116 VOL. 11, No. 4PHILIPS TECHNIC4L R)j;VIEW, \

CONSTRUCTIÓNAND APPLICATIONS OF A NEW DESIGN OF ~THEPHILIPS VACUUM GAUGE

'.... '

by F. M. PENNING and K. NIENHUIS 531.7~8.732

The Philips vacuum gauge, with which the pressure ofa gas is measured by means of theintensity of a gas discharge in a magnetic field, was introduced 12 year§ ago and has nowfound application in many fields and proved to be a most useful instrument. Meaitwhile theneed has also beenfelt for such a simple vacuum gáuge for measuring still lotoergas pressures.This need has been met by changing the construction of the present gauge. In its new design:this instrument is also quite useful as leale detector.

A description of the Philips vacuum gauge wasgiven iD. this journal 12 years ago 1). The tubeused in' this gauge contains two parallel plates ofzirconium as cathode and a ring as anode (fig. la)and- is placed between the poles of a permanentmagnet with its field perpendicular to the cathode.Owing to the peculiar course of the electrical and

.~ magnetic lines of force inside the tube the electronstravel to and fro an immense number of timesbefore reaching the anode, thus making many.jnore collisions with gas molecules than would bethe case if they travelled directly from the cathode, to ,the anode. This large number of collisions. calf- he interpreted as ~eing derived from an apparentelevation of the gas pressure.Consequently it is possible for a gas discharge

to take plaée hetween c'old electrodes at a pressureabout 1000 times lower than is possible in a tube ofa similar construction without magnetic field.

. '

II I ,I I

I IAcfH.

58016

Fig. 1. Diagrammatic representation of the position of the'cathode K and the anode A .of the 'Philips vacuum gauge.

• G represents the glass envelope and H the direction of themagnetic field. a) The older design, b) the new design.

1) F. M. Penning, High-vacuumgauges;PhilipsTechn.Rev.2, 201-208, 1937. '

It has been found experimentally 'that the magni-tude of the discharge current is a good measure forthe pressure.

1Mn M

·~s--.--(~~~--)--~====~~ @ 58017

Fig. 2. The circuit diagram of the vacuum gauge M in serieswith the glow lamp E, acting as current Indicator, and amicro-ammeter A. This circuit is applied both in the. olddesign and in .the new one.

This vacuum gauge, the circuit diagram of which-is given infig. 2, has in recent years been extensiv~lyapplied in the. m~mufacture of electronic valvesand in apparatus in which the vacuum has to bemaintained by means of a pump, as is the case. with two kinds of apparatus previously describedin this journal: the 'neutron generator 2) ~nd theelectron microscope 3). It can also he used, asdescribed in literature on the subject" in combin-ation with a ·"Pirahi" manometer 4) or w;ith athermocouple manometer 5), in which c~se l;l largerpressure range is covered. Further, this instrumentis very suitable for starting a certain signal G) oractivating a relay <1) 7) at the moment that thepressure exceeds a certain limit. ' ,

2) F. A. Heyn and A. Bouwers, An,apparatus forthe trans-,mutation of atomic nuclei, Philips Techn. Rev. 6, 46-53,1941.

3), A. C. Dorsten, W. J. Oosterkamp and J. B.le Poole,'An experimental electron microscope for 400 kilovolts,Philips Techn. Rev. 9, 193-201, 1947.

4) N. C. Pi car d , P. C. Smith and S. M. Zollers, A reliablehigh vacuum gauge and control system, Rev. sci. Instr. 17,125-129; 1946. ,

6) R. 1. Garrod'and K. A. Gross, A combined thermocoupleand cold-cathode vacuum gauge, J. sci. Instr. 25, 378-383'-194.8.

6) ~. S. Mackey, Non-linear indicator for vacuum gauge,. "Electronics, Febr. 1948, p. 140. "

7) H. A. Thomas, T. W. Williams and J. A. HippIe,A mass spectrometer type of leak detector, Rev: sci. Instr.17, 368-372,' 1946; Detecting vacuum leaks electronically,Wes~ghouse Engineer, July 1946, p. 108.

OCTOBER 1949 NEW PHILIPS VACUUM GAUGE 117

For such applications the pressure range of 2 X 10-3to 10-5 mm Hg covered by this vacuum gauge isgenerally sufficient, but in the manufacture ofvalves which are exceptionally sensrtrve to gasstill lower pressures have to be measured.

