375 OCR Output

For a given installed pumping speed this determines the static pressure(iv) vacuum outgassing rate

atmospheric pressure and hence the frequency changeprevent collapse due to buckling (b) the elastic deformation of the cavity under externalThis is characterised by Young's modulus and determines (a) the minimum thickness to(iii) mechanical stiffness

For a given power dissipation this determines the temperature levels(ii) thermal conductivity

For a given geometry and frequency this detemiines the max. possible Q( i) electrical conductivity

The important material properties are:

Niobium or niobium-sputtered copper (TRISTAN and LEP SC cavities).(PEP)Aluminium

Copper—clad steel (SPS , ..... )Copper—electroplated steel (KEK APS, GSI UNILAC...)Bulk OFHC-type coppers (SLAC, DORIS, PETRA, LEP .... )

with cavities made from these materials are given as examples in brackets.The most commonly used materials are given below. The names of some of the machines

for example the inside surface had to be covered with 10-20 nm of titanium nitride [1].has a high coefficient of secondary emission. To reduce multipactoring in the PEP cavitiesbut aluminium cavities tend to suffer from multipactor problems since the oxidised surfacedeposited onto, or attached to, a stronger material such as steel. Aluminium has been usedlow temperature superconducting structures. For strength reasons the copper is often

Copper is the preferred RF surface for room temperature applications and niobium for

2. MATERIALS

accelerators.process in a simple way and then by giving at least one example taken from present dayprocess but rather to unveil the panoply of possibilities that exist by first describing theused to construct and cool them. The aim is not to give a rigorous presentation of eachthe materials, joining techniques, fabrication methods and cooling schemes that are normally

This paper describes for those with little or no experience in the design of RF cavities

1. INTRODUCTION

present day accelerators.described. The processes are illustrated by examples taken fromfrequency cavities for high energy particle accelerators areschemes that are normally used in the construction of radioThe materials, joining techniques, fabrication methods and coolingABSTRACT

CERN, Geneva, Switzerland

I. Wi I s 0 n

CAVITY CONSTRUCTION TECHNIQUES

376 OCR Output

process.

electrolytically deposited copper may be as much as a factor 10 worse depending on the1.5 x 10*12 (baked) Torr l s‘1 cm*2 after 100 h pumping at 20 OC. The value forTable 1 have all approximately the same outgassing rate of 1.5 x 10*11 (unbaked) andIn the mechanically worked condition, chemically cleaned surfaces of the metals given in

the metal is plastically deformedgrain size. Small grains reduce the risk of crack propagation along grain boundaries whenbe mechanically worked by forging, rolling or extruding to eliminate porosity and reduceMaterial in the as-cast state is not recommended for cavity construction. Cast billets should

Silver 4.16.2 x 10 0.76 x 104

Aluminium 2.23.7 x 10 0.72 x 104

Mild steel 0.55.5 x 10’ 2.10 x 104

Stainless steel 0.21.5 x 10 2.10 x 104

OFHC copper 4.05.8 x 10 1.20 x 104

(W / cm /°C) (daN I mml)(9*1 cm*1)

Material Electrical conductivity | Thermal conductivity | Young's Modulus

Typical material propertiesTable 1

given in Table l for the most commonly used materials at room temperature.Typical values for electrical conductivity, thermal conductivity and mechanical stiffness are

This is important only if the cavity is subject to extemal magnetic fields.(xi) magnetic permeability

example Ref. [2]a danger of surface fatigue damage due to RF impulse heating and cooling - see fornew generation of high gradient pulsed accelerating structures for linear colliders there isIn certain circumstances this may determine the lifetime of the cavity: example - in the(x) mechanical fatigue

5 or morethickness ratio of 200 contains 0.1% Ag which increases the creep resistance by a factorexample - the OFHC copper of the LEP storage cavities which have a diameter-toThis may be important to prevent the buckling of thin-walled structures dtuing bakeout:(ix) mechanical creep resistance

circumstances - see examples:The following material properties are not usually important but may be in special

