FINDING TUTOR WITH ENGINEERING BACKGROUND

281

HISTORY OF VACUUM DEVICES

P.A. Redhead National Research Council Ottawa, Canada

Abstract An outli ne of the early history of vacuum devices (pumps and pressure gauges) is presented, from the first recorded experiments on vacuum in the 1640s to the start of the modern period of vacuum technology when ultrahigh vacuum was introduced in 1950.

1. INTRODUCTION

The first experiments to produce a vacuum were proposed in 1631 by Reneri of Leiden in correspondence with Descartes [1]. The first recorded experiment to produce a vacuum appears to have taken place in about 1641 when Gasparo Berti experimented with a water barometer [2], see Fig. 1. The lead tube AB was about 11 metres long, the tube and reservoir were fill ed with water and the valves G, D, and B were then closed. The valve B was then opened and the water level dropped to L, a vacuum was left in the chamber above L. The experiment was not very convincing since it was assumed that sound would not travel through a vacuum yet a bell M rung in the vacuum space could be clearly heard, the sound was probably transmitted through the supports of the bell . In 1644 Vincenzio Viviani repeated Berti’ s experiment using a mercury-fill ed glass tube which was inverted with its open end in a reservoir of mercury. The experiment was probably planned by Evangeli sta Toricelli who was credited by earli er historians with performing the famous experiment in 1643. The Toricelli an experiments convinced most scholars at the time that a vacuum had been created. This was the time of the dual papacy and the concept of vacuum was still considered anathema, thus it was dangerous to continue experiments on vacuum in countries accepting the Roman pope so the experiments were pursued in the reformed countries and in France, where the French pope was incli ned to ignore the opinions of the Roman pope.

A few years after these initi al experiments on vacuum the first vacuum pumps were created by Otto von Guericke the Burgomaster of Magdeburg, this is when the history of vacuum devices begins. In this paper we examine the development of vacuum devices, in particular vacuum pumps and gauges, from 1650 to the start of modern vacuum technology in 1950 when ultrahigh vacuum was first achieved and measured.

Several histories of vacuum technology may be consulted for further detail s [3-11].

Fig. 1 The apparatus used by Gasparo Berti i n about 1641 to produce a vacuum

282

2. VACUUM FOR FUN AND SCIENCE, THE FIRST TWO CENTURIES, 1650 TO 1850

The vacuum pump was one of the six instruments invented in the 17th century that had a profound impact on experimental science, the others were the pendulum clock, the telescope, the thermometer, the barometer and the microscope. Otto von Guericke started work on air-pumps in the 1640s and his work was first reported by Caspar Schott [12] in 1657. Von Guericke’ s first two pumps are shown in Fig. 2. These pumps were essentiall y water pumps being worked backwards. In his first experiment in the 1640s he fill ed a well -caulked wooden barrel with water and two strong men pumped the water out, as air rushed through the pores in the wood into the empty space a noise was heard. In the second experiment he used a large copper sphere attached to the pump; the water was omitted and the air pumped out directly. When most of the air was pumped out the sphere coll apsed with a loud noise, von Guericke reali sed that this was caused by atmospheric pressure. A more accurately spherical vessel was made and the experiment was then successful.

The news of von Guericke’ s experiments was spread through Europe by Schott’ s book and the next improved pump was designed by Robert Boyle [13] and built by Robert Hooke in England in 1658-9. This pump is shown in Fig. 3, it used a rack and pinion to move the piston thus greatly reducing the eff ort needed to operate the pump. Boyle was the first to make a measurement of vacuum with a mercury manometer placed in a bell -jar (see Fig. 4), his pump was capable of reaching a pressure of about 1/4 inch of Hg (6 Torr). Thus Boyle was the first to design a vacuum pump and combine it with a pressure gauge. The piston pump and the mercury manometer were to remain the only type of vacuum pump and pressure gauge for about two hundred years.

Fig. 3 Piston pump designed by Robert Boyle and built by Robert Hooke in 1658-9.

Fig. 4 The first measurement of a sub-atmospheric pressure by Robert Boyle c.1660. A beaker of mercury with a manometer tube more than 32 inches long was sealed in a bell j ar and evacuated by the pump of Fig. 3.

Fig. 2 The pumps of von Guericke in the 1640s. (Top) The first pump with a wooden barrel. (Bottom) The second pump with a copper sphere.