The new vacuum gauge

The new vacuum gauge specially designed forlower pressures has a different electrode system_(fig. lb). The cathode plates are round discs parallel

Fig. 3. Photograph of the new vacuum gauge.

to the axis of the glass tube, whilst the anode ISIII the form of a cylindrical jacket perpendicularthereto. Where the cathode plates are mounted theglass tube is pinched. Thanks to this constructionthe same permanent magnet can be used with thenew vacuum gauge as with the old one, whilst .theelectrical circuit (fig. 2) also remains exactly thesame. Fig. 3 is a photograph of the new gauge.Compared with the old one it has the advantagethat the whole of the discharge space is enveloped bythe electrodes and the discharge can no longer beaffected by the glass wall and the thin layer ofmetal deposited upon it by sputtering of thecathode.

Infig. 4 the current :flowing through the vacuumgauge has been plotted as a function of the gas

pressure for the voltage (2000V) and -serres reSIS-tance (1 megohm) usually applying also with theold type. This curve applies for air 8), its trenddepending more or less upon the nature of the gas.A comparison with the analogous curve (dottedline) for the old gauge (fig. la) shows that thesensitivity of the new vacuum gauge is about tentimes greater, due, inter alia, to the larger effectivesurface area of the cathode. The curve for the newgauge has been drawn as a continuous line down toa pressure of about 10-6 mm Hg, but actually thedischarge continues to take place at still lowerpressures, inwhich range the sensitivity is represent-ed approximately by the broken line; at such lowpressures calibration is very difficult.

In the first description of the Philips vacuumgauge 9) it was observed that when the pressure isreduced there is sometimes a sudden jump in theintensity of the current. With several specimensof the new design such a jump has been observedround about 10-4 mm Hg, thus at the upper limitof the pressure range. Below that pressure therewere only small jumps.

Just as with the old construction, for currentsbetween 10 and 1000 mA a small glowlamp (b.g.Philips 4662) can be used as indicator (see fig. 2),

pA11 3v

/t--

[7101=

't== .-1=

07' =8 -7 -6 -5 tO--4mmHgto 10 10 1058018

Fig. 4. The continuous line represents the current flowingthrough -the new vacuum gauge plotted as a function of thegas pressure for the usual values of voltage and seriesresistance(2000 V, 1 megohm). The probable extension of the curvetowards lower pressures is represented by the broken line;the dotted line relates to the old vacuum gauge.

8) The curve plotted is the average for three different tubes;the greatest deviation was 30%.

9) F. M. Penning, Ein neues Manometer für niedrige Gas-drücke, insbesondere zwischen 10----3und 10--.5mm, Physica,The Hague, 4, 71-75, 1937.

;,

, .

, '

..'~

.... , .

, ,

'':".'

.'

.' ,

PHILlPS TECHNICAL REVIEW VO!,.: 11, No. 4'"118

but for currents below 10 mA 'a more sensrtrvedevice is needed, A light-spot galvanometer withsay 100 scale divisions 'for i (J.A is highly suitablefor this purpose, one scale division correspondingto a difference in pressure of the order of 1O~ to 10,..9,

, . mm Hg. In' order to get results at all reproducible, ',,_in the range below 10-5 mm Hg it is of course neces_..

. sary to degas carefully all glass walls andelectrodes,Including the. vacuum I!:auge tube itself.

Leak detectors

."

" '

..':-

For smaller leaks other meth~ds have. beendevised 10), which were of great importance innuclear research for testing the extensive vacuumapparatus required for separating U235 11). Forthis purpose the apparatus was fitted with ä massspectrograph 7) sensitive especially to helium arid"

58019

Fig. 5. A vacuum gauge (old design), used as leak detector,provided with a window through which the colour of the dis-charge can be seen to change when the leak is sprayed withsome gas other than air. , '

, I

In any pumping 'installation the vacuum gauge:. can also be used for detecting and tracing leaks. '" To find the' origin of a leak in the' object to be

evacuated (for instance a transmitting valve orthe pumping installation itself) one usually goesover the whole apparatus part for part, replacing theatmospheric air' by some other substance. 'When the vacuum was tested by, spraying with thisthe leak is reached this substance flows in and the gas. This method has the great advantage that