(viii) cost.(vii) weldability or brazability(vi) formability or machinability

incident electron — this influences the degree of multipactoring(v) coefficient of secondary emission of the surface - number of electrons produced per

377 OCR Output

interferometric feedback with 25 nm resolution, vibration isolation and air bearing spindlesfinish Ra = 15 nm. These Pneumo machines have CNC control, closed-loop laserthe state of the art in diamond machining. Tolerances of i 2 um were obtained with a surfacePneumo MSG 325 lathe for the CLIC 30 GHz prototype accelerating sections and representThe copper cups in Fig. 1 for example were machined, to the dimensions shown, on a

Roughness obtained with special care

Roughness obtained with usual workshop practice

POL ISH ING

LAPPING

MILLING

DIAMOND TURNING

TURNING

DRILLING

PL ANING

Ra (pm) | 3.2 [1.6 |0.8 |0.4 |0.2 |O.I |0.05 |0.025

SURFACE FINISH I N8 I N7 I N6 I N5 I N4 I N3 I N2 I N1

Typical surface finishes for classical machining operationsTable 2

surface finishes better than Nl.lathes (see below) are capable of machining components to tolerances of a few um with

of mono-crystal diamond cutting tools. Such tools used in conjunction with sophisticatedFor applications requiring the very best dimensional accuracy and surface finish use is made

cutting copper in particular should be sulphur—free.the cutting fluid. Otherwise the fluid should be water-based as opposed to oil—based and forfinal machining of the internal vacuum surfaces of cavities should be dry or with alcohol asthe mean arithmetic distance of the true profile from the average profile. Whenever possibleprocesses. As a general rule of thumb the finish required is Ra = (skin depth /4) where Ra isand surface finishes that are normally associated with the various classical machiningone of the cheapest and most accurate machining techniques. Table 2 gives the tolerancesaxisymmetric geometries be used so that the components can be turned on a lathe. Tuming is

For cavities made by machining it is recommended that whereever possible

3.1 Machining

3. FORMING TECHNIQUES

of copper to niobium has been used extensively at Argonne National Lab. [3].Copper—clad steel is produced by explosion bonding or hot rolling. The explosive bonding

has also been successfully used - for example in LEP accelerating cavities.electronic copper which is de-oxidised with phosphorus and has up to 30 ppm of oxygenThe oxygen content of OFHC copper is normally restricted to 5 ppm but the BE 58

380 OCR Output

Fig. 4 Fabrication of SPS cavities from two half-shell pressed parts

0 100 200mm

$.!P.¤&¤¤s/

R F shi

~ gg;

Leann

Welded hl

Qld fvmgd M6

deformation in the longtudinal direction.assembly welding is made by stretching or compressing the cavity to produce plasticstandards). The main dimensions are given in Fig. 4. Overall tuning of the cavity after iinalsuch half-shell pressed parts for which tooling was available in industry (pressure vesselThe SPS sin gle-cell 200 MHz copper cavities for LEP operation were consmicted from two

Fig. 3 Forming outlet ports in cavity walls by pressing

pre—cut hole as shown in Fig. 3.made in the main wall of the cavity by pressing or drawing a hemispherical tool through ain this case the outside surface of the cavity. Ports for tuners and other accessories are easilyfabrication of a few cavities. The reference surface provided by the surface of the mould 1sOCR OutputOCR Outputdcvclopmcnt of suitablc tooling makes this method less attractive than others for the

382 OCR Output

a thick layer of electrolytically deposited material onto a metal mandrel which isIn the electroforming technique the main body of the cavity is produced by building up

3.8 Electroforming

value.measured Q values of the cavities corresponded to 89% of the calculated SUPERFISHwith a 200 um copper surface layer deposited using a pyrophosphoric acid bath. The

