New Instruments and Methods of Measurement

Modern Fractionating Pumps

A. P. Nevlev

A speed of action of 7000 l/sec*), an ultimate pressure of \(10^{-9}\) mm without the use of traps, a maximum working fore-vacuum pressure of 1 mm, and the ability of the oil to withstand heating in the boiler for as long as desired while the pump is connected with the atmosphere[^2]—these are the latest achievements of vacuum technology, made possible by the use of fractionating pumps and modern high-quality oils. To be sure, 7000 l/sec was measured at \(10^{-8}\) mm, so that the pump throughput, determined by the product of the speed of action and the pressure at which it was measured, is only about 10 mm\(^3\) at atmospheric pressure; but in comparison with 1 l/sec for ordinary mercury laboratory pumps this is an enormous value.

High-vacuum pumps with so great a speed of action prove necessary in those areas of science and technology where it is required to pump out very large volumes and it is difficult to create absolutely reliable vacuum seals.

The speed of action of the first mercury condensation pump constructed by Langmuir[^31], 4 l/sec, in its time fully satisfied requirements, surpassing in this respect all other known pumps. The pressure of \(10^{-6}\)—\(10^{-7}\) mm attainable when mercury vapor was frozen out with liquid air was the limit of the dreams of physicists working in the field of high vacuum.

A substantial shortcoming was the pump’s need for a low fore-vacuum pressure. It proved possible to overcome this difficulty by employing a series connection of pumps, so that the pumps (or pump) operating on the low-vacuum side created (with a poor fore-vacuum) a pressure sufficient for the normal operation of the high-vacuum pump. As a result, designs appeared in which two or more pumps were combined into one. Such are the multistage pumps of Volmer[^32], Gaede and Beust[^33], Dunoyer[^34], the steel pump of Gaede[^34], and others. In this case the clearance at the nozzle through which the produc—

*) By the time of publication of the present article, speeds of action of 12,000 l/sec and higher had become known[^36].

the pumping-out, in each stage of the pump, became broader and broader as one approached the high-vacuum side. This contributed to an increase in the pumping speed and at the same time made it possible to use a higher fore-vacuum pressure. In Paine’s two-stage pump, improved in this way,^35 it proved possible to obtain a pumping speed of 15.5 l/sec at a fore-vacuum pressure of 3 mm and a limiting vacuum of \(3 \cdot 10^{-7}\) mm.

Nevertheless, despite all the advantages of the improved pumps, there remained one shortcoming that could not be eliminated. It lay in the mercury itself. Without the use of artificial freezing with liquid air, by means of mercury condensation pumps one could obtain only a pressure equal to the vapor pressure of mercury at the lowest temperature present in some part of the system being evacuated. But even freezing did not lead to the rapid attainment of a good vacuum. Several hours were required before the mercury-vapor pressure corresponding to the temperature of liquid air was reached. In addition, mercury vapor inside the vacuum system amalgamates the metals that form part of the instruments being evacuated, which is undesirable.

All this compelled a search for other working liquids, especially for powerful high-throughput pumps of large dimensions. Such liquids were soon found. They proved to be various kinds of oils and ethers with a vapor pressure (at room temperature) a thousand or more times lower than the vapor pressure of mercury.

Since for a long time mercury had been the only working liquid in condensation pumps, when the transition to oil was made the established “mercury traditions” were at first mechanically transferred to it. However, from a consideration of the properties of mercury and of oil (ether) there follows the need for an entirely different approach to oil.

PRINCIPLE OF FRACTIONATION

If mercury is poured into the reservoir of a condensation pump that can operate both with mercury and with oil, and, with a sufficiently well operating fore-vacuum pump, it is heated, then at first, up to a certain temperature determined by the construction, we shall observe the absence of any action of the pump. Upon reaching this temperature the pump will begin to operate, and with a further rise in the temperature of the mercury and with good action of the water cooling the pump will continue to operate well. The jet of mercury vapor issuing from the nozzle will overcome the opposing tendency of gas molecules to penetrate from the fore-vacuum into the volume being evacuated, and the limiting pressure that a mercury condensation pump can create will be determined only by the partial pressure of impurities in the mercury and by the vapor pressure of the mercury in the coldest part of the vacuum system.

If, now, some oil used in practice is poured into the reservoir of this same pump, the picture will be different. As the temperature is raised, just as for mercury, a certain lower temperature limit will be reached (lying, incidentally, lower than for mercury), after which the pump will begin to operate. However, with a further rise in the temperature of the oil, the pressure on the high-vacuum side (beginning at a certain critical temperature) will begin to increase, and, finally, the pump will cease to operate altogether. This upper temperature limit for oil (which mercury does not have) is connected with the decomposition of the heavy hydrocarbons of which it consists into lighter gaseous noncondensable products, which enter the high-vacuum side and lead to cessation of the pump’s operation.

If one takes as the criterion for the beginning and cessation of operation of the pump the pressure of permanent gases on the high-vacuum side of \(10^{-5}\) mm, then, for example, for the old Hickman and Sanford pump \(^{16}\) the working temperature range will be limited to only two degrees Celsius. Ptitsyn \(^{21}\) gives, for apiezon oil, a curve yielding a working temperature range, under the same restriction, of \(30^\circ\) C. It is natural to strive to extend these limits, but, as it turns out, this cannot be done by the simple choice of an oil; rather, it is necessary to turn to a study of the very mechanism of operation of the vapor-oil pump and to its improvement.

