306 PHILIPS TECHNICAL REVIEW Vol. 2, No. 10

was developed, in which the intermediate-frequencyamplifier with the second detector and the low-'

axis of the paraboloid, so that as little shadowas possible was caused.

Fig. IS. Photograph of the long receiver in the paraboloid.

frequency amplifier were mounted end to end, andthe whole combined with the mixing stage and thefeeder to give a unit which was set up along the

In fig. 14 may be seen a picture of the longreceiver and infig. 15 a photograph of aerial systemused at the receiving end.

JOINTS BETWEEN METAL AND GLASS

by H. J. MEERKAMP VAN EMBDEN.

Summary. A survey is given of metal-glass joints and the demands made upon themin practice. The application of chrome iron to glass seals (particularly in gasdischarge tubes) is discussed in some detail.

Introduction

Since Plücker and Geissier carried outtheir ex-periments with the forerunners of the modern neontubes, gas discharge tubes have passed through avast development, and are not only used forillumination but are finding a continually widerapplication III the field of radio and heavycurrent technique. Although originally the mainobj ect in using Gei s si e r tubes was the study ofinteresting phenomena, in connection with tech-nical applications it is now often important to beable to dissipate as great an electrical power as pos-sible within a tube of given dimensions. This power,as in the case of practically all electrical machines, islimited by the heating permissible in relation to theconstruction and kind of material. Examples ofheavily loaded discharge tubes have been describedrepeatedly in this periodical, some of these arethe metal rectifier with mercury cathode 1), thehigh pressure mercury lamp 2), the water-cooledtransmitter valves 3), etc.

1) A new metal valve with a mercury cathode, Philips Techn.Rev. 1, 65, (1936).

2) The mercury lamp HP 300, Philips Techn. Rev. 1, 129(1936 ).

3) The development and manufacture of the moderntransmitter valves, Philips Techn. Rev. 2, 122, (1937).

The choice of material for heavily loaded. tubes

The endeavour to attain maximum power in thetube leads to new aspects in the choice ofmaterial.The wall of a discharge tube must be competely

impervious and the current leads must pass throughthis wall without the possibility of gas leakage ineither direction.At the same time the wall must be of material

suitable to insulate the electrodes from each other.But in addition to this the wall fulfills still anotherfunction: if the tube is used for illumination pur-poses, it must transmit the radiation, while itmust give up to its surroundings by conductionand radiation the energy freed as heat in the tube.Because of this the use of glass-like substancesfor thc walls of the tube is indicated.

In the technical development of dischargetubes, as is clear from the above, limitations areencountered especiallywith regard to the last point.Upon increasing the power the energy dissipatedas heat increases also and the wall becomessteadily hotter.When glass apparatus is used the danger of

softening or cracking of the glass with great heatis considerable, and if an attempt is made to

OCTOBER 1937 JOINTS BETWEEN METAL AND GLASS 307

decrease the amount of energy given off per unitof surface by making the tubes larger, the 'limitof the technically possible' is soon reached, sincethe implosion of large evacuated containers is verydangerous.

10-620

18

16

14

12

10

8

6

42o

FeD 10 20 30 40 50 60 10 80 90 100%Ni

2S6~

)

\..,......... ~/

1/

\ 1v

Fig. 1. Coefficient of expansion of nickel iron at 20° C as afunction of the composition. With ahout 36% Ni (invar) thecoefficient of expansion has a sharp minimum. A coefficientof 100 X 10-7rC, thus equal to that of ordinary glass, is

. reached with about 50% Ni. .Int, Crit. Tables 2, 465, 1927.