TRISTAN at KEK have used this method. 'I'he APS cavities are made from low carbon steelBoth the 508 MHz disk—and-washer and 508 MHz aperiodic structures (APS) for

cm*Z after 100 h pumping at 20 °C.a conductivity of 58 x 10*4 §2*1cm*l and an outgassing rate of about 3 x 10*Torr 1 $*10 1and involved - the GSI plating process for example has 16 steps but produces a copper withand care is needed for a successful application of this technique which is often quite longmild steel for the fabrication of their Wideroe and Alvarez tanks [6]. Considerable expertiseThe GSI at Darmstadt in particular has invested considerable effort in the copper plating of

however that nickel is magnetic!layer before copper plating is normally necessary to obtain a satisfactory adhesion - notethe best OFHC-type coppers. In the case of stainless steel a pre-plating with a 4-5 pm nickel

high outgassing rates, this technique can now produce copper qualities which rival those ofRF fields in cavities. At one time avoided for justified complaints of low conductivity andconductivity material (usually copper or silver) on steel surfaces which would otherwise see

This is the standard method for putting thin coatings (say up to 100 pm) of high

3 . 7 Electroplating

as substrates for the fabrication of niobium-sputtered superconducting cavities.sources for electron emission makes these multi-cell structures particularly attractive for useThe absence of welds which are normally associated with surface irregularities and hence

Fig. 7 Expansion in a multi-part die

nm

_

cx § ur. c muassuruz M ·. `UIHHV" ' ' ¢ GM \

expansion as the length of the tube decreases.tube is put into a multipart die which is initially open but which closes progressively duringOCR OutputThe expansion phase, using pressures up to 200 bars, is shown schematically in Fig. 7 - the

383 OCR Output

Fig. 8 The Daresbury Laboratory 500 MHz electroformed cavity

mm

0 100

Tuner

1,./ ,.-l-. -\

Water inlet Water outlet

Beam centre line

q l

?·*f.»·*" F'?

warg; inlet Water outlet

§77I\W

Input window

onto a steel mandrel from which they were released by heat after completion.into its walls and the mandrel etched out chemically. The half shells were electrodeposited13.5 mm. After electroplating of the inner shell was finished cooling channels were milledincorporated into the main body of the cavity as the electrodeposited layer was built up totuners and other accessories were fixed onto the aluminium mandrel and becamewelded together to form the water cooling jacket. Four machined copper outlet ports forsingle-piece electroformed shell surrounded by two electroformed half shells which are EBGeV electron storage ring were made by this process [7]. The cavity consists of an inner,The Daresbury Laboratory 500 MHz single-cell copper cavities shown in Fig. 8 for the 2

Difficulties in brazing and welding electroformed copper.Unless made in two halves each cavity requires a formerRelatively slow rates of deposition (30—50 um/h)Tends to be expensive

The disadvantages are:

Cooling ducts can be incorporated into the walls.Single piece cavities can be made without brazed or welded jointsDimensions of the mandrel are reproduced exactlyAbility to produce complicated shapesHigh purity and good characteristics of deposited material

The main advantages of this technique are:aluminium.subsequently removed in the case of steel for example or dissolved out in the case of

384 OCR Output

Fig. 9 Schematic assembly for magnetron sputtering

GAS tnaectton

puupnnc svstsu

stA|ni.Ess TUBE

cAtH0¤E

niosnuu

Euactnomncnst

Faison r=L0w

pnson input

coppzp RF cAv1tv

csmuvnc |nsui.At0n

1 ,', M0vEABLE wmcnst suppont

Faeon output Fneon input

in Fig. 9. The time to deposit 1 um of niobium was l h.