,~ ',: gauge usually registers 'a change' in pressure. As helium can never occur as al!:impurity in the Ilppar-substitute one often uses a liquid, for instance I atus itself, such in contrast, for instance, to hydro-water, a~d if the apparatu~is fitted with a liquid gen. In most countries, however, helium: is tooair trap, in which' the penetrating water vapour' 'expensive to be used' for this purposé, whilst onimmediately condenses, the pressure in the apparatus ' ,the other hand the mass speetrograph is a ra~her'will /drop to; a very: low Ievel. . complicated apparatus and the currents have to. The use of it liquid, as substitute 'for air has "be amplified before being measured:its objections. It is not always possible to càrry the Simpler in many respects is: a method where <'

liquid to the suspected places, and moreov~r the advantage is tak~n of the fact .that a heated wall'method becomes very cumbersome when some ,of palladium is only permeable 'for hydrogen_atmospheric air' remains between the liquid and a~d not for ah:. With ihi~ method hydrogen has.the' leak' and first has to he pumped away. It is therefore to be used as substitute for air. The.therefore better to use instead of a' liquid a gas principle of the method is illustrated, in fig· 6:.which can he sprayed upon the suspected places.In recent years this principle has in fact been appliedin various ways. Fo; this' r~ason we sha:ll confine

.. our further considerations to gases as substitutes'fo~ air. The, usual method, whereby a leak is detectedthrough a' difference in t~e pressure registered bythe vacuum gfluge when the suspected place issprayed, ,'0ll he reverted: to farther on, after wehave first 'dealt with some other methods:. One can use as indication, for instance, the changein colour of a gas' discharge. Hitherto this dischargehàs usually heen brought about' by means of aTesla coil, but this requires a rather high pressure

.' '-3(at least 10 IÏ!Ill Hg). ,At lower pressures thechange in eolou~ of the discharge can be observedin the' Philips v;;tcuum gauge (old design) itselfif'thè: tube is provided with a window far enoughremoved from the discharge to he kept fr_ee.from

.. any sputtered material (see fig. 5). When a leakis sprayed for instance with hydrogen one can clearlysee thé discharge inside the vacuum 'gll;uge change'éolour, .. , " .

\ . ,;~8020

Fig. 6. Principle of a ,Îeak detector with palladium wall.M is the vacuum gauge which is evacuated vin the tube A;sealed by fusing at B and then connected at C to the pump.P is the palladium tube which can be heated by meáns of thefilament F.

the previously evacuated vacuum ~gauge M' (e.g.an ioni:r.ation vacuum gauge) is separated from theapparatus being' tested by the palladium tube P -.which can De heated by means of the' filamènt F;

. . ,

_lO) For a more extensive review see' the recent publication ofS. Dushman, Scientific foundation of high vacuum tech-nique, John Wiley and Sons, New York 1949. ,

11) R. B. J acobs and H. F. Zuhr, New developments in '. vacuum engineering, J. appl. .Phys, 18, 34-48, 1947. "

"

OCTOBER 1949 NEW PHILlPS VACUUM GAUGE 119

as soon as the jet of hydrogen strikes the leakhydrogen passes through P into M and raises thepressure.

The new vacuum gauge as leak detector

Owing to its high degree of sensitivity the newvacuum gauge described above is highly suitablefor detecting smalileaks. One can use; for instance,the arrangement shown in fig. 6, where M thenrepresents the new gauge, which is first evacuatedvia the tube A and then sealed off at B, afterwhich the whole device is connected at C to thepumping installation and the apparatus to bechecked.

permanent magnet and a bulb contammg a palla-dium tube, the latter in this case being heated bythe direct passage of current. The whole device isconnected on the right-hand side to the apparatusbeing tested.

In fig. 8 some examples are given of the variationsin current as a function of time when a leak in anobject (total volume about 2 litres) is exposedalternately to air and to hydrogen. The sprayingwith hydrogen begins where the arrow points up-wards and ceases where the arrow points down-wards. It is seen that the pressure continues torise for some time after the spraying has beenstopped. This i" due to the fact that the leakage

Fig. 7. The new vacuum gauge used as leak detector according to the principle of fig. 6.On the left tbe new gauge with magnet, on the right tbe envelope witb the palladium tubeheated by the direct flow of current via the contacts provided at the top of the envelope.

For tracing a leak the palladium tube is heatedwith the -filament F and current is sent throughthe vacuum gauge. Since the discharge causes gasto disappear, after some time there will he a lowequilibrium pressure, which is not zero becauseas a rule some hydrogen will still be passing throughthe heated palladium owing to hydrogen or hydrogencompounds being present in the pumping instal-lation. To localize the leakage the pump is shutoff by means of a cock and one goes over theoutside of the apparatus with a jet of hydrogen inthe manner described above. As soon as the currentflowing through the vacuum gauge rises this givesan indication that the leak has been found. If thereis any risk of an explosion a mixture of hydrogenwith some other gas can be used, but then of coursethis reduces the sensitivity of the method.