283

Improvements were made to the piston pump in the next two hundred years but the basic design remained the same. The ultimate pressure achieved by these pumps decreased only sli ghtly, the mechanical nature of the vacuum experiments of the time did not require significantly lower pressures, but the new pumps were easier to use. Hawksbee [14] built a pump in 1704 with two cyli nders in which the pistons were balanced against each other as they were driven in opposite directions by the rack and pinion (see Fig. 5), this pump was capable of reaching 1.9 Torr in two minutes. At the Great Exhibiti on of 1851 in London the first prize for vacuum pumps was won by Newman [15] whose pump achieved a pressure of 0.5 Torr. Figure 6 shows a simil ar commercial pump of about 1850 that is littl e changed from the construction of the Hawksbee pump of 150 years before.

Fig. 5 The double-piston pump of Hawksbee (1704). Fig. 6 A commercial double-piston pump from about 1850.

Figure 7 shows the ultimate pressure achieved by the various pump designs in the period from 1660 to 1900 and clearly indicates the limit ed reduction in attainable pressure in the first two hundred years. The vacuum pump made important contributions to science in its first two centuries but throughout this period it was widely used as a source of entertainment and instruction because vacuum was a novel and fascinating subject.

3. VACUUM FOR SCIENCE, THE GREAT LEAP FORWARD, 1850 TO 1900

The 50 years from 1850 to the turn of the century produced an intense activity in the development of vacuum technology driven initi all y by the needs of scientific research and later by the demands of the incandescent lamp industry. In the fifty year period the attainable pressure was reduced by six decades and vacuum equipment, though still awkward, was in use in industry. The need to thoroughly outgas the equipment and to avoid the use of rubber tubing was well establi shed.

284

Fig. 7 The ulti mate vacuum from 1660 to1900. Note the break in the time scale. The references to the experiments are; Boyle [13], Hawksbee [14], Geissler [17], Sprengel [18], Crookes [19], Edison [22], Fleuss [24], Gimingham [20], Kahlbaum [23].

The development of mercury-piston pumps was the major factor in achieving lower pressures. The first step occurred in 1854 when Juli us Plücker of the University of Bonn asked Heinrich Geissler, his glass blower, to design a glass pump using a mercury piston to permit experiments on low pressure gas discharges; Geissler built a pump in 1855 capable of reaching 0.1 Torr which was first described in a pamphlet by Mayer [16] publi shed in 1858. In 1862 Töpler [17] invented an improved form of the Geissler pump and in 1865 Sprengel [18] devised a pump in which a train of mercury droplets trapped packets of gas in a glass tube and carried the gas away.

In the 1870s Willi am Crookes, with his assistant Charles Gimingham, attempted to achieve a vacuum “ approaching perfection” . Figure 8 shows Crookes’ f irst pumping system [19] using a Sprengel pump. An improved version [20] of this pumping system using seven fall t ubes was capable of achieving a pressure of about 2 x 10-5 Torr as measured by a McLeod gauge. Crookes greatly improved the vacuum conditi ons by replacing all rubber tube connections with ground glass joints and by heating the apparatus to degas it.

The vital step in pressure measurement was McLeod’ s invention of his vacuum gauge [21] in 1874 which was based on the compression of the gas by a mercury column to an easil y measured higher pressure, and the use of Boyle’ s law to calculate the original pressure (see Fig. 9). This permitted pressure measurements down to the 10-6 Torr range. Since the McLeod gauge does not measure the pressure of condensable gases, such as water, the true pressure may have been higher than those reported in this period.

285

Fig. 8 Crookes’ f irst vacuum system using a single Sprengel pump (1872).

Fig. 9 The first McLeod gauge (1874).

By the end of the 1870s Crookes’ vacuum techniques had moved into industry and were being used by Edison to improve the vacuum in incandescent lamps. Figure 10 shows Edison’ s vacuum system which contained two Sprengel pumps, a Geissler pump, and a McLeod gauge; pressures of about 10-3 Torr could be produced [22]. Edison’ s pumps were operated manuall y and were slow and clumsy for industrial use. By1894 Kahlbaum [23] was able to obtain pressures as low as 3 x 10-6 Torr with a Sprengel pump.

In the same period the soli d-piston pump was greatly improved, by 1892 Fleuss had manufactured a pump with an oil -sealed piston and valves that were moved mechanicall y [24]. This pump was known as the Geryk pump and was capable of 2 x 10-4 Torr, it was widely used in the lamp industry to replace the manuall y operated mercury-piston pumps until t he invention of the rotary mercury pump by Gaede in 1905. The Geryk pump and the rotary mercury pump could be motor- driven and thus had the advantage over the older li quid-piston pumps for industrial use. By 1900 pressures of the order of 10-6 Torr could be achieved and measured and the pumps for industrial vacuum systems were being motor driven.