by this technique [8]. A schematic drawing of the set-up for magnetron sputtering is shownTwo four-cell 352 Ml-Iz superconducting cavities for SPS/LEP injector operation were made

fields to sputter off niobium atoms which condense on the cavity walls.and using the impact of the positive argon ions driven onto niobium cathodes by electricThis method consists of producing an argon gas plasma at a pressure of about 5 x 10*2 Torr

inclusions at the surface and lower cost.has the advantages of higher thermal stability, elimination of macroscopic resistiveThe alternative solution in which a l um Nb layer is sputtered onto OFHC substrate cavities

exceeding the critical temperature due to local heating at surface defects.at liquid helium temperatures which is often insufficient to prevent the inner surface fromand welding. This type of cavity suffers from the relatively poor thermal conductivity of Nb

Traditionally superconducting cavities are made of high purity Nb sheets by spinning

3 . 9 Sputtering

partially formed wall and then plating over it to complete the process.Cooling channels can also be built into the cavity walls by putting wax beading onto the

385 OCR Output

Some recommended joint geometries are shown in Fig. ll. The brazed joint should beThe size of the gap in which the braze flows is typically a few hundredths of a millimetre.

given in Appendix 1.Recommended cleaning procedures for OFHC copper and stainless steel components arethe surface and this is only possible after elimination of grease and oxide films.In order to wet well, a molten metal must be capable of alloying with some of the metal on

former, flows by capillary action through a defmed gap and solidifies on cooling.metal or alloy, having a lower melting point than the base materials, melts and wets theBrazing is a joining technique in which, through heating to a certain temperature, a filler

two-mile linac was made this way.Brazing is one of the most common ways of assembling cavities — as an example the SLAC

4.1 Brazing

4 . J OINING TECHNIQUES

point of view of formability compared to techniques where the force is applied slowly.Cornell [4] that the high stress rate during explosive forming was disadvantageous from thestorage cavities in this way. For niobium however it was concluded from tests made atby the Societe Nationale des Poudres et Explosives in Paris to fabricate the LEP 352 MHzThe technique is apparently suitable for copper components - an offer was made to CERN

Particularly suited to one-off large dimension components.Low tooling costsConservation of intemal surface iinishExcellent dimensional conformity to the dieAbility to exceed the limits of conventional forming processes

The advantages claimed for this process are:

Fig. l0 Explosive forming

me

msc

FLANGE 1-/·\·-§|| fi;"` "

cuupme" "` ""°"°:`>*

·—·me-..$’$'ii“?i'"‘ ··i. Hcc. \ é

blanks into extemal dies.by detonating an explosive charge. The force of the wave is used to deform sheet metalcavities. The principle is shown in Fig. 10. A shock wave is created in an underwater vessel

Explosion forming is a little known and, to-date, unused technique for fabricating

386 OCR Output

the same temperature without melting the previous brazings.is changed (alloying with the base metal) it is even possible to make more than one braze atbe made on the same component. In some circumstances where the composition of the fillerSuitable choice of filler alloys allows multiple brazes at successively reduced temperatures to

metalised ceramic/metalCu/SS (Ni plated)79072Ag/`28CuCu/Cu

Cu/SS (Ni plated

81768Ag/27Cu/5Pd Cu/Cu

Cu/SS (Ni plated

86059Ag/31Cu/10Pd Cu/Cu

8652OAg/60AuQOCu Cu/Cu

91080Au/20Cu metalised ceramic/Nb

96080Au/20Cu SS/Nb

1065 SS/SS76Ni/14Cr/10P

990 Cu/Cu50Au/5OCu

1030 Cu/Cu35Au/65Cu

1175 SS/SS92.5N i/4.5Si/3B

FH..LER ALLOY TYPICAL BRAZE TEMP. (°C) I BASE METALS

Commonly used brazin g alloysTable 3

eventually form part of the cavity vacuum surface.which have high vapour pressures should not be used to braze components which willin Table 3 for the base metals indicated. Filler alloys containing metals such as Zn andSome of the commonly used brazin g alloys and their brazing temperatures are given

Fig. ll Some recommended joint geometries for brazing

m ate r I al

F i I I e r

be on the outside.cylindrical components the material with the higher coefficient of expansion should thereforedesigned if possible such that on cooling it is under compression rather than in tension. For