Fig. 7 is an illustration of such a leak detectorconsisting of a vacuum gauge in the field of a

channel is still filled with hydrogen, the pressureceasing to rise when all the hydrogen in this channelhas been replaced by air and diffused throughthe pumping installation.

For these and subsequent measurements the leak wasformed artificially by means of a very narrow glass capillaryopen to the outside air. Such a capillary is made by firstreducing the diameter of the tube in a flame until the airchannel inside is just visible, after which the tube is drawnout further and broken at its narrowest part. The leak can beenlarged by breaking off a piece of the capillary, and thetube can be sealed by holding the extremity in the flame fora moment. The passage is previously measured by allowingair to leak in for a considerable length of time. In this way itis possible to make capillaries passing through less than 10-7

cm" X mm Hg air per second.

The amount of hydrogen passing through thepalladium depends upon the difference in pressure'between the two spaces and thus upon the increase

120 PHILIPS TECHNICAL REVIEW VOL. 11, No. 4

of the pressure in the object. For a given leak thesensitivity of this method is therefore roughlyinversely proportional to the volume of the object.

i

t

t

~

-t.>....--

.c } 3x10-6mm Hgl

)/)

sec

)2XI0-7mmffg/sec

}5x10-7

mmHgjsec

o fat

20 _t(sec) 3056087

Fig. 8. Some examples of the current variation asfunction of time when a leak is alternately exposed to ail'and to hydrogen. Spraying with hydrogen begins at the arrowdirected upwards and ceases at the arrow directed down-wards. The initial current in air for 0 < I < 10 sec is indicatedat an arbitrary height above the z-axis, An idea of the sensiti-vity can be formed from the size of the scale divisions of thecurrent meter given in the diagram. To the right of the curvesthe size of the leak in air is indicated in mm Hg per second.

As already stated, the principle of combining aheated palladium wall with an ionization vacuumgauge with hot cathode has been applied before 12),the pressure being deduced from the ratio of theion current to the electron current. The methoddescribed here is simpler because there is onlyone current to be measured and moreover this isgreater than the non-amplified ion current inthe other method. Another advantage is thatwhen a leak is detected the hydrogen gas thathas penetrated into the vacuum gauge can bemore easily removed, so that one can pass overmore quickly to the next test. In the method witha cold cathode the discharge itself will "burnaway" the gas more quickly than when a hotcathode is used.For the detection of still smaller leaks in valves

which can be sealed off from the pumping device

12) H. Nelson, The hydrogen gauge, an ultrasensitive devicefor location of air leaks in vacuum-device envelopes,Rev. sci. Instr. 16, 273-275, 1945.

another method, without palladium tube, is tobe preferred. A vacuum gauge is fused onto thevalve to be tested and both of these are then evac-uated as well as possible, after which they aresealed off from the pump (see for instance fig. 9).The current through the vacuum gauge is thenswitched on until it reaches a constant value. Asalready remarked, owing to the discharge a certainamount of gas disappears. In the state of equilibriumthe amount burnt away per second by the dischargeis obviously equal to the sum of that leaking inand that released from the walls, etc. If the amountof the gas released is negligible some idea of thesize of the leak can already be formed from thecurrent flowing through the gauge. On a former

58052

Fig. 9. Photograph of a transrruttmg valve with vacuumgauge fused on, by which means a leak was found in one ofthe metal caps for the current lead-ins.

occasion 9) it had been observed that in a welldegased tube about 20 i cm3 X mm Hg air persecond is burnt away, i representing the currentin amperes. In the case just assum~d this is in factthe amount leaking in per second.

In order to localize the leak one can now spraythe valve to be tested with some gas other than air.

OCTOBER 1949 NEW PHILIPS VACUUM GAUGE

The current passing through the gauge will gener-ally depend upon the nature of the gas used, becauseboth the rate at which the gas leaks in and thatat which the burning away takes place are related

" thereto, whilst the sensitivity of the instrument,.also _depen?s "upon the kind of gas. Good results

i (fiA)10

5

2

.:

0.,5

0,05

0,02

0,01o 5

a

, - '

have been reached for instance with argon ,andhydrogen as substitutes. Fig. lOa gives someexamples of the results obtained with argon. The .~~ak was again in the' form of a capillary shaped I

as described above, and this was directly fusedonto the vacuum gauge as indicated in fig. lOb.Jus:t as in fig. 8, the arro ...vs indicate the beginninga~d the end of the spraying of the le~k with thegas. When spraying for only one second the currentreturns to its initial value within a few seconds -.owing to the argon being burnt away. When spray-ing for 10 seconds (curves and arrows drawn in

- dotted lines) the current reaches within a fewseconds a new st~te of equilibrium whereby justas much argon is burnt away as flows in.