It is obvious that a better' material must besought for heavily loaded tubes. For high pressuremercury lamps the use of quartz, for example,which softens only at higher temperatures thanglass, is an important step. In general, however,glass seems to be an outstandingly suitable material,which should only be avoided at those points wherethe greatest heating effect occurs. If, for instance,one could substitute a inetal for the glass at thepoints of greatest heat development, the dangerof cracking and breaking would be avoided, whilethe removal of heat could be made much more rapidby the use. of a water jacket. Moreover the

10-620. 18

16

1412

fO

8

6

42oo

.25656

/ /V

5O%Ni j //

~'/0\0~/

I--600 800°C2fX) , 400

Fig. 2. Coefficient of expansion of nickel iron as a functionof the temperature. a) 36010 Ni. The coefficient of expansionremains very low up to 300°C. b) 50 "!«Ni. The coefficientof expansion is uniform up to 500°C. Int. Grit. Tahles 2,465, 1927.

accuracy with which metal can be worked is muchgreater than can be obtained with glass, so thatthe construction and installation of a partly metaltube is much simpler. The sealing of glass to a metalis not so simple however. Due to the difference incoefficients of expa~sion. the glass usually breaksupon cooling when it is fused to a metal.

In order to prevent this there are two possibilities:1) make the metal so 'thin, that upon cooling it

may be plastieally deformed without causingthe glass to break, .

2) find a metal whose expansion corresponds withthat of the glass 4) ..Both methods are employed; for the first copper

(linear expansion coëfficient hetween 0 and 300 oe,average 168 X 10-7 î" C) is usually used as the metal.It must be very thin (about 0.1 mm) in order to bewelded to glass. The procedure is, however, quitedifficult and expensive, due to the tendency of thecopper to be oxidized, while 'great strength cannotbe expected of the thin material near the weld.

%010

.25657

~., //I

: .~

I~

\:;: //-:V

./ /~

7~ V',~ °l~

1/ t77 -V

9

8

7

6

5

4

3

2

1

oo 600

Fig. 3. Change in length as a function of the temperature .a) with 36 "!« Ni. b)with 50 "l« Ni. Int. Crit. Tables 2, 465,1927.

For the second method, the use of a metal adaptedto the expansion of glass, we shall first discuss the.requirements which the metal must satisfy. Theyare the following:1) Uniform expansion throughout a long temper- ,

ature range, from room temperature to above •the softening point of the glass.

4) This does not necessarily mean that the coefficients of "expansion must be exactly the same. Conversely" withsimilar coefficients of expansion, differences in expansionmay occur, because of the fact that the metal and theglass cool off at different rates. In this connection it isdesirable that the metal does not have too great a heatconduction. -

308 PHILlPS TECHNICAL REVIEW Vol. 2, No. 10

2) The metal must be sufficiently stable duringthe sealing-in.

3) The adhesion to the glass must be good.oe

180025658

1400

S

S+G( ......- I---.....~ t.... ~ "I-~

f\.~ 0(r \\1\ ~

~ ~ ~ot+l I \

11\

~ ~ ó~ ~ ~+ +:IS lS

1600

1200

1000

800

600

"lOO

200

oFe 0 10 20 JO 40 50 60 70 80 90 100%Cr

Fig. 4. The iron-chromium diagram. Taken from: Ed.Houdremont, Einführung in die SonderstahJkunde, Springer1935, p. 182.

4) The amount of gas given off by the metalmust be small.

5) The price must be reasonable.The significanee of these requirements will be

briefly explained in the following.

quently a metal which is suitable for sealing into normal calcium glass (coefficient of expansionabout 100 X 10-7;0 C), is not suitahle for sealing

Fig. 5. Example illustrating the possibility of making thickchrome iron-glass joints.

into hard glass with a coefficient of expansion of40 to 50 X 10-7;0 C. The metal to be sealed inmust possess uniformity of expansion over a widerange, from cold to the softening point.If allotropie transitions occur in a metal upon

passing certain temperatures, for example a tran-sition from a to !~-structure, the transitions are

Fig. 6. Chrome iron tube with glassbulb (for transmitting valve) on the sealing-inmachine.