Every mineral oil has its own temperature limit of decomposition and (in contrast to mercury), being a mixture of various components, has neither a definite boiling point nor a definite freezing point. Making use of the differences in the physical properties of these components, one can, by successively heating the oil to different temperatures, separate them from one another in the form of fractions boiling in the temperature intervals taken.

Let us choose such temperature intervals within which, by distillation, cuts of the oil or foreign impurities in it are separated, having properties related to one another, and let us denote the quantities of these cuts within the chosen temperature intervals (or the amount of substance in each of the fractions thus obtained) by the letters \(A, B, C\), etc. As Hickman showed \(^{4}\), any oil can then be represented as the sum of a series of terms

\[ A + B + (C + D + \ldots + W + X + Y) + Z. \]

Here \(A\) is the fraction consisting of water vapor and of permanent gases dissolved in the oil and produced from it upon its decomposition (assuming that the boiling points of the remaining fractions lie higher). \(B\) is the light fractions, whose participation in the operation of the pump is undesirable because of the high elasticity of their vapors; \((C + D + \ldots + W + X + Y)\) are the working fractions, whose boiling points lie in the working zone—above the initial working temperature and below the temperature at which decomposition of the oil begins; and \(Z\) is the heavy fractions with boil—

stones with boiling points above the temperature limit of decomposition, resins, and polymerized products.

Since the ultimately attainable pressure is determined by the vapor pressure of the working fluid in the high-vacuum pump, it is obvious that it is desirable to find a way to use, on the high-vacuum side, the heavy oil fractions with the lowest possible vapor pressure, having separated them in the pump from the light fractions. Such separation of fractions can be carried out continuously in a fractionating pump in two ways: by fractionation in the gaseous and in the liquid phases, to the consideration of which we shall now turn.

Fig. 1.

Let a boiler be located at the bottom of the pumped-out tube, from which, upward along the tube, vapors of mixed fractions, constituting the oil, rise. After the tube has been heated to certain equilibrium temperatures—highest in its lower part and lowest in its upper part—its temperature regime will be established. The lightest oil fractions, condensing at the lowest temperatures, will rise along the tube to the greatest height. The heaviest fractions, having the highest condensation temperature, will rise only to a minimum height and will condense there on the walls of the tube. Thus, in the lower part of the tube there will at all times be vapors of all fractions. As the vapors rise upward along the tube, the number of fractions in them will decrease until only the lightest remain. Such separation of oil fractions in the vapor state is the simplest type of fractionation and is called fractionation in the gaseous phase. This type of fractionation is widely used in practice.

To clarify the role of fractionation in the gaseous phase in practice, let us critically examine the operation of various types of vapor-oil-

...oil pumps. In doing so we shall take into account the slight solubility in oil during condensation of the gases and vapors pumped by the pump, and their easy liberation during the subsequent heating of the condensate.

Figure 1 shows the four most widespread types of oil condensation pumps. Arrows with captions indicate the places where gaseous products and light fractions \(A\) and \(B\), entrained together with the vapors of the working fractions, are liberated and redissolved. It is obviously desirable that the amount of liberated fractions \(A\) and \(B\) should always be greater than the amount redissolved, and that the oil should thereby be freed of volatile constituents. Redissolution is most strongly expressed in pumps of the inverted type (Fig. 1—2, 3, 4), where components \(A\) and \(B\) are released at the top and absorbed in the lower part of the pump. The best with respect to redissolution is the simple straight-through pump (Fig. 1—1). The vertical tube of its condenser forms a kind of miniature fractionating column, at the top of which the undesirable volatile fractions \(A\) and \(B\) accumulate and are gradually pumped away by the fore-vacuum pump. The light fractions condensing and dissolving in the condensate in this pump, flowing downward along the condenser tube, evaporate again owing to heating by the opposing hot exiting vapors and are kept away from the region of high vacuum. The two-stage pump of the inverted type (Fig. 1—4) is the worst of all four as regards correct utilization of the working fractions and redissolution.

Fig. 2.

In order to create in a pump of any design the conditions for fractionation in the gaseous phase, it is evidently sufficient to maintain the fore-vacuum tube of this pump hot or, at least, not to subject it to strong cooling. Figure 2 shows a two-stage pump of the inverted type improved in this way, with a hot tube \(^{4}\). In addition to the fore-vacuum tube being insulated, in this pump it is provided with five tiers of nested small bulbs for retaining the light fractions flowing downward. The use of such bulbs considerably improves the operation of the pump.

Fractionation in the gaseous phase is a considerable advantage, since it makes it possible to remove from the evacuated system not only gases, but also comparatively easily condensing vapors. Proceeding from this, one should distinguish between the normal speed of action and the speed of removal of a pump[^4].

