LEJAY, Y. and PAIGNE, I.. Le Vide, ^9_, 436 C1964). 2. ARMAND, G., LAFUJODLADE, J. and PAIGNE, I,, Trans. 2nd Internat. Vacuum CongreBS, Pergaoon Press, Oxford, 1962, p. 1091. 3. ARMAND G., and LEJAY, Y,, Rep CEA PA-PIEL/RT Ho. 229, Saclay, 1967, 4. BOULLOuD, J.P. and SCHWEITZER, J., Le Vide, 14_, 241 (1959).

5. BRIDGE, H.t et al., Adv. Vac, Scl. Technology, Pergamon Press, Oxford, I960, p. 481. . 6. PISCHHOEF E. et al., Le Vide, 17_, 195 (1962). 7. GALE, R.F. and MACHI1I, C.F., J, Sci. Instrum. 30_, 97 (1953), 8. GARROD, R.I. and NANKIVELL, I.F., Vacuum, 11, 139 (1961).

9. GITZEHDANNER, L.G„ and RATHBUN, F.O., Rep. Genera.1. Electric, 65 GL73, 1965. 10. GUTHRIE, A, and WACKERLING, K.K., Vacuum Equipment and Techniques, M<- Gra»-Hill, New York, 1949. 11. HOCH, H., Vaku«a-Technik, 10_, 235 (1961). 12. HOLDE11, I., HOLLAND, L. and LAURENSON, L. J. Sci. Instrua., 3£. 281 (1959). 13. JORDAN, I.R., Trans, 2nd Internat. Vacuum Congress, Pergamon Press, Oxford, 1962, p.1302. 14. KOBAYASHI, S. and YADA, K., Adv. Vac. Scl. Technology, Pergsmon Press, Oxford, 1960, p.248. 15. LAMGE, W.I. AECU - 3889/3890, 1957. 16. MARK, I.I. and DREYER, K., Trent. 6th Amer. Vacuum Symp., Pergamon Press, Oxford, 1960, p.177. 17. MONDAY, G.L., Nucl. Inst rum. Methods, 4., 367 (1959). IB. RATHBUN, F.O., NASA N63 - 18159/1963. 19. REDMAN, I.D. at al., ORNL -3472. -580-

20. BOTH, A., VACUUM Sealing Techniques, Pergamon Press, Oxford. 1966. 21. BOTH, A. and WEAK, A., Vacuum, IS, 309 (196B). 22. BOTH, A., Proc. 3rd Internal. Conf. Fluid Staling, Cambridge BHRA, 1967, paper C2. 23. BOTH, A., Vacuum, 20, 431 (1970). 2*. TU1HBR, IA., PHXARD, H.M. and B0FIHAK, C.R., J. Set. Inetrum. M, 26 (1962). 25. HEITZB., D.B., Rev. Set. Inotrun., 31, 1350 (I960). 26, WEEELER, K.E. and CARLSON, H.A., Trans. 2nd International Vacuum Congress, Pergemou Press, Oxford, 1962, p. 1309.

OSttGASSISC {WMOGRAM 2)

1. AH0ICM0W, I. and COUHXAUD, I.P., Le Vide, 24, 181 (1969). 2. BARTON, R.S. and COVIER, R.P., Proc. 4th Internat. Vacuum Congress, Manchester, 1968, p. 775. 3. BARTON, R.S. and COPIES, B.P., J, Vac. Sci. Techn. 2., 113 (1965). 4. BARTON, R.S. and SOVIEE, R.P., Vacuum, 20, 1 C1970). 5. BARRE, K.R. and WHGODIN, 6., La Vide, 12,, 195 (1957). 6. BASALAEVA, K.I., Soviet recta. Pbya. 3, 1027 (1958). 7. BECDUW, U., Vacuus, 13, 349 (1963). 8. BLEARS, I., 0REER, E.I. and NIGHTIHCALE, I., Proc, 1st Internet. Vacuum Congress, Pergaaon Press, Oxford, I960, p. 473. 9. BOULASSIER, J.C., Le Vide, 14, 39 (1959). 10. BX0HM, R.D. Vacuum, 17, 505 (1967). 11. CALDEX, X. and LEHIX, G., Brit. J. Appl. Mljre. 18, 1459 (1967). 12. CRAWLEY, D.I. and de CSEMATOire, L., Vacuum, 14_, 7 (1964). 13. CSEKKAIOKT, L. de Vacuum 16, 427 (1966). 14. DAS, D.K., AEDC, TDK - 62 - 19/1962. 15. DAYTON, B,B,, Trans. 6th Amer, Vacuum Symp., Pargamon Press, Oxford. 1960, p.101. -581-

