Project Description

Per gentile concessione di Varian S.p.a.

Tratto da : “High Vacuum Technology: A Practical Guide” di Mars Hablanian

Ultrahigh-vacuum techniques and system components have evolved together with special pumping methods uniquely adapted to the requirements of extreme cleanliness, bakeability, and pumping speed. The principal pumping mechanisms employed are chemical transformation, whereby gases are chemically combined into solid compounds having very low vapor pressure, and a method involving an ionization step that permits acceleration of the atoms or molecules in an electric field to drive the ions directly into a solid surface, where they remained captured.


At high-vacuum conditions a surface can hold large quantities of gases compared to the amount of gas present in the space. This produces a pumping action either by physisorption or gettering, which refers to a chemical combination between the surface and the pumped gas. Many chemically active materials can be used for gettering. In vacuum systems the material commonly used as titanium because it is chemically reactive with most gases when it is deposited on a surface as a pure metallic film, but it is rather inert in bulk form because of the tenacious oxide film covering its surface.

To produce a pumping action, all that is needed is a source of titanium and a means of producing a fresh (unoxidized) layer preferably on a large surface. Various forms of heating can be used to deposit a titanium film by evaporation or sublimation. The fresh titanium, deposited on surfaces surrounding the source, forms stable, solid compounds with chemically active gas atoms or molecules that strike the surface. This capture process can be continuous if new layers al titanium are constantly produced.

Figure 1 shows a typical arrangement.


The pumping speed versus pressure curve for any titanium pump is shown in Figure 2. Pumping speed is maximum when a complete film of titanium is maintained on the pumping surface. Pumping speed, then, is maximum at low pressure. The number al gas molecules striking the titanium surface decreases with decreasing pressure. Therefore, below some pressure point a particular titanium source is able to supply titanium fast enough to sustain a complete film of fresh titanium. In this pressure range, the pumping speed is limited only by the intrinsic speed Sj of the titanium surface area and the gas conductance from the chamber into the pump.

It should be noted that at law pressure. periodic sublimation is employed to allow greater saturation of the titanium film before more titanium is deposited. The life of a titanium source is finite, but with judicious use (to be discussed later) it can be made to serve for a relatively long period. When pressure increases, and hence the density of gas molecules per unit volume is greater, pumping speed decreases and is dependent on the rate of continuous titanium sublimation.


At higher pressures, the impinging gas molecules combine with the titanium virtually as fast as it is deposited and a complete film of fresh titanium is never formed. Consequently, the probability of capture for gas molecules decreases. Throughput is constant in this pressure region because the pumping action is limited by the rate of titanium deposition.

As in the case of cryopumps, the pumping speeds of gettering pumps cannot be defined precisely because it depends somewhat on the mixture of gases, the presence of gases that are not chemically pumped, the temperature, and the previous history of a surface.


The following method can be used for computing an approximate pumping speed. Surface area, speed for specific gases, and the conductance of the titanium sublimation pump inlet aperture are the variables considered (see Table 1).

The area-limited speed for specific gas is:


To use titanium economically, it is essential to determine the time required to deposit a complete but nonwasteful film and then cycle power to the filament accordingly.where A is the area of gettering surface in square centimeters and Sj is the intrinsic speed per square centimeter of titanium surface for a specific gas, in liters/second.

Figure 3 shows typical pressure rise characteristics for a system in which the titanium sublimation pump yields a pressure reduction of a factor of 10 before saturation of titanium causes the pressure to rise. At successively lower pressures, saturation occurs often a longer period of time because the number of reacting gas molecules is decreased proportionally.


Since the mechanism of titanium pumping is chemical combination, the theoretical maximum throughput can be found by using the appropriate chemical formula and the rate of titanium sublimation in atoms per second. To use as an example pumping nitrogen (compound: TiN), the typical sublimation rate from a titanium source is 3×1017 atoms/s since one titanium atom can combine with one nitrogen atom, the theoretical maximum nitrogen throughput is

3x1017 atoms/s

However, nitrogen forms molecules with two atoms. Therefore, the throughput in molecules per second is 1.5×1017 mol/swhich is converted to the normal units for throughput by using Avogadro’s number:


There are 6.023 x 1023 molecules in a mole of gas that at atmospheric pressure (760 torr) has a volume of 22.4 l. Titanium evaporation can be of great assistance in vacuum systems by providing high pumping speeds at relatively high pressures. Typical pumping speeds for pumps al different diameters are shown in Figure 4.



In Figure 1 the source al titanium is shown in the form of a wire. In practice, there are several ways to produce a mechanically stable evaporating surface. For example, molybdenum-titanium alloy wire will retain its shape at an elevated temperature while preferentially evaporating titanium because it has a higher vapor pressure compared to molybdenum. One or more filaments can be used. If one al the filaments has been depleted, the other can be used by external switching. To provide a larger storage of titanium, a hollow titanium spheroid piece can be heated by an internal heater (Figure 5).

The titanium piece is removable and should be replaced before it gets too thin to retain its shape.

A titanium sublimation control unit provides a regulated power source for operating either a filament sublimation cartridge or a radiation-heated titanium sphere. Essentially, it is a current-regulated power supply (low-voltage ac power) which is applied directly to the filaments of the sublimation cartridge or to the heaters inside the Ti-Ball sphere. Power is adjustable according to the desired evaporation rate.

In the manual mode it can provide 0 to 50 A either continuously or periodically (a maximum of 10 V). In the automatic mode the current is supplied for a period of a few minutes and then switched all or reduced to a standby value for a period that depends on the pressure in the vacuum system.

This period is adjusted according to the signal received from the pressure gauge. Often, a life indicator is used to show the amount of titanium expended as measured through an ampere-hour integrator.

When a new filament is put in operation, it must be degassed to reduce the initial burst of gas, which is released when the sublimator is heated. For pressures above 5×10-6 torr, it is recommended to maintain sublimation continuously.


Manufacturers usually recommend a certain waiting period between evaporations, in the manual mode, according to a system pressure. However, the optimum “off” period may vary for different systems because it depends on the geometric relationships and the size of the surface receiving the evaporating gettering material. Typically, a monolayer of titanium is evaporated during 2 min of sublimation time. Typical sublimation rates for a filament and a Varian Mini-Ti-Bali are shown in Figure 6 and 7. Chemical gettering is not possible with inert gases and it is almost useless with a gas such as methane (Table 1).

A technique has been developed to pump inert gases which includes an ionization process. Usually, the ionization is produced as a result of a collision with electrons. The electrons may be produced either by cold cathode field emission or thermoionic emission from hot filaments. The electrons and then the ions are accelerated through an electric field of a few thousand volts. The velocities obtained are sufficient to drive the ions into a solid metal surface. They are “buried” a few atomic layers deep and