The normal speed of action of a condensation pump refers to its normal operation in pumping permanent gases, which can only be temporarily adsorbed by the surface of the condensate. It is not affected by moderate cooling of the vapor of the working liquid in the pump refrigerator, since the gases being pumped do not condense under such cooling. The concept of speed of removal, however, refers to the vapors of liquids that readily condense in the refrigerators of common pumps. In order to increase the speed of removal, it is necessary to create such conditions in the pump refrigerator that only the vapors of the working liquid (oil) condense in it, while the vapors of the substances that must be removed remain in the gaseous state. Thus, the speed of removal may be defined as the number of liters of vapor of the condensing liquid removed by the pump in 1 sec at the vapor pressure determined by the temperature of the coldest part of the system, from the evacuated volume to the pump refrigerator. From this definition it follows that an increase in the speed of removal can be achieved by raising the temperature of the refrigerator and of the fore-vacuum tube of the pump, i.e., by the conditions of fractionation in the gaseous phase.

Hickman[^4] carried out such an experiment. Two pumps were taken. One of them was a direct-action pump with fractionation in the gaseous phase (Fig. 1—1), and the other an ordinary inverted-type pump without fractionation (Fig. 1—3). The first had a speed of action for air of 3 l/sec and the second 10 l/sec. When these pumps were used for evacuating a system evolving condensing vapors instead of air, their relative speeds were reversed, since only the direct-action pump with fractionation in the gaseous phase could remove vapors.

Without taking into account the distinction between the normal speed of action and the speed of removal, gross errors are sometimes made in designing condensation pumps, when a refrigerator is also placed on the fore-vacuum tube of the pump in order to prevent removal of the vapors of the working liquid itself. Pumps with such a refrigerator accumulate readily condensing vapors and oil within themselves in liquid form, and the oil in them thereby quickly ceases to be suitable for operation. The speed of removal is of primary importance in chemical work when pumping vapors. In physical work it also plays a very large role, since condensing vapors—for example, water vapor evolved by glass—are present in all installations. Therefore a pump with a high speed of removal should always be preferred.

It should also be noted that pumps with fractionation in the vapor phase and with a high pumping speed have yet another advantage. The oil in them, during operation, is not contaminated by condensing volatile liquids, as in ordinary pumps, and therefore does not require such frequent replacement as in the latter.

Let the boilers \(1, 2, 3, 4\), etc. (Fig. 3), connected to one another by narrow heat-insulated tubes, be filled with mixed oil. If it is heated simultaneously in all the boilers, it will evaporate, and its vapors will rise through the vertical tubes and leave through nozzles into the inclined tube \(A\), which is pumped out through \(B\) by a rotary pump. Condensing on the walls of tube \(A\), the oil, owing to its inclination, will flow in the direction indicated by the arrows and, through tube \(a\), will enter boiler \(1\).

Fig. 3.

Since the condensate of the vapors that have emerged from all the nozzles enters it, while in the other boilers the oil only decreases owing to evaporation, after some time in boilers \(1\) and \(2\), and then in all the others, a certain difference of levels will be established, and the oil will continuously flow through the connecting tubes from boiler \(1\) into boilers \(2, 3, 4\), etc. Thus there will be established a continuous motion of the oil vapors along tube \(A\) from left to right and of the condensate from one boiler to another—from right to left.

Having entered boiler \(1\), the mixed condensate is heated, and from it there first evaporate the light fractions with the highest vapor pressure and the lowest boiling point. After this, only the remaining fractions, already in a heated state, enter boiler \(2\), and in it one or more fractions with a lower vapor pressure and a higher boiling point evaporate from them, and so on.

Thus only the heaviest fractions, with the lowest vapor pressure and the highest boiling point, enter the last boiler. Owing to the continuous flow of the oil and the successive evaporation of fractions, the same fractions will at all times enter each boiler, and the same fractions will also leave it, except for those evaporated in the boiler. Such separation of fractions in the liquid state constitutes fractionation in the liquid phase.

Since the limiting attainable pressure in the pumped-out apparatus is determined by the vapor pressure of the working liquid, in a multistage pump using oil it is advantageous for the working liquid in the high-vacuum stage to have the lowest vapor pressure.

Then the limiting attainable pressure will also be the lowest. Fractions with a higher vapor elasticity will not be able to penetrate by diffusion into the evacuated system, because this is prevented by the jet of vapor emerging from the nozzle of the last high-vacuum stage.

Let us now consider the practical application of fractionation in the liquid phase.

The sum of the terms of the series proposed by Hickman can be represented as the sum of only three terms \((A + B) + (C + D + \ldots + W + X + Y) + Z\), differing from one another in their basic properties. The amount of substance included in the first and last terms, in an oil suitable for operation, is usually small. The main mass of the substance is made up of the fractions of the second term—the working fractions.

Fig. 4.

The fractionating pump separating these three fractions from one another and using the fractions of the second term for operation must therefore consist of three parts. The first—of small volume, for retaining the light fractions \((A + B)\)—must be located in the fore-vacuum part of the pump unit, consisting of successively connected pumps. The second—the working part of large volume for the working fractions \((C + D + \ldots + W + X + Y)\)—must constitute the working part of the unit. The third—of small volume, for retaining the heavy non-working fractions of oil and resins—must be located somewhere near the part of the pump unit that uses the heaviest fractions for operation.