4. VACUUM FOR INDUSTRY, 1900 TO 1950

The change in ultimate vacuum in the period 1900 to 1950 is shown in Fig. 11. Vacuum technology made rapid advances in the period 1900 to 1920, the two figures that dominated this period were Gaede in Germany and Langmuir in the USA. The first improvement in high vacuum pump design was the invention of the rotary mercury pump by Gaede [25] in 1905, this was a rotary mercury pump which could be motor driven (see Fig. 12). It could produce pressures in the 10-6 Torr range and was manufactured in large quantiti es and widely used in the lamp and vacuum-tube industries.

286

Fig. 10 Diagram of Edison’ s vacuum system for the production of incandescent lamps (1870s).

Fig. 11 The ulti mate vacuum from 1900 to 1950. The references to the experiments are; Gaede (1905) [25], Gaede (1913) [27], Sherwood [34].

Fig. 12 Gaede’ s rotary mercury pump (1905). Fig. 13 Rotary oil pump designed by Gaede (1907) compared with Prince Rupert’ s “ Water-Bolt” , a water pump invented in about 1650.

In 1907 Gaede developed the rotary oil pump (see Fig. 13) primaril y for use as a fore-pump for the rotary mercury pump [26], its close resemblance to the water pump invented by Prince Rupert in about 1650 is evident. The principle of this type of waterpump goes back to at least 1588 with the vane pump of Rumelli .

287

The next major advance in pumping methods was the invention of the molecular-drag pump by Gaede [27] in1913 (see Fig. 14). With a fore pressure of 2 x 10-2 Torr a pressure of 4 x 10-7 Torr was measured by Dushman [28] at a rotation speed of 8,000 rpm. This pump was used in the vacuum- tube industry until t he arrival of the diff usion pump. Improved versions of the molecular-drag pump with higher pumping speeds were developed over the next 30 years, notably by Holweck [29] in 1923 and Seigbahn [30] in the 1920’ s, pressures in the 10-7 Torr range were typicall y obtained. Molecular- drag pumps were not much used until t he 1970s because their pumping speed was low and their reli abilit y poor as a result of the very close spacing between rotor and stator. They were ahead of their time because the technology for making high-speed rotating machinery with close spacings was still rudimentary. In recent times the molecular-drag pump has been successfull y re-introduced, usuall y in combination with a turbo-pump.

The mercury vapour diff usion pump was the first vacuum pump to have no moving parts. It was invented independently by Gaede [31] (see Fig. 15) and Langmuir [32] (see Fig. 16) in 1915-16. Langmuir’ s design with its large nozzle had much higher pumping speed and was quickly taken up for industrial use. The use of low-vapour-pressure oil s rather than mercury in diff usion pumps was introduced by Burch in 1928 [33]. The diff usion pump became the most widely used high vacuum pump until t he sputter-ion pump became avail able in 1958. Both mercury and oil diff usion pumps appeared to have an ultimate pressure of about 10-8 Torr as measured by a triode ionisation gauge, the curves of pumping speed of diff usion pumps as a function of pressure in manufacturers’ lit erature showed the pumping speed going to zero at about 10-8 Torr (such curves were still being publi shed as late as the 1960s). The early results using mercury diff usion pumps and ionisation gauges are typified by the work of Sherwood [34] in 1918 who measured an ultimate pressure of 2 x 10-8 Torr.

Fig. 15 Gaede’ s first mercury vapour diff usion pump (1915).

Fig. 16 Langmuir’ s original mercury vapour diff usion pump (1916).

Fig. 14 Schematic diagram of Gaede’ s molecular-drag pump; 1 pumping port, 2 stators, 3 rotors (1913).

288

Several types of pressure gauge were invented after the McLeod gauge in 1874. In 1897 Sutherland [35] invented a gauge depending on the viscosity of the gas, it could measure from 1 to 10-4 Torr. This principle was extended by Langmuir [36] in 1913 who developed a gauge using a quartz fibre which was made to oscill ate in the gas, the decrement in amplit ude of the oscill ations was a measure of the pressure, its pressure range was simil ar to Sutherland’ s gauge. Viscosity gauges were not very widely used. In 1906 two gauges which depended on the thermal conductivity of the gas were invented covering the range from atmosphere to 10-4 Torr. The Pirani [37] gauge measured the heat loss from a hot fil ament to the gas by the change in resistance of the gas. The thermocouple gauge [38] measured the change in temperature of the fil ament with a thermocouple. In 1910 Knudsen invented the radiometer gauge [39], which with care could measure to 10-6 Torr. Only the Pirani and the thermocouple gauge are still i n use. Various mechanical gauges were developed including the Bourdon and the diaphragm gauge, the only one to survive to the present day was the capacitance manometer [40] invented in 1929

