At low pressures the difference of pressure between the two ends of a connecting tube is necessarily small, and for rapid pumping it is desirable that the gas shall pass easily through the tube under this minute difference. Knudsen, who has carried out classical investigations on the behaviour of gases at very low pressures, has dealt with this problem, and established the laws which govern the passage. The application of the formulae derived by him for the case when the pressure is sufficiently low for the mean free path to be large compared to the radius of the tube shows at once the necessity of using very wide connecting tubes in exhausting vessels at low pressures, a fact of prime importance in the design of pumps and pumping systems. To take an example, suppose that the gas in question is air at C, and that the vessel to be exhausted has a volume of 2 litres. Let the connecting tube be 50 cm. long, and 5 mm. in diameter, which might be imagined, at first sight, to give ample freedom for the gas to pass. Let the pressure in the vessel at a given instant be io microbars and suppose the pump to maintain a pressure of •i microbar at the other end of the tube. The time required for the pressure in the vessel to fall to 2 microbar, twice that produced by the pump, is found from the formula to be over 5 minutes. Accordingly, whenever very low pressures are to be produced short wide tubes must be used for joining the pump to the system to be evacuated, and any taps which may form part of the connecting system must also be of large bore, for even with a tube only a centimetre or two long the time required for the fall of pressure specified in the example just taken will be a few minutes if the diameter is only 2 mm. Special large glass taps of very wide aperture are used in modern work.
One of the great troubles in producing very high vacua is to get rid of the vapours of any volatile liquids or soft solids which may be present, even in small quantities. The chief sources of such vapours are the mercury used in mercury pumps and the greases which must be used on taps to make them airtight. The vapour pressure of mercury is low, being a few ten-thousandths of a mm. of mercury at room temperatures, but such pressures cannot be tolerated in high vacuum work. To prevent the vapour arriving from the pump or from mercury surfaces of any kind which hap pen to be embodied in the apparatus, it is usual to insert "traps" between the mercury surfaces and the actual vessel in which the vacuum is required. Such traps may consist either of tubes which can be cooled by immersion in a cold liquid or of tubes coated on the inside with a metal having no great affinity for mercury, such as sodium or potassium. In the case of a condensation trap the temperature is best maintained well below o° C, say at — 78° C by solid carbon dioxide mixed with ether or alcohol, or, better still, at — i8o° C or so, by liquid air, since at very low pressures condensation is difficult to initiate.
As regards tap grease, there are special compositions, such as "Ramsay fat" (4 parts of pure unvulcanised rubber, 2 parts of vaseline, and i part of white paraffin) which have a very low vapour pressure, and should be used if some form of grease is inevitable. For some work the greased tap is replaced by a mer cury seal, in which, by the lifting of a reservoir, a mercury level is raised so as to cut off communication between two tubes. A number of special cements and waxes are in use for joining to gether metal and glass parts of vacuum systems, such cements requiring low melting point combined with low vapour pressure. Sealing wax, which must be of best quality, has long been widely used for this purpose : a black cement called picein, similar in composition and appearance to Chatterton compound, is much used in Germany, while many American workers favour Khotin ski cements, made of gum lac with a small percentage of a coal tar distillate. Whenever possible, the use of cements is avoided in high vacuum works. Glass parts are fused together, and metal parts soldered together, while recently a technique has been worked out, and is widely employed, which permits glass-to-metal joints to be made by fusion.
Both glass and metals contain large quantities of occluded gases and gases held at the surface by forces whose precise nature is undetermined, water vapour being particularly persistent in the case of glass. When a vessel is evacuated to low pressures, say to a ten-thousandth of a millimetre of mercury, the surface gases are liberated, slowly if the vessel is at atmospheric temperature, much more rapidly if it be strongly heated. Dunoyer quotes a case of a vessel of 9 litres capacity, which was evacuated down to a pressure of i microbar and sealed off. In to hours the pressure had risen to be nearly a hundred times as great, namely 95 micro bars. If therefore a vessel be evacuated and sealed off without any special precautions, the vacuum will rapidly deteriorate. To get rid of the surface gases as far as possible tubes which are to be sealed once for all, such as X-ray tubes, are subjected to a prolonged baking in special ovens while being evacuated, the temperature being taken as high as is consistent with the safety of the glass. Metals heated in vacuo also give out gases, mainly water vapour and hydrogen. The behaviour of these occluded gases has been extensively studied in connection with the manu facture of electric lamps and thermionic valves, for example by Langmuir and his colleagues in the laboratories of the General Electric in America, and by Norman Campbell and his colleagues in the laboratories of the General Electric Company, Limited, in England.
Pumps.—The points which have just been reviewed demand consideration no matter what type of pump be used to produce the vacuum. Of recent years a large number of pumps of different types have been evolved for the rapid production of low vacua, which will now be described.