Figure 4 shows a pump unit constructed by Hickman for investigating and improving the conditions of continuous fractionation of this kind. It consists of two successively connected direct-acting pumps \(A\) and \(B\), a small volume \(C\) for collecting fractions \(Z\), and a small volume \(D\) for collecting the light fractions \((A + B)\), the latter being a system of the already mentioned nested bulbs in the form of a harmonica. To remove the lightest fractions from the oil composition, a section with a test tube is soldered to the upper bulb. The light fractions that enter this test tube can either be poured out by breaking off the glass tip drawn out at its lower end, or removed by unsoldering the entire test tube with these fractions.

Both pumps \(A\) and \(B\) are direct-acting, with fractionation in the gas phase. Light fractions from pump \(B\) partly enter pump \(A\) through an inclined tube. Condensates from pumps \(A\) and \(B\) go through tubes \(a\) and \(b\) to the point where they join; this point contains a steel ball, moved from outside by means of a permanent magnet into one of the two “pockets” \(a'\) and \(b'\), or into the recess \(c\) between them. The boilers of both pumps are connected by tube \(d\), and the pockets—with the boilers of the corresponding pumps. A third tube, not designated by a letter in the figure, connects the lower part of the boiler of pump \(B\) with the small boiler \(C\), from which there is no direct outlet for liquid, but only for vapor (tube \(C\)) and condensate (tube \(f\)). The boilers have electric heating, not shown in the figure.

Thanks to the device with the steel ball, three operating cases of the pump unit are possible:

1. Worst series connection. The ball is pushed into the left pocket \(a'\), and the oil from the condenser of the fore-vacuum pump \(A\) first enters the boiler of the high-vacuum pump \(B\), and only after this returns through tube \(a\) to pump \(A\).

2. Normal series connection. The ball is in recess \(C\), and each of the pumps uses its own condensate.

3. Best series connection. The ball is pushed into the right pocket \(b'\), and the oil from the condenser of pump \(B\) first enters the boiler of the fore-vacuum pump \(A\), and then returns through tube \(d\) back into the boiler of pump \(B\).

Table 1 gives the values of the ultimate pressures obtained with different sequences of oil passage for different working fluids and condenser temperatures. The pressures are given in mm of mercury.

Table 1

The data in the table clearly show the advantage of the best sequence, corresponding to the implementation of additional continuous fractionation in the liquid phase.

In the most modern pumps, both types of continuous fractionation are used, in both the gaseous and liquid phases. A schematic

the design of such a pump with combined fractionation would be obtained if, as an extension of tube \(B\) (Fig. 3), one added the vertical fractionating tube of pump \(A\) of the experimental apparatus (Fig. 4—the volume \(D\) for collecting the light fractions).

Single-stage pumps can be fractionating only in the gaseous phase, since fractionation in the liquid phase in principle requires the use of several stages. Successful fractionation in the gaseous phase is possible only in a direct-action pump.

DESIGN OF PUMPS

The question of the choice of nozzle arrangement is very important. Because of great structural difficulties, up to the present time there have been no multistage direct-action fractionating pumps, either metal or glass, with a vertical arrangement of the nozzles. All such pumps are made of the inverted type and therefore always have a relatively low pumping speed. In them it is impossible successfully to carry out fractionation in the gaseous phase, whereas a horizontal arrangement makes it possible to have combined fractionation as well.

The advantage of the vertical design is the ease of its manufacture from metal. The material used is ordinary seamless tubing, machined entirely on a lathe. Its major disadvantages are:

1. The fore-vacuum nozzle with the narrowest gap between it and the condenser has a small diameter. This requires great precision in manufacture and careful development of the design.

2. The width of the concentric boilers is small in comparison with their length. This leads to undesirable mixing of oil in neighboring boilers, facilitates the transfer of heat from one boiler to another, and makes their operation dependent on one another.

3. The areas of the boilers are limited by the dimensions of the nozzles. These areas cannot be increased without widening the tube at the base of the pump, and this destroys the simplicity of the design.

Advantages of horizontal metal pumps:

1. The areas of the boilers do not depend on the diameters of the nozzles.

2. The supply of heat to the boiler of each stage is independent and can be regulated.

3. The high-vacuum end of the pump can be made from a large-diameter tube, and the low-vacuum end from a small-diameter tube, without complicating the design.

4. The pump itself can serve as a straight connecting tube between the rotary pump and the volume being evacuated, or other parts of the vacuum apparatus, more often than a vertical pump can.

Despite all considerations in favor of the horizontal metal construction, the vertical construction is the most widespread. This can

MODERN FRACTIONATING PUMPS

can be explained only by the conservatism and poverty of equipment of the workshops manufacturing pumps. Glass pumps, requiring only a gas burner for their manufacture, are often made as combined pumps—with a horizontal arrangement of the high-vacuum nozzles and a vertical arrangement of the fore-vacuum nozzle, with fractionation in the gaseous phase.