Measurement of pressure in the high vacuum range prior to 1916 was diff icult, the only avail able gauges were the Mcleod gauge and Dushman’ s rotating disk gauge [41] (1915), the latter was the precursor to the spinning-rotor gauge, these were both awkward to use and limit ed to about 10-7 Torr. In 1916 Buckley described the hot-cathode ionisation gauge [42] (von Baeyer had reported the measurement of pressure in a hot-fil ament triode used as an ionisation gauge [43] in 1909; however, his work was not foll owed up and Buckley is generall y credited with the invention of the hot-cathode ionisation gauge). The triode hot-cathode ionisation gauge was almost universall y used to measure high vacuum until 1950. This type of gauge had a cyli ndrical i on coll ector of large surface area surrounding the cyli ndrical grid and axial fil ament, an early example [44] is shown in Fig. 17. With the use of this type of hot-cathode ionisation gauge the measured ultimate pressure in almost all vacuum systems was about 10-8 Torr. The ultimate pressure had hit a plateau that lasted for more than 30 years.

The limit at 10-8 Torr was generall y assumed to result from a fail ure of the pumps rather than the gauge, there was initi all y no suspicion that there might be a limiti ng process in the gauge at low pressures. The search for a method to reduce the ultimate vacuum in this period has been described in some detail [45] and will not be repeated here. Suff ice to say there was considerable evidence by the late 1930s that the hot- cathode ionisation gauge was not capable of measuring below 10-8 Torr. The reason for this limit (the x-ray limit as we now know) was not understood by workers in vacuum technology although by the late 1930s the production of photoelectron emission from the grid of a high-power vacuum tube, as a result of soft x-rays produced by electron bombardment of the anode, was well understood by vacuum tube engineer (see Ref. 40 and references therein). The two communiti es, although closely associated, apparently did not communicate on this subject.

Fig. 17 Triode ionization gauge of Dushman and Found (1921)

Fig. 18 Diagram of the Penning gauge, the first magnetic, cold-cathode, ionization gauge. P, cathodes, R, anode, H, magnet pole-face (1937).

289

The cold-cathode discharge in a magnetic field was first demonstrated by Philli ps [46] in 1898 but it was not until 1937 that Penning invented the cold-cathode ionisation gauge in a magnetic field, known as the Penning or Philli ps gauge [47] (see Fig. 18). This gauge was rugged and reli able but the ion current vs. pressure relation was non-li near and the discharge went out at about 10-7 Torr and thus it did not help to solve the pressure limit at 10-8 Torr. The gauge was also found to have a significant pumping speed and was the precursor of the sputter-ion pumps designed in the late 1950s. The fact that an electric discharge caused a pumping action had been known since 1858 when Plücker noted [48] “ Certain gases react with the platinum cathode and the resulti ng compounds deposit on the wall s. So we approach the absolute vacuum” . In 1916 Vegard observed [49] that there was a change in the pumping speed of a cold-cathode discharge when using diff erent materials for the cathode but not the anode. Penning noted “at high pressures a discharge with high currents should be avoided because of cathodic sputtering ... one has to take care of the fact that the discharge absorbs gas, in air about 1 lit re at By 1950 the minimum measured pressure was still 10-8 Torr, the same as it had been in 1920. Some experimenters had undoubtedly achieved pressures much lower than 10-8 Torr (e.g. W.B. Nottingham [50] and P.A. Anderson [51]) but were not able to make definiti ve measurements of these lower pressures. It had been suggested that the hot-cathode ionisation gauge had a limit t o the lowest measurable pressure set by an x-ray photocurrent but this had not been proven.

1950 is the end of our journey when the revolution in vacuum technology occurred (the Bayard- Alpert gauge [52]) which led to ultrahigh vacuum and the modern era of vacuum devices. Virtuall y all the new or improved vacuum devices developed since 1950 find their roots in devices developed between 1850 and 1950.

REFERENCES [1] Rene Descartes, Oeuvres, eds. Chas Adams and Paul Tannery, vol. 1, p. 205 (Paris, 1897).

[2] W.E.K. Middleton, The History of the Barometer (John Hopkins Press, Baltimore, 1964).

[3] E.N. da C. Andrade, The History of the Vacuum Pump, Adv. Vac. Sci. Tech., 1, 14 (Pergamon, Oxford, 1960).

[4] M.J. Sparnaay, Adventures in Vacuum (North Holl and, Amsterdam, 1992).

[5] L. Fabel, Vakuum-Technik, 35, 128 (1986).

[6] A.C. van Helden, Tractrix, 3, 149 (1991).