121

In this way it is possible to determine extremelysmall leaks of about 2 ?< 10-7 ems X mm. Hgfsec;'corresponding to an increase in 'pressure' in thevolume of the manometer of 0.0025 mm per month.Thè sensitivity is several hundred times greaterthan that given for the mass spectrograph method.

\ .

c

. -..,'

3" ')'-5 m mn Ha, secri\'

- T -,J \.".. ~ •-

1

- - - - --xx. -u ~-, T ,

IJ \ \/'- '.. ,

Ij '\ 2x p~~I'..

1

1..... ,..- -- - -- ....,/' r....... 6)( 10- ,

/ -, .....,-I

Fig. 10. a) Some examples of the detection of leaks in a valve evacuated with thevacuum gauge to which it was fused (the substitute gas was argon). The. size of theleak is indicated at the right of the curves in cm3 X mm Hg per second. When no moregas is released from the glass wall and the electrodes then the value at which the currentthrough the gauge reaches an equilibrium, with air around the tube, (t< 4 sec in the diagram)already gives an idea of the extent of the leakage.' b) illustrates the manner inwhich the artificialleak in the form of a capillary C was connected to the vacuum gauge,

la-t(sec)

1558022

b

"

An advantage of the latter method, however; is'that' the inflowing helium can also he measured, '.'when there 'is already a fairly high pressure of I -

other gases in the apparatus. Moreover, to "reachthe maximum sensitivity with the method described.here it is ess~ntial th~t both the object being testedand the manometer together must be well degased,so that the residual press1l!e is determined onlyby the leakage.

In most cases the valves to' be tested will havea volume of the same _order of size or greater thanthat of the tube _of the vacuum gauge, in whichcase it will take longer for the air to be "replaced bythe gas injected. Fig. lla gives an example where.there is a -volume of about 1 litre between the

....

\- .

,I..r

PHILIPS TÉCHNICAL REVIEW:, _,~

VOL. 11, No. 4 -,~.:~.- "_

leakage capillary and the vacuum gauge (sée also .: the' leak::When; ho~ever, a volume of i litre isfig. Uh). For the same sizeof leak both the initial placed in between -the vacuum' gauge and thevalue before spraying and the ultimate value capillary;the transition from the initial vaiue toafter spraying will be the same as in the case,.of the ultimate value takes place much more slowly.fig: 10. In both caseswhen,the stateof equilibrium This may be seen when courparing fig. Ua withis reached the amount of gas burnt away by the fig. lOa. In fig. Ua it has moreoverbeen indicated.discharge equals the amount flowing in through . by a dotted line what increase in current might be

expected if the volume of 1 litre were oniÏtted.,With larger volumes it thus takes much longer todetect a 'leak, but apart from this the sensitiVityis just as great as in the case of smallervolumes"

~c=====~J_~~--~_~m~------df~-t(sec) é. ',s. 58023

Fig. 11. a) An experiment similar to that ,in fig.10 but with avolume V of 1 litre between the vacuum gauge and the capil-lary, The broken line shows the trend to he expected for this

: leak if the capillary were attached in the manner indicated in "fig. lOb. b) The manner in which the artificial Ieak (capillary C) ,'was affixed to the vacuum gauge. '

, 122

i_";

I (PA)6~-------.---------r--------.

S~ary. 'The Philips vacuum gauge previously describedin this journal is suitable for gas pressures of 2X10--:<to 10-'5mm Hg, In order to produce a vacuum gauge based upon the'same principle but useful for lower pressures, the electrodesystem has been replaced by another consisting of two parallelround metal discs as cathode and an anode in the form of acylindrical jacket. To keep the distance between the poleshoesof the permanent magnet enveloping the glasstube as shortas possible this tube has been pinched at that point. Theseimprovements have given the new instrument a sensitivity. about ten times as great: it ,covers a range from 10-4 to 10-6mm _Hg•and can probably he used for pressures even below10-7mmHg. Vacuum gauges are also used as leak detectorsand some methods followed, for this purpose are briefly dis"cussed, It is indicated how the new design of vacuum gaugescan best b~ employed for such an applica~on.

.:

_ •• 4

:... I -

-,

"

, _'

" .

.~ -..