387 OCR Output

Fig. 12. A typical heating cycle for brazing

At; = 1 min - 40 minAtl = 10 min - 2 h

(dT/dt)2=10O—40O°C/h

(dT/dt)1=20O— 500°C/h

Ah Ar; TIM E1-—-—·—» 4-•·

FREE COOLING

( d T/ d Y) 1

( d T/ d 1 ) 2

be heated. A typical heating cycle is shown in Fig. 12.The heating rates and temperature stabilization times are chosen according to the masses to

material mismatches.subsequent contraction of the brazed joint due to differential heating and cooling rates, andExtemal alignment or support devices tend to create constraints on the free expansion andWhenever possible the components to be brazed should be se1f—aligning and seltisupporting

Possible reduction of strength and grain growth due to relatively high temperatures.Disadvantages:

Multiple joints can be made in one operation.Dissimilar materials can be joined (metals to ceramics)Little or no distortion of the components

Advantages of brazing:

materials due to its strong reducing properties.more rapid and uniform heating of the components but can cause embrittlement in somesubsequent outgassing rate is significantly reduced. A hydrogen atmosphere produces aelevated temperatures), and the hydrogen content of the base materials and hence theiratmosphere of hydrogen. Under vacuum, surface contaminants are removed (at least atThe brazing process is usually carried out either under vacuum (10*0 Torr minimum) or in an

388 OCR Output

for tuning is necessary.for this cold joining method is that the overall geometry is so well defined that no allowanceelectroplating an extemal envelope of copper as shown in Fig. 14. The advantage claimedcylinders of the disk-loaded waveguide structure were stacked together and consolidated byGeV electron linac of the Photon Factory at KEK [10]. The diamond-machined disks and

This technique was used to make the ttavellin g wave accelerating sections for the 2.5

4 . 3 Joining by electroplating

outside. Q-values of 95% of theoretical have been obtained using this technique.cycle under slight pressure. A similar ring at the outer surface prevents braze leakage to theclean mirror—f1nish contacting surfaces (Ra = 0.025 mm) is made during the normal brazingelectrical contact and blocks the flow of the brazing alloy. The diffusion bond between theA 1 mm wide diffusion-bonded annular surface at the inside edge of the discs provides

Fig. 13 Assembly of CLIC 30 GHz cups by brazin g and diffusion bonding

MATERIAL OUT OF CAVITY

BlDCKSFLOWOFE(CESS BRAZE

OUTER DIFFUSION BONDED SURFMI

ALLOY FOR BRAZING

WATER CHANNELS USED T0 HOUSE

PARTIALLY DRILLED COOLING

CAVITY

OF EXCESS BRAZE MATERIAL INTO

SURFACE AND BLOCKS FLOW \ I VQmovmss ELECTRICAL corrmcr AT INNERCRE/\TEDD\JRIT·K3BRAZINGCY(].E K 'Z XDIITUSION BONDED SURFACEMIRROR FINISH COPPERICOPPER

excess braze material into the very small 30 GHz cells. The scheme is shown in Fig. 13.together, a direct copper-to-copper diffusion bond has been used to prevent the flow ofthe CLIC prototype accelerating sections [9]. Although the cups for the sections are brazedon copper surfaces without the silver. This observation has been used in the fabrication ofIf the contacting surfaces are very clean and very flat a diffusion bond can even be obtained

formed by diffusion of the silver into the copper.together for 1 h with a pressure of 1.3 daN/mm2 at a temperature of 350 °C. The bond isThe surfaces to be joined were first coated with a 6 ttm layer of silver and then pressed

wave accelerating sections of the LH. (LEP Injector Linac).This technique has been used to assemble the copper cups that form the travelling

4 . 2 Diffusion bonding

389 OCR Output

vacuum IS necessary.