The pumping speed of a pump is determined by the well-known formula \(S = \frac{V}{P}\frac{dP}{dt}\) in \(l/\mathrm{sec}\); here the pressure \(P\) at which it is measured is indicated. For oil-vapor fractionating pumps, \(S\) remains practically constant only in the pressure interval from \(10^{-4}\) to \(10^{-6}\) mm. In calculating a pumping unit consisting of several pumps connected in series, the following requirements must be satisfied: 1. That each preceding pump create a sufficiently low limiting pressure \(P_{np}\) to serve as the fore-vacuum for the subsequent pump. 2. A sufficiently high throughput of each of the pumps (the product \(S \cdot P_{np}\)), ensuring the removal by the preceding pump of all the gas discharged during evacuation by the subsequent pump. This can be formulated, by numbering the pumps in order from low vacuum to high vacuum, in the form of the inequality

\[ S_1P_1 \geq S_2P_2 \geq S_3P_3 \geq \cdots \geq S_nP_n. \]

The pumping speed of the unit for which the indicated requirement is fulfilled will be determined by the pumping speed \(S_n\) of its last high-vacuum stage, so that \(S_{\mathrm{agr}} = S_n\). One might think that, in trying to increase \(S_n\), we would come up against a limitation on the part of \(S_1\), the rotary pump. The following example shows that this is not true.

Let \(S_1 = 0.2\ l/\mathrm{sec}\), \(P_1 = 0.5\) mm. It is asked what \(S_n\) could be provided by the operation of such a rotary pump at \(P_n = 10^{-6}\) mm. From the equality \(S_1P_1 = S_nP_n\) we have \(S_n = 100\,000\ l/\mathrm{sec}\). The question, therefore, lies not in limitation by the fore-vacuum pump, but in the ability to provide \(S_n = 100\,000\ l/\mathrm{sec}\).

Work on obtaining such large \(S_n\) has proceeded up to the present time in the following directions: 1. Increasing the dimensions of the pump, giving an increase in \(S_n\) proportional to the square of the increase in the “scale” of the pump, but reducing the maximum permissible \(P_{n-1}\) (\(P_1\) for modern rotary pumps is \(5\text{–}10 \cdot 10^{-4}\)). 2. Increasing the cross-sectional area of the gap between the edges of the nozzle and the condenser, giving a linear increase in \(S_n\) according to the simplified formula of Gaede[^10]: \(S_n = 11.7 A_n\ l/\mathrm{sec}\), where \(A_n\) is the area of the gap section in \(\mathrm{cm}^2\), but still more rapidly reducing the maximum permissible \(P_{n-1}\). 3. Increasing the diffuse access of gases from the evacuated space into the stream of vapor carrying away gas molecules, while simultaneously reducing the access of vapor from the nozzle toward the high vacuum.

In the latter direction many investigators worked, solving the stated problem by an appropriate choice of the shape of the nozzles. Thus Crawford[^5], Ho[^6], Lauritsen[^6], Zabel[^7], Amdur[^8] proposed using, instead of one nozzle, several conical nozzles in each stage.

Fig. 5 shows Ho’s multi-nozzle pump. As a criterion of the quality of nozzle operation Ho[^9] introduced the concept of the nozzle velocity coefficient as the ratio of the experimentally found \(S_n\) to that calculated by Goede’s formula[^10] (from \(11.7 A \lambda/\sec\)), proceeding from the geometric dimensions of the nozzle gap. This coefficient proved, for most mercury-vapor and oil-vapor pumps, to lie within only 2 to 10%. Copley, Simpson, Tenney, and Phipps[^11] found that single-nozzle pumps with a conical jet of vapor, calculated on the basis of Laval turbine-nozzle theory, have a higher velocity coefficient (up to 56%) than multi-nozzle pumps. Holtsmark, Ramm, and Ustin[^12], for a mushroom-shaped nozzle in the form of a truncated cone, found the most advantageous angle at the apex of the cone to be 70–75°, giving a conical, downward-directed jet of vapor. Embry proposed a promising nozzle, shown in Fig. 6, of the mushroom type, with internal expansion of the vapor jet, giving a velocity coefficient of 30%.

Fig. 5.

Fig. 6.

No one has yet carried out a sufficiently thorough investigation of nozzle operation.

TYPES OF PUMPS

The history of oil-vapor pumps begins with Burch[^13,^14], who in 1928 was the first to use oil and one of the first to express the basic ideas of fractionation in the gas-like phase[^15]. He was followed by the works of Hickman and Sanford[^16], Becker and Jaycox[^17], Estermann and Beek[^18], Brandenstein and Klumb[^19], Henderson[^20], Ptitsyn[^21], Burden[^22], Stozharov[^23], and others. Oil-vapor pumps began to be built by the firms “Metropolitan-Vickers” in England, Geff, Gallo, and Pilon in France, etc. In our country oil-vapor pumps were produced serially by the “Svetlana” plant.

The first models of fractionating pumps of the modern type were built by Hickman^4 in 1935. One such model is shown in Fig. 7. It is the simplest horizontal glass-and-metal fractionating pump in the form of a glass tube containing metal nozzles and partially filled with oil heated from below during operation. Such a pump gave a pressure of \(10^{-7}\) mm, although it also required a fore-vacuum pressure of less than \(0.05\) mm.

Hickman’s further work in the laboratory of the firm “Distillation Products” proved very fruitful. Fig. 8 shows glass pumps with combined fractionation: \(1\)—a vertical high-pressure nozzle, \(2\)—a spur preventing the return of the bulk of light contaminants back into the boiler, \(3\)—an intermediate horizontal nozzle, \(4\)—a high-vacuum nozzle, and \(5\)—a receiver for heavy fractions, resins, and polymerized products not taking part in the operation. The heaters are made of bare wire immersed directly in the oil.