16. DAYTON, B.B., Trans. 9eh tear. Vacuum Syap. Macslllan, N.Y., 1963, p. 293. 17. DAYTON, B.B., Trans. 2nd Internat. Vacuus Congress, Pergaaoa Press, Oxford, 1962, p. 42. 18. GAME, S. and CHRISTIANS, K. Vakuum - Tecbnik. LI, 9 (1962). 19. GELLER, R., Le Vide, 13, 71 (1958). 20. HAEFER, R. and UINKLER, 0., Vakuum-Tecnnlk , 5, 149 (1956). 21. HEHRY, R.P., Le Vide, 2£, 316 (1969).

22. JAECKEL, R0, Trans. 2nd Internat. Vacuus Congress, perga=on Press, Oxford, 1962, p. 17. 23. MARKLEY, F., ROMAN, R, and VOSECEK, R., Trans. 2nd Interoac, Vacuum Congress, Pergasnn Press, Oxford, 1962, p. 7S. 24. MUHCHHADSEN, H,, ion and SCHITTKO, F,J,, Vacuo. 13, 549 (1963). 25. POWER, B.D. and CRAWLEY, D.J., Proc. 1st Internat. Vacuus Congress Pergaann Press, Oxford, 1960, p.206. 26. POUER. B.D, and R0B50N, P.C., Trans. 2nd Internat. Vacuus Congress, Pergason Press, Oxford, 1962, p.1175.

27. SANTELER, DBJ», Trans, 5ch Aver. Vacuum Syap., Fergaawn Press, Oxford, 1958, p. 1. 28. SCHITTKO, F.J,, Vacuum, 13, 525 (1963). 29. SCHRAM, A., Le Vide, 18.. 55 (1963). 30. STRAUSSER, Y.E., Proc. 4^h Internat. Vacuus Congress, Manchester, 1968, p.469. 31. THIEHE, G., Vacuus 13., 137 (1963). 32. IHIBAULT, J.J,, R0USSEL. J. and J0URSAN, H.. Le Vide, 22_, 309 (1967). 33. ZMLNIN, V.S., ZHILNIHA, L.P. and KUZNIN, A.A., Proc. 4th Internee. Vacuust Congress, Manchester, 1968, p.801. -582-

TEMUalO) (MOHOCMH 3)

1. CMHLEr, O.J. and CSEBNAXOOT, L., Vacuw, 14_. 7 (1964). 2. DcMION, K.B., Brit. J. Appl. Hijra. £, 414 (1957). 3. ZSCHBACH, H.I.., GxOSS, ». and SCHtLIEH, S., Vacuus), 13, 5*3, (1963). 4. HOBXON, F.J., trans, 2nd Incarnat. Vacuus Congress, Pcrgeaou Press, Oxford, 1962, p.8. 5. KOGEKS, C, at al., J. Tachn. Assoc. Pulp * Paper, Ind. 39_ , 737 (19S6). 6. WALDSOWIDT, E., Metallurgy, £, 749 (1954). 7. TOtiBG, J.R. and UHETXEN, H.R., Trans. 2nd Internat. Vacuum Congress, Pergaston Press, Oxford, 1962, p. 625.