1) The expansion of glass is uniform from roomtemperature to the softening point, but vanesconsiderably for different kinds of glass. Conse-

usually accompanied by volume changes. Stressesin the metal occur which may cause cracking ofthe glass. It is therefore necessary to choose a

OCTOBER 1937 JOINTS BETWEEN METAL AND GLASS 309

metal or an alloy which exhibits no allotropietransitions in the temperature range in which theglass is hard.2) The sealing-in to the glass takes place at

high temperatures at which glass flows; at thesetemperatures the metal must not yet he soft andmust not burn.3) The seal may be sufficient in some cases

due to adhesion alone. The classic example of thisis the glass-platinum seal. The metal remainsquite bright in welding and provides an excellentseal. In the case of most metals and alloys, however,

Fig. 7. A water-cooled metal valve with a mercury cathode,described in detail in Philips Techn. Rev. 1, 65 (1936).Average current strength 75 A, peak current 500 A. The chro-me iron cylinder is water-cooled. It contains a pool'of mercuryand serves as cathode. The upper half of glass is necessaryfor insulation of the auxiliary anode (upper left) and the mainanode (upper right) which is introduced by means of a chromeiron cap.

this is not the case and the adhesion between metaland glass is not sufficient, while at the same timethe metal becomes covered with an oxide layer uponheating. This film may in some cases serve to pro-mote a good seal when it functions as an interme-diate layer between glass and metal. On one sidethe film is firmly bound to the metal, on the otherit is to some extent dissolved in the glass forminga silicate and produces a gradual transition at this

point. If, however, the oxid~ layer is thick andporous or has the tendency to form loose flakes, it

Fig. 8. The rectifier valve DCG 10/15, example of a morecomplicated construction of glass and chrome iron. Lowerbulb: cathode space, upper bulb: anode space. The thirdbulb serves as a condensation chamber for the mercuryevaporating in the lower bulb. The mercury pressure nearthe anode is thus kept Iow (little back ignition) with a highmercury pressure in the neighbourhood of the cathode (Iowcathode drop). The chrome iron tubes serve as auxiliaryelectrodes. They must be at a positive potential to en-sure ignition. At the same time there is therefore thepossibility of using the rectifier as a relay valve.

is obvious that there can be no question of atight seal.4) During the

the metal should

gas-

sealing, and afterwards too,contain no gasses in the adsorbed

r:

Fig. 9. Anode of a metal transmitting valve with chrome iron-glass seal.

310 PHILIPS TECHNICAL REVIEW Vol. 2, No. 10

or dissolved state, which could be freed duringworking. This would prevent a perfect seal. Atthe interface between glass and metal gas bubbleswould occur which could cause leakage.Also after the seal is finished no more gas may

escape from the metal, since it would spoil thevacuum or the gas atmosphere in the tube.

It has long been known of platinum, 'with alinear coefficient of expansion of 90 X 10-\ thatit gives vacuum-tight seals, and indeed the firstelectric lamps had platinum leads. The price is how-ever so high that attempts were immediately madeto find cheaper alloys, and at present the platinumlead is of interest only in the lahoratory because

of the fact that the sealing-in of such leads is sosimple, the ends of the wires are always bright andductile and the metal is so extraordinarily stablechemically.

Of the nickel-iron alloys invar, with about36 per cent of nickel, is known for its vcry slightcoefficient of expansion. Figs. 1 and 2 give the co-efficients of expansion of Fe-Ni alloys as a functionof the composition and of the temperature. If weheat such alloys we see that the coefficient of ex-pansion between room temperature and 1500 C ispractically equal to zero, but that above thattemperature (fig. 3) we obtain a horizontal lineup to 200 0 which turns sharply upward at 200 o.

If we now add more nickel to the alloy we findthat the horizontal line begins to slope upwardand the point where the slope suddenly changesis displaced to higher temperatures. An alloy withabout 50 per cent of nickel between room tem-perature and 500 0 has a coefficient of expansionequal to that of glass, above the latter temperatureit is somewhat higher. This is however not danger-ous, since at those higher temperatures the glassis soft and therefore will not crack.