The main disadvantages of EB welding is that in situ welding is not possible because a

Good reproducibility (fully automatic process)Can weld dissimilar materials (S S/Cu for example)Chemically pure welds due to the vacuum environmentHigh penetration single pass welds at high speedsWelding almost inaccessible places (line of sight only needed)Low total heat input resulting in virtually no distortionHigh quality welds with good joint strengthsExtremely narrow fusion and heat affected zone

Advantages of EB welding:

workpiece — enough to weld 200 mm of steel - with spot sizes down to 0.2 mm.A typical large EB machine using 150 keV electrons can produce 75kW of heat energy at the

as the electron beam is moved along the joint.rather than by melting. The vapour filled hole has liquid walls which close in on themselvesthe electron beam is so high that the it pierces the material being welded by vaporisationpenetrate for a distance of a few atomic diameters into solid material, the energy density ofwhich gives up its energy to the workpiece in the form of heat. Although electrons only

The welding agent for this technique is a finely focused beam of high-speed electrons

4 . 4 Electron-beam (EB) welding

Fig. 14 Joining by electro-plating

EX'l'ERNAL ENVELOPE OF ELECTRO-PLATED METAL

VACUUM DUCT

COOLING WATER

DISK

I cvt.m¤mcAi. sncsn ( tjWATER INPUT

390 OCR Output

Fig. 16 Geometry of EB weld for LEP 352 NH12 accelerating cavities

(a) (bl

1.0 mm

5 I LOCAL POROSITY

sroppan &

Typical dimensions of the EB weld are shown in Fig. 16 (b).edge of the cavity. Weld shrinkages of about 0.5 mm were measured for this geometry.weld a very superficial cosmetic pass is necessary to produce a smooth surface at the outerthird of full penetration) to minimise distortions and then welded. After the full penetrationis kept safely away from vacuum surfaces. The pieces were first tack welded together (oneFig. 16) also ensures that the small zone of porosity present at the root of copper EB weldsvery rough and irregular surface the weld is fired into a "stopper". This stopper (seeSince full penetration EB welds in copper tend to splash through into the cavity leaving ageometry of the joint between the disks and cylinders of the cavities is shown in Fig. 16 (a).The 352 MHz copper cavities for the LEP main ring were assembled by this technique. The

Fig. 15 EB butt-weld geometries for cavity work

PLAIN BUTT LAP BUTT

cavity work.Plain butt-weld or lap butt-weld geometries as shown in Fig. 15 are usually preferred for

391 OCR Output

analysed in detail in Ref. [12]. It was found in PETRA for example that the operationalfrom the temperature-dependent variation in electrical conductivity. This problem has beenthat can be tolerated without creating an unacceptable loss in shunt impedance that resultsamount of heat which has to be carried away, but also the increase of surface temperature

The design of the cooling system for RF cavities must take into account not only the total

5 . COOLING

diameter. For later constructions this technique was abandoned in favour of brazing.waveguide tube. The resulting interference tit at room temperature was about 0.5% of theand after cooling to liquid nitrogen temperature were put into the steam heated outer[1 l]. The discs of the disk-loaded waveguide structure were mounted on an invar mandrel

This method was used at Stanford to construct the original 1 GeV (MARK 3) accelerator

4 . 6 Shrinking

together by this technique.The disks and cylinders of the 353 MHz aluminium cavities for PEP at SLAC were joined

Dissimilar metals cannot be welded.Dcformations in thick components can be largeLarge heat affected zoneMultiple passes required for thick componentsPoor penetration power

Disadvantages:

Process can be fully automated.Ability to do in situ weldsLow cost processHigh quality welds with good joint strengths

Advantages of TIG or Plasma welding:

the weld area from the normal oxidising atmosphere.flowing around it to produce a constriction of the arc. Argon or helium gas is used to shielduse a non-consumable tungsten electrode which in the plasma process has a high—speed gasTIG (Tungsten Inert Gas) and PLASMA welding are normally preferred. Both techniquesworkpiece and an electrode using a constant or pulsed voltage supply. For high quality work