[7] M.H. Hablanian, Comments on the History of Vacuum Pumps, J. Vac. Sci. Technol. A, 2, 118 (1984).

[8] T.E. Madey, Early Appli cations of Vacuum, from Aristotle to Langmuir, J. Vac. Sci. Technol. A, 2, 110 (1984).

[9] P.A. Redhead, The Measurement of Vacuum Pressures, J. Vac. Sci. Technol. A, 2, 132 (1984).

[10] S. Shapin and S. Schaeff er, Leviathan and the Air Pump (Princeton University Press, 1985).

[11] P.A. Redhead, The Ultimate Vacuum, Vacuum, 53, 137 (1999).

[12] Caspar Schott, Mechanica Hydrauli co-pneuimatica (Wurzburg, 1657).

[13] Robert Boyle, new Experiments Physico-Mechanical, Touching the Spring of the Air and its eff ects, etc. (Oxford, 1660).

[14] F. Hawksbee, Physico-Mechanical Experiments on Various Subjects (London, 1709, Johnson Reprint Corp., New York, 1970).

[15] J.A. Bennett, Science at the great Exhibiti on (Whipple Museum, Cambridge, 1983).

[16] W.H.T. Mayer, Über das geschichtete elektrische Licht (Berli n 1858).

[17] A. Töpler, Dingler’ s Polytechnisches Jour., 163, 426 (1862).

[18] H.T.P. Sprengel, J. Chem. Soc., 16, 9 (1865).

290

[19] W. Crookes, Phil . Trans. Roy. Soc., 163,277 (1873).

[20] C.H. Gimingham, J. Soc. Chem. Ind., 3, 83 (1884).

[21] H.G. Mcleod, Phil . Mag, 48, 110 (1874).

[22] M. Josephson, Edison (McGraw-Hill , New York, 1959).

[23] G.W.A. Kahlbaum, Wied. Ann., 53, 199 (1894).

[24] H.A. Fleuss, German Patent, No. 76147, 13/8/1893.

[25] W. Gaede, Phys. Zeit., 6, 758 (1905).

[26] G. Meyer, Verh. d. deutsch Phys. Ges., 10, 753 (1907).

[27] W. Gaede, Ann. Physik, 41, 337 (1913).

[28] S. Dushman, Phys. Rev., 5, 224 (1915).

[29] M. Holweck, C.R. Hebd. Seances Acad. Sci., 177, 43 (1923).

[30] M. Seigbahn, Arch. Math. Astron. Fys., 30B, 17, (1944).

[31] W. Gaede, Ann. Phys. (Leipzig), 46, 357 (1915).

[32] I. Langmuir, Gen. Elec. Rev., 19, 1060 (1916); J. Frankli n Inst., 182,719 (1916).

[33] C.R. Burch, Nature, 122, 729 (1928).

[34] R.G. Sherwood, Phys. Rev., 12, 70 (1918).

[35] W. Sutherland, Phil . Mag., 43, 83 (1887).

[36] I. Langmuir, J. Amer. Chem. Soc., 35,107 (1913).

[37] M. Pirani, Verh. Physik Ges., 8, 686 (1906).

[38] W. Voege, Phys. Zeit., 7, 498 (1906).

[39] M. Knudsen, Ann. Phys. (Leipzig), 31, 633 (1910).

[40] A.R. Olsen and L.L. Hurst, J. Amer. Chem Soc., 51, 2378 (1929).

[41] S. Dushman, Phys. Rev., 5, 212 (1915).

[42] O.E. Buckley, Proc. Nat. Acad. Sci., 2, 683 (1916).

[43] O. von Baeyer, Phys. Zeit., 10,169 (1909).

[44] S. Dushman and G.C. Found, Phys. Rev., 17, 7 (1921).

[45] P.A. Redhead, in Vacuum Science and Technology, ed. P.A. Redhead, (A.I.P. Press, New York, 1994) p. 133.

[46] C.E.S. Philli ps, Proc. Roy. Soc., 64, 172 (1898).

[47] F.M. Penning, Physica, 4, 71 (1937).

[48] J. Plucker, Poggendorf’ s Ann. 103, 88 (1858); 105, 84 (1858).

[49] L. Vegard, Ann. Phys. (Leipzig), 4, 769 (1916).

[50] W.B. Nottingham, Phys. Rev. 49, 78 (1936); J. Appl. Phys., 8, 762 (1937).

[51] P.A. Andersen, Phys. Rev. 47, 958 (1935); 49, 320 (1936); 54, 753 (1938); 57, 122 (1940).

[52] R.T. Bayard and D. Alpert, Rev. Sci. Instr., 21, 571 (1950).