Fig. 7.

Cooling of the horizontal tube of pump \(C\) may be by water or by blown air. The speed of action at \(10^{-5}\)—\(10^{-6}\) mm is \(5\ \text{l/sec}\). The fore-vacuum pressure is \(0.01\) mm. These values depend on the dimensions selected. The limiting pressure depends on the grade of oil and, with a trap at \(25^\circ\text{C}\), may be obtained at \(5 \cdot 10^{-8}\) mm. Fig. 9 shows a section of a metal fractionating pump manufactured in series by the firm “Distillation Products,” with working-tube diameters of 4.8 and 21 inches. The three nozzles of the pump are arranged inside the inclined tube at an angle of \(5\)—\(10^\circ\). The first fore-vacuum nozzle enters the branch pipe of the condenser cooled by water and connected by a flange to the working tube. The lightest fractions, through this branch pipe, enter the fractionating box with shields and annular baffles, where they are collected. Only the heaviest of these light fractions pass through the tube into the condensate receiver. The condensate of the second and third nozzles also enters it, and nonvolatile substances collect there. The condensate receiver is connec-

opening with the vapor chamber of the first vessel; it with the chamber of the second vessel, and the latter with the chamber of the third. Inside each of the

Fig. 8.

chambers partitions are arranged, forming a labyrinth, so that the working oil must travel a long path before it reaches the high-vacuum stage. The speed of action of a pump of diameter

…with a diameter of 4 inches is equal to 250 l/sec, and with a diameter of 21 inches—7000 l/sec, at a pressure of \(10^{-4}—10^{-6}\) mm. Figure 10 shows the external appearance of the 4- and 8-inch pumps, and Fig. 11—the pump with a diameter of 21 inches. It is interesting to note that the increase in pumping speed here was obtained entirely by increasing the scale.

Indeed,

\[ \frac{7000}{250} = \frac{21^2}{4^2} = 28, \]

i.e., the pumping speeds are related as the squares of the diameters.

Fig. 9.

Fig. 10.

Fig. 11.

Vertical designs of metallic fractionating pumps by Lokenvitz \(^{24}\), Malter and Markuvitz \(^{25}\), and the Kharkov Phys—

13 UFN, vol. XXIX, issue 1–2

Physical-Chemical Institute of the Ukrainian Academy of Sciences³ are shown in Figs. 12, 13, and 14, respectively.

Their design is clear from the drawings. The pump in Fig. 14 is of great interest owing to its high speed coefficient,

Fig. 12.

Fig. 13.

Fig. 14.

which approaches unity (0.9). Its pumping speed with a pipe diameter of 129 mm is \(1200\ l/\text{sec}\) at a pressure of \(10^{-6}\) mm. It is easy to see that the pump represents a step forward in the design of a high-vacuum nozzle, even in comparison with

21-inch Hickman pump. In addition to those mentioned, other vertical designs are known, for example, those of Sikes and Bancroft[^26] and others, which are not of fundamental interest.

TRAPS

Even when oils with the lowest vapor elasticity are used, the gradual penetration of oil into the evacuated apparatus occurs. This phenomenon of diffusion made it necessary to use various kinds of traps in the path of vapor penetration. The purpose of traps is twofold. They must[^27]: 1. Prevent the movement of oil vapor into the evacuated system. 2. Reduce the amount of vapor moving in this direction to concentrations lying considerably below saturation. This task can, of course, be successfully solved only when oils with low vapor elasticity are used.

In the broadest sense, a trap may be not only an obstacle, but may also have its own pumping speed. Accordingly, traps may be divided into dynamic ones—actively accelerating pumping—and static ones, which have no pumping speed of their own.

The former include traps with an adsorbent[^28], cold traps[^28], and electric shields[^27]. The latter include mechanical shields without cooling or with moderate cooling[^27], and hot shields[^29].

Where pumps with a high speed of action are used, it is desirable to employ traps satisfying the following general requirements:

1. Completely eliminate all harmful penetration of oil in the reverse direction. 2. Be simple in design and manufacture. 3. Require little maintenance and attention during operation. 4. Introduce minimal losses in the speed of action of the apparatus. These requirements are particularly well met by mechanical shields, which have recently found wide use in combination with moderate cooling by circulating water.

Figure 15 shows the construction of one variety of high-capacity mechanical shield. Its action consists in the fact that not a single molecule of oil from the pump can penetrate into the evacuated space without striking the surface of one of the rings of the shield (in Fig. 15 they are shown by horizontal dashes) and temporarily condensing on it. This greatly slows the diffusion of oil vapor toward the high vacuum. The design shown, which does not appreciably reduce the speed of action, was proposed by Morse[^27] for a horizontal metal pump. Morse observed, during operation of a horizontal fractionating pump without a shield, the spreading of a visible oil film on the walls of the tube connecting the pump with the ionization manometer.

...at a rate that could easily be recorded over the course of several hours. After installing a mechanical baffle, however, of the kind shown in Fig. 15, Morse was unable to detect any visible traces of oil under the baffle even after continuous operation of the pump for two months.

