Vacuum, Vol. I I No. 4 October, 1952

L E T T E R S T O T H E E D I T O R

An Investigation o f a Metal Knudsen M a n o m e t e r

Sommaire Br6ve description d 'une jauge de Knudsen m&allique. Les d6tails du proc6d6 d'etallonnage comprenant les pressions entre IO -a et IO 4 mm. Hg sont donn6s. A la fin de la lettre des effets du ddgasage sont discutds.

IN A PREVIOUS paper W. Steckelmacher I has described and investigated a metal Knudsen manometer of his own construction. A similar instrument has been worked with in this laboratory.

This gauge is constructed in mainly the same way as Steckelmacher's although there are certain differences. Thus the vane is suspended in a single- ended tungsten wire, the necessary damping being achieved by means of two or three permanent magnets. The heaters consist of two helical resist- ance wires of a temperature running at about 30°C.

The connection to the vacuum system is made by a standard taper joint at the top of the envelope. The reading device is of the common lamp-and-

/ 1 Z 5 ~.

~A i r L

, / ~ ' ~ / I He[ / J / um

/ • I

-:7// / .

/ F

5 6 7 8 PressuPe mm H 9

10- 10 "~

Fig. I . Calibration curve for dry air and hydrogen.

scale type, and pressure readings may be taken by means of calibration curves belonging to the instru- ment. It has two ranges, 10-3-10 -4 and 10 -a- 10-Smm.Hg, controllable by a switch on the auxiliary transformer.

The present work included a calibration of the gauge in connection with a general survey of its properties. The calibration was made by means of a pressure-reducing device similar to that used by Knudsen in his fundamental works °-. It consisted of a number of glass bulbs with known volumes, joined together by a tube fitted with stopcocks or mercury cut-offs, which at one end was connected to a barometer and at the other to a mercury diffusion pump and liquid air trap. In this way the Knudsen gauge could be filled with gas up to a pressure which could be computed by Boyle's law. Before the Knudsen gauge was mounted in the system it was made sure by comparison with a precise McLeod manometer that the pressure obtained did agree with that computed by Boyle's law. This was found, in fact, to be the case throughout the range considered here. Our reason for performing the calibration in this way instead of in the one most commonly used was, that we wanted a calibration as absolute as possible without any auxiliary device.

The results of the calibration for dry air and hydro- gen are shown on the attached Graph (Fig. 1.) Solid curves (for air and helium respectively) are due to a calibration made in the usual way, the dotted ones (for air and hydrogen) refer to this report. It should be noticed, however, that on account of a breakage the suspension wire had to be shortened by about 10% before the measurements were made, so the results should be expected to be correspondingly too l O W .

The calibration of the low range 10 -5 mm.Hg proved to be impracticable in the present way because of a most disturbing gas development, which must be ascribed partly to the internal metal surfaces and partly to a possible minute leak. From our results we realize that the pressure in the gauge after cutting it off for reading will increase remarkable during the reading. In fact, the rate of increase was found to be on average 1 × 10 -5 mm.Hg per minute.

As pointed out in Steckelmacher's article metal gauges are designed for use in large metal systems where gassing defects are overcome by increasing the pumping speed. In the light ofthis it is not surprising, that working with the gauge under these circumstances will meet with difficulties. The example merely shows what will happen when doing so. However, this does not prevent metal gauges from giving correct results when properly used. We consider it

888

Octohet 1952 Vacuum, Vot. I1 No. 4

Fig. I. Back-s t r eaming of fluid f rom an 02 oil diffusion p u m p t h r o u g h a s t ra ight pipe.

essential that it has a very wide bore connection just as in Steckelmacher's mentioned above, as correct pressure measurements in metal systems depend upon the fact that the pressure gradient in the connecting tubes is negligibly small.

This work has been performed at the Royal Technical College, Copenhagen, under supervision of Professor Dr. R. E. H. Rasmiissen, to whom I would convey my best thanks for steady encourage- ment and words of instruction.

H. G. JENSEN.

Royal Veterinary and Agricultural College, Copenhagen, Denmark.

27th April, 1953.

x STECKELMACHER, W., Vacuum, I, (Oct. I95I) , 266.

z KNUDSEN, M. , Ann. Phys. Lpz., 444, (I914), 581.

On the Assessment o f Back-Streaming in Vacuum Systems

Sommaire L ' a u t c u r d6crit une md t hode pou r essayer l'6fficacitd de pibges refrig6r6s dans des sys t~mes ~ vide. L a chose est par t icu l ic rement impor t an t quand , dans u n apparei l metal l i ser sous vide par example , la piece est m a i n t e n u e

une t emp6ra tu re inf6rieure h l 'enceinte . E n vue de se rendre compte de la quant i t6 d 'hu i le rdtro-6vapor6e de la p o m p e , et pouvan t con tamine r la surface ~t recouvrir , une cu re d ' A d a m a 6td employ6e, con tenan t de l ' eau don t la surface a dt6 soupoudrde de talc pu lv6ru len t trbs fin. U n e aire connue de la surface fi recouvri r est t r empde darts l ' eau et l 'hu i le se r epend pou r fo rmer u n f i lm monomol6cu la i re h t ravers la surface. L ' hu i l e repousse le talk et la surface qui en est recouver t est ddfinie par des arr~tes tr~s nettes. A l 'aide de m o y e n s addi t ionels de calibrage la quant i td d 'hu i le est ddterminde. Les r6sultats de l ' enqu6te son t repr6sentds par des graphiques .

THE DANGERS which arise with diffusion pumps because the molecules of the boiling fluid diffuse or ' back-stream' into the evaporation chamber have been realised for some time 1 and various arrange- ments of baffles are commonly used to overcome this

Fig. 2. Back - s t r eaming of oil f r o m an 02 oil diffusion p u m p t h r o u g h a double r ight angled ben t pipe.

,~ 12Of

~-8oF

(5' ' - a b ' - z b o ' "TEM~ (*c3

difficulty 2. The problem may be particularly serious ff it is necessary to maintain a substrate in vacuo at a temperature significantly lower than that of the surroundings. In such cases the use of cooled traps is obvious, but means of checking the efficacy of such safeguards is desirable. The purpose of this note is to report a method used in studying the behaviour of oil diffusion pumps which is simple and easy to apply.

The very small quantity of oil which is deposited on a substrate can be measured using an Adam trough. A uniform layer of a fine powder is sprinkled over a clean water surface in the trough 3. A known area of the substrate is dipped into the water and the oil spreads to form a monomolecular film across the surface. The oil pushes back the powder and the area of the film is defined by a sharp boundary. I f two parallel glass barriers, waxed to prevent wetting, are arranged some distance apart and the oil deposited between them, then the area of the resulting rect- angular film can be readily measured. The most satisfactory powder in our experience is fine un- scented talc.* I t is important not to use too much powder. I f it is desired to convert the areas of the

4C

,_1 >o o I I i I I I i

- 2 0 - -40 -GO TIE:MR (°C.)

889