With regard to expansion therefore the latteralloy is a suitable one, and is in fact very often usedas sealing-in wire for lamps but it has howeverseveral disadvantages particularly with regardto adhesion. Upon sealing it in, the wire isstrongly oxidized and the oxide has a greattendency to flake off, while in addition it is very

Fig. 10. The heavy anode of a "Metalix" X-Ray tube is sealed to the glass with a chromeiron ring.

5) The high costs in many cases prevent theuse of a noble metal such as platinum.

Let us now list the metals which may be con-sidered suitable for sealing to glass. For ordinaryglass with a coefficient of expansion of 85 to 100 X

10-7;0 C we find the following possibilities:

platinumnickel ironchrome iron

For hard glass with a lower coefficient of expan-sion there are other suitable metals and alloys whichcorrespond in expansion with hard glass. A discussionof these would lead us too far afield. Use is usuallymade of ternary iron-nickel-cobalt alloys and alsopure tungsten and molybdenum.

Fig. ll. The middle section of a "Metalix" X-Ray tuhe ismade of chrome iron for screening- off undesired radiation.The X-rays are emitted exclusively through thc glass window.

,OCTOBER 1937 JOINTS BETWEEN METAL AND GLASS 311

difficult to make this material gasfree. Nickelabsorbs all kinds of gasscs very easily (its use asa catalyst is an example of the application of thisproperty), and these gases are easily given offagain. That is the reason why the sealing-inwires are often covered with a thin layer of copper

Fig. 12. Simple lead-in in a rectifier, consisting of a chromeiron plate with terminals soldered on to both sides.

(borated copper), since the adhesion of glass to cop-per by means of the film of copper oxide is very firm.Since copper has a much higher coefficient of ex-pansion, the copper coating must be very thinso that it may be plastically deformed withoutcracking the glass. For larger seals nickel ironis out of the question.

On the other hand for large seals chromeiron has found extended application. Before weconsider its application we shall first study the iron-chromium system. The diagram of the iron-chromiumsystem is given in fig. 4. When pure iron is heated,it passes over into another modification, y-iron,at 906°C. y-iron has a different atomic arrangementthan the a-iron 5) stable below this temperature.

Upon further heating y-iron passes over at1401° C into a-iron again which is identical with thea-non stable below 906°C. a-iron melts at 1528 °C.Upon cooling the transitions occur at the samepoints, solid at 1528°, a-iron to 1401°, y-iron to906° and a-iron lower. If we add chromium as al-loying element we find that the two transitionpoints approach each other with increasing chro-mium content; with about 15 per cent of chromiumthey coincide. An alloy with more than 15 per cent

5) The atomic arrangement for a and y-structure is shownin Philips Techn. Rev. 2, 253, (1937).

chromium and the rest iron consists therefore onlyof a-crystals between room temperature and themelting point.That means that the expansion due to change

of temperature proceeds quite uniformly withoutthe transitions a ....,.y which are accompaniedby large changes in volume. Moreover the coef-ficient of expansion is 100 X 10-7 and correspondstherefore very well with that of glass.

TechnJcal procedure in sealing

The alloy is melted in an electric furnace for whichthe electric are furnace as well as the more modernhigh frequency furnace may be used. It is thenpoured into casting moulds and forms ingots.The ingots are now rolled into rods or sheets andthe pieces intended for sealing in to glass are madefrom these by turning, drawing, etc. Thin sheetcan very well be drawn, care must he taken how-ever in rolling that the material is not given ananisotropic texture by cold working, since thiswould give bad results in the drawing 6).

Chrome iron is fairly resistant to air. This stabilityis ascribed to the formation of an extremely thinfilm of oxide which entirely covers the surface andprotects it from further attack by the atmosphere.Various chromium alloys are on the market asstainless steel. In our case the oxide film forms a

Fig. 13. Three electrodes of a rectifier valve are led in bymeans of sector-shaped chrome iron plates. The three platesare first welded together with a glass edge to give a circularplate and then sealed on to the circular opening of therectifier bulb.