A mechanical baffle may operate with water cooling of the tube in which it is placed (see Fig. 15), or without it, or else with direct cooling of the rings of which it consists. The rings of a cooled baffle are usually made of aluminum or another material with good thermal conductivity. A baffle with water cooling thus becomes a kind of additional condenser to which all oil molecules that have penetrated toward the high-vacuum side are, as it were, “attracted.” As a result, the baffle quickly “fogs up.” It becomes necessary to stop operation of the pump, remove the baffle, and clean the oil from its plates. This is its disadvantage.

Fig. 15.

In Fig. 16 is shown a glass trap—a mechanical baffle with water cooling, proposed by Hickman30 for phthalates and apiezons, which does not reduce the conductance of the vacuum system. Its principle of operation is the same as that of the water-cooled mechanical baffle.

In Fig. 17 is shown the device of an electric baffle, which, it is true, has not yet found wide application. A system consisting of a heated cathode in the form of a spiral of tungsten wire and an anode made of metal mesh is inserted inside the pumping tube on the high-vacuum side. In the space between these electrodes, ionization takes place of oil molecules moving from the pump in the direction opposite to pumping. The ionized molecules may either be attracted by some special negative electrode situated near the pump nozzle and then gradually pumped away by the pump, or they may fall on the surface of the incandescent cathode and undergo cracking there, with the formation of light noncondensable hydrocarbons pumped away by the pump.

If a mechanical baffle is installed on the high-vacuum side and its plates are heated sufficiently so that oil does not

condensed, then there is obtained the so-called hot baffle, which, owing to the reverse evaporation of oil molecules, reduces the rate of their penetration into the apparatus being evacuated. However, a certain fraction of the oil molecules, owing to re-evaporation from the reverse surfaces of the baffle rings, will nevertheless enter the evacuated apparatus and will condense on its cold parts. This is the disadvantage of such a hot baffle.

Birch29 proposed heating the plates of the baffle to a much higher temperature, sufficient to cause cracking of the heavy hydrocarbons with the formation of lighter ones, as also occurs in electric baffles. Such

For amyl and octyl phthalates

For heptyl and benzyl phthalates and Apiezon oils

Fig. 16.

Fig. 17.

high-temperature hot baffles are especially suitable for prolonged evacuation of high-voltage X-ray tubes and generator tubes.

Metal traps with liquid air, glass traps with water cooling, and traps with charcoal are well known.

Of interest is a new method of reducing the elasticity of oil vapors above the high-vacuum nozzle, mentioned by Hickman1. It consists in the fact that a nonvolatile component is specially added to the pump fluid. During operation of the pump, this component separates from the oil and enters the last chamber of the high-vacuum nozzle, in which, upon condensation, it virtually “dries” the residual oil vapors, apparently dissolving them in itself.

A. P. NEVLIEV

OILS

The basic principles that should be followed in choosing an oil for fractionating pumps may be formulated, on the basis of operating experience, as the following requirements:

1. Low hygroscopicity and a low capacity for dissolving gases and vapors. 2. Stability under overheating. 3. A wide range of operating temperatures. 4. Low vapor pressure at room temperature. 5. Chemical inertness with respect to substances introduced into the vacuum system and to gases.

Organic liquids used in pumps are inferior to mercury in two closely related respects[^30]: during the first pumping they release a large quantity of dissolved gases and deteriorate when they come into contact, in the hot state, with air. Damage is done to the oil in two ways: 1) when it is heated in the presence of air at the beginning of pumping; 2) after the disappearance of the vacuum, when the oil comes into contact with air while being strongly heated.

Fig. 18.

As a result of such contact with air, volatile substances appear in the oil that reduce the limiting vacuum, as well as oxidation products that contaminate the oil. Consequently, after air has entered a hot pump, it may cease to operate normally for a long time. A good oil must withstand the test for the so-called “degree of poor handling,” which consists in admitting air at atmospheric pressure into a hot pump operating with this oil for, for example, 2 minutes, and then, immediately after pumping is begun again, giving the same vacuum as before the test.

The working temperature range of good modern oils should cover an interval of at least 50°. The permissible fore-vacuum pressure for the oil should be not lower than 0.1 mm; the vapor pressure at room temperature, not higher than \(10^{-6}\) mm. These requirements are met by isobutyl sebacate, widely used in America, isoamyl sebacate (Amoil S), diethylhexyl phthalate (Octoil), diethylhexyl sebacate (Octoil S), and the worst of these—butyl phthalate. With Octoil S in fractionating pumps, without the use of traps, pressures of the order of \(10^{-9}\) mm have been obtained. This is the latest achievement in the field of limiting pressure obtained by simple means. In the field of fore-vacuum pressure, the latest achievement is 0.6 mm for butyl phthalate and 1 mm

for some having a high density of chlorinated hydrocarbons, the composition of which is not indicated in their manufacturers’ company catalogs. These oils operated in the booster pump of Fig. 18, having a speed of action of \(4\ l/sec\) at \(10^{-2}\ mm\) at first with elevated heater wattages and \(10^{-3}\ mm\) later at reduced power. Until recently we had been using apiezons, which withstand rather poorly the test for degree of ill-treatment, as well as fractions of Vaseline oil distilled off in vacuum. At the same time we also have oils rich in promise, such as the diisoamyl ester of azelaic acid and others, obtained by Candidate of Chemical Sciences, Docent M. A. Zakutskaya in the laboratory of the Department of Organic Chemistry of the Central Asian State University in Tashkent, headed by Corresponding Member of the Academy of Sciences of the Uzbek SSR, Professor I. P. Tsukervanik*).