In this process the heat for welding is produced by striking an arc between the

4 . 5 Inert-gas arc welding

done at CERN for 352 MHz niobium cavities.surfaces, EB welding can, when space permits, be made from the inside. This has beenIn superconducting cavities where it is very important to have smooth defect—free internal

200 MHz cavities for injection into LEP.the EB weld be made last to avoid cracking the weld by thermal stress as occured in the SPSFor constructions requiring both EB welding and brazing operations it is recommended that

= l.3Weld thickness= 560Working distance d= IO mm/sWelding speed= 58Welding current

kVGun voltageTypical welding parameters for this weld are:

392 OCR Output

pressures.

some form of channeling to be effective and submits the inner shell to much higher extemalJacket cooling is also used (see for example the Daresbury electroformed cavity) but requires

can be created by an interconnecting series of drilled holes as shown in Fig. 18.For more solid pieces such as forged discs or nose-cone blocks, an intemal cooling channel

Fig. 17 Typical surface heat transfer coefficients for water at 25 UC

WATER SPEED ( m / s )

0.2

0.4

0.6

D=20mm0.8

D=15mm

D=10mm

1.0D= 5mm

1.2

L4 D= lmm

1.6

NN., = 0.023 NP. 0-4 NRC 0-8

given in Fig. 17 using the empirical relationship given in Appendix 2Typical values for surface heat transfer coefficients for water at 25 OC in round pipes areFor effective cooling the water flow should be turbulent (Reynold's number > 2300).

CERN by metal spraying.pre-stripper tank [14]. Copper pipes have also been attached to prototype copper cavities atheat-conducting paste was used at the Lawrence Radiation Laboratory in Berkeley to cool ainterference clamping by shrinking of a pre-formed array of pipes onto the cavity wall via asoldering (example - SPS LEP single-cell cavities). The very unorthodox technique ofwhich are tixw to the external surface by brazing (example - LEP storage cavities) or by softNormal conducting sheet metal cavities are usually cooled by water flowing through pipes

The basic theory goveming thermal flow in water-cooled solids is given in Appendix 2.

effects [13].shunt impedance was 18.9% lowcr than thc assumed room temperature value due to thermal

393 OCR Output

[15] A. Klinkenberg and H.H. Mooy, Chem. Eng. Progr., 44, 17-36, (1948).

[14] J.M. Haughian and R.M. Main, IEEE NS 16 (1969), 3, 395.

[13] G. Gaede et al., DESY M-80/08 (1980).

H. Henke and I. Wilson, CERN ISR-RF·GE/81-25 (1981).[12]

[1]] M. Chodorow et al.,Rev. Sci. Instr. 26, (1955) 134.

Photon Factory Activity Report 1983/84, KEK (1984).[10]

[9] I. Wilson et al., CERN SL/90-84 (1990).

C. Benvenuti et al., CERN LEP-VA/88-35 (1988).[8]

[7] A.R. Wilson and W. Quartz, Engineering, Nov. (1978) 1214.

[6] D. Boehne, IEEE NS16 (1969), 3, 390.

[5] S. Dujardin etal., CERN MT-ES/90-7 (1990).

Ithaca: Comell Univ., Nov. (1987).[4] J .L. Kirchgessner, Forming and welding of niobium for superconducting cavities,

[3] K.W. Shepard, Proc. Second Workshop on RF Superconductivity, CERN, (1984) 9.

[2] I. Wilson et al., CERN LEP-RF/88-50 (1988).

[1] M.A. Allen et al., SLAC-PUB-1898 (1977).

REFERENCES

Fig. 18 Intcrconnccting drillings for cooling water in LEP 352 MHz accelerating cavities

L2 S6 m m

1r.,2

s0”

sm `HI !

%a°

I 0uT

69mm

394 OCR Output

Dry in hot airRinse in cold demineralised water

CERNand used in a concentration of 20 g/l of demineralised water has proved very effective atThe detergent P3 VR 580 17 supplied by Henkel and Co. SA (Pratteln, Switzerland)Ultrasonic clean in an alkaline detergent solution (pH = 9.7) at 60 °C.Degrcase in perchloroethylene

RECOMMENDED CLEANING PROCESS FOR STAINLESS STEEL

steps 4-6 should be omitted.In order not to deteriorate the very high surface finish on diamond—machined components

Air dry after rinsing in hot demineralised water or alcohol at room temperatureRinse in cold demineralised wateraciPagivate in a solution containing 80 g/1 of chromium trioxide and 3 cc/1 of sulphuricRinse in cold demineralised waterdepending on the thickness of the oxide layer to be removedDeoxidise in a 50% by volume hydrochloric acid bath at room temperature for 5-15 minsRinse in cold demineralised waterconcentration of 10 g/l of demineralised water has proved very effective at CERNThe detergent NGL 17.40 supplied by Cleaning Technology SA (Geneva) and used in aUltrasonic clean in an alkaline dctcrgcnt solution (pH = 9.7) at 60°C for 5-10 mins.Dcgrcasc in pcrchlorocthylcnc

RECOMMENDED CLEANING PROCESS FOR OFHC COPPER

APPENDIX 1

395 OCR Output

than the bulk temperature. This must of course be the case for any efficient transfer of heat.correspond to high-temperature gradients, the surface temperature is considerably higherlayer the flow is turbulent with only small velocity gradients. Since high-velocity gradientspipe in which the flow is laminar and in which the velocity increases rapidly. Outside thisDuring turbulent flow, despite its name, a very thin surface layer is created at the wall of the

In this particular formulation all physical properties are evaluated at the bulk temperature.

Ngc Reynolds number.Npr Prandtl number

where NND Nusselt number

NNU = 0.023 Np, 0-4 NRC 0-8

successful:normally chooses a turbulent flow for which the following empirical relationship has provedwell—established turbulent flow can be assumed. In order to obtain good cooling oneNR., < 2300 laminar flow exists over the entire cross—section, whereas for NRC > 10000 ais determined by the Reynolds number NRC. For forced flow of water in round tubes andboundary; this in tum depends on whether the flow is laminar or turbulent. The flow regimeThe value of h is influenced considerably by the velocity profile of the fluid close to the solid

tubes of diameter D is NNU = h D / k.dimensionless Nusselt number which for flow of water of thermal conductivity lc in round'l`he surface heat transfer coefficient h is usually found in the literature hidden in the

T2 usually the bulk fluid temperature.T1 temperature of the wallh surface heat transfer coefficient

where q heat flow per unit area

Q = h (T1 - T2)

relationship proposed by Newton in 1701 is used:special case of heat transfer between a moving fluid and a solid surface the following simpletransfer analysis are given in Ref. [15] together with their physical significances. In thecertain logical grouping of the variables. Most of the groups encountered in convective heatexpressing the empirical relationships in terms of dimensionless groups which establish adata covering an equally vast range of variables. This data is in general rationalised bydoes not exist and this has necessarily led to the accumulation of vast amounts of empiricaldescription. A differential equation describing the law of convective heat transfer thereforecomplicated mixing motion of fluids for which there is no complete mathematicalForced and natural convection on the other hand involve the exchange of heat due to theexactly by the mathematical laws of Fourier (1822) and Boltzmann (1884) respectively.Steady state conduction and radiation present no difficulties since they can be expressed

(iii) by natural convection and radiation to the atmosphere.(ii) by forced convection to the bulk of the cooling water, and(i) by conduction tluoughout the cavity,

cavity walls is transportedThe heat that is dissipated within the very thin surface layer at the inside surface of the

THERMAL FLOW IN WATER-COOLED SOLIDS

APPENDIX 2