6) See Philips Techn. Rev. 2, 156, (1937). for the drawingof chrome iron cups (Example 23 for X-Ray testing ofmaterials ).

312 PHILlPS TECHNICAL REVIEW Vol. 2, No. 10

connecting substance between glass and metal, andit is obtained in the proper thickness for goodadhesion by heating the chrome iron with the flamedirectly before the sealing into the glass.Fig. 5 shows that even thick pieces of glass and

chrome iron may adhere very well to each other.The fact that such a sealing-in may be carriedout with chrome iron is due in part to the fact thatthe heat conduction of this alloy is very small.For sealing large pieces a special lathe has been

made (.fig. 7) with two ch u ck s opposite eachother. In one ch u ck the chrome iron in tubeform is held, in the other the glass part. While

Fig. 14. The use of chrome iron anodes in neon tubes leadsto a very simple construction.

both are turning the edge of the chrome iron isfirst heated and oxidized, after which a thin layerof glass is deposited by hand upon the hot edge byapplying a thin glass rod to it. After that the edgeof the piece of glass to be welded is heated and thetwo chucks are brought together, still rotating.The glass seal is then obtained as an ordinaryglass butt weld by fusing. The edge of the chromeiron tube in the above described process need notbe sharp or even thin, a fact which is favourableto the strength of the glass-metal seal. In thisrespect chrome iron is distinct from all otherordinary sealing-in alloys which can only be sealedinto glass in the form of thin, sharply bevelededges. Figs. 7 and 8 show two important examples

of the glass-chrome iron butt weld in the construc-tion of large rectifiers. In the case of the very largewater-cooled transmitting valves the sealing ringon which the glass is fused is made of chrome iron(jig. 9). This ring is soldered to a copper tube whichis water-cooled. The heat conduction of copper ismuch better than that of chrome iron, so that moreheat can be conducted away in this manner. Thesame principle is used in the case of the anodesof X-ray tubes (jig. 10), where a very high heatconduction is required. The anode therefore con-sists of a block of copper which is fused to one edgeof a chrome iron ring, the other end of which isthen joined to the glass.

The middle sections of the Philips "Metalix"X -ray tubes are also made of chrome iron (jig. 11)and a window for transmitting the beam of raysis sealed in. This would not be so simple with metalsother than chrome iron. The tube is mountedin the apparatus by this metal middle section.If it is desired to make current lead wires,

for very heavy currents one may take a chromeiron cap with a turned edge on which a glass tubecan be sealed in the manner described. The capis then provided with a lead wire in the middle ofthe inner side which is soldered to the cap, and witha second wire on the outer side. Afterwards thewhole can be sealed into the glass tube.

For smaller current strengths, for example in therectifier given in jig. 12, a chrome iron plate issufficient, which is provided on both sides withglass. It is then sealed into the glass tube. Severalsuch plates may be "joined to a foot", whichis sealed into the rectifier valve as a whole afterassembly. Fig. 13 gives an example of this.Another example may be seen in the electrodes

of a neon tube (jig. 14) in which a light cup ofchrome iron which serves as electrode is sealedat its edge directly to the end of the glass tubeso that the external current supply can takeplace directly through the base.

For additional technical applications referenceis made to the possibility of introducing glasswindows into pressure or vacuum boilers, and tothe use of glass connecting pieces on metal reactionchambers, while the special application for telescopemirrors 7) has already been mentioned.

7) Telescope mirrors, Philips Tech. Rev. 1, 358 (1936).

OCTOBER 1937 313

Illumination of the Eiffel-tower at the Paris World Exhibition (1937). The greater part ofthe electrical plant has been equipped by, or with materialof, S. A. Philips - Paris.