CONCLUSION

The ever wider and wider spread of fractionating pumps, due to their advantages over simple oil-vapor and mercury condensation pumps, will apparently soon lead to their complete displacement of all other high-vacuum pumps. Apiezons and other oils similar to them will have to give way to more perfect ethers with lower vapor elasticity. Fractions in a narrow temperature interval will have to give way to liquids operating in a wide range of working temperatures, with a large working temperature zone. To extend the working zone, it is possible to use specially composed mixtures with the upper limit of the working zone of one component being the lower limit of the working zone of another. In this way the best use of the oil at the various stages of fractionation can be achieved.

In the electrovacuum industry, fractionating pumps will in many cases make it possible to use almost untreated “crude” oil, which will cheapen the cost of obtaining vacuum. The enormous speeds of action achieved by fractionating pumps, when they are further increased, will completely change existing pumping equipment and will permit the creation of new equipment in which it will be possible not to pay such great attention to the impermeability of all joints and to leakage. The capacity of future pumps will make it possible to have even small openings in the vacuum system without deterioration of the vacuum below that required.

It is possible that the high speeds of fractionating pumps will enable them to find application also in the future atomic-energy industry.

*) Mention should be made of the organosilicon compounds developed by VEI\(^ {37}\), known under the name of silicones and, apparently, far surpassing all known pump liquids\(^ {38}\).

References

1. K. C. D. Hickman, Journ. of Applied Physics, 11, 303, 1940.
2. L. Malter and N. Marcuvitz, Rev. Scient. Instr., 9, 92, 1938.
3. K. D. Sinelnikov, A. K. Val’ter, L. N. Ulezko, and A. N. Yamnitskii, ZhTF, vol. XI, no. 10, p. 879, 1941.
4. K. C. D. Hickman, Journ. of Frankl. Inst., Febr., 215, 1936.
5. W. W. Crawford, Phys. Rev., 10, 557, 1917.
6. T. L. Ho, Rev. Scient. Instr., 3, 133, 1932.
7. R. M. Zabel, Rev. Scient. Instr., 5, 54, 1935.
8. I. Amdur, Rev. Scient. Instr., 7, 395, 1936.
9. T. L. Ho, Physics, 2, 386, 1932.
10. W. Gaede, Zs. f. Phys., 4, 350, 1923.
11. M. I. Copley, O. C. Simpson, H. M. Tenney, and T. E. Phipps, Rev. Scient. Instr., 6, 265, 1935.
12. J. Holstmark, W. Ramm, and S. Westin, Rev. Scient. Instr., 8, 90, 1937.
13. C. R. Burch, Nature, 122, 729, 1928.
14. C. R. Burch, Proc. Roy. Soc., (A), 123, 271, 1929.
15. C. R. Burch and F. E. Bancroft, Brit. Pat. 346243, January 30, 1930.
16. K. C. D. Hickman and C. R. Sanford, Rev. Scient. Instr., 1, 140, 1930.
17. J. A. Becker and E. R. Jaycox, Rev. Scient. Instr., 2, 773, 1931.
18. I. Estermann and H. T. Byck, Rev. Scient. Instr., 3, 482, 1932.
19. M. von Brandenstein and H. K. Klumb, Phys. Zs., 33, 81, 1932.
20. J. E. Henderson, Rev. Scient. Instr., 6, 66, 1935.
21. S. V. Ptitsyn, ZhTF, vol. II, no. 7—8, p. 660, 1932.
22. J. A. Bearden, Rev. Scient. Instr., 6, 276, 1935.
23. A. I. Stozharov, ZhTF, vol. IV, no. 9, p. 1763, 1934.
24. A. E. Lockenwitz, Rev. Scient. Instr., 8, 322, 1937.
25. L. Malter and N. Marcuvitz, Rev. Scient. Instr., 9, 92, 1938.
26. Sykes and Bancroft, Brit. Pat. 477013, June 20, 1936.
27. R. S. Morse, Rev. Scient. Instr., 11, 277, 1940.
28. K. R. More, R. F. Humpreys, and W. W. Watson, Rev. Scient. Instr., 8, 263, 1937.
29. C. R. Burch, Brit. Pat. 396205.
30. K. C. D. Hickman, Journ. of Frankl. Inst., March, 383, 1936.
31. I. Langmuir, Phys. Rev., 8, 48, 1916; Journ. of Frankl. Inst., 182, 719, 1916.
32. M. Folmer, Ber. d. Phys. Ges., 52, 804, 1919.
33. M. Folmer, Zs. f. angew. Chemie, 34, 149, 1921.
34. N. A. Kaptsov, Physical Phenomena in Vacuum and Rarefied Gases, ONTI NKTP USSR, p. 68, 1937.
35. S. Dushman, Journ. of Frankl. Inst., 211, 689, 1931.
36. Chem. and Metallurg. Eng., Febr. (1946).
37. K. A. Andrianov, O. I. Gribanova, D. A. Gambarova, and A. L. Karpova, Electricity, 4, 39 (1946).
38. C. R. Brown, Rev. Sci. Instr., 16, 316 (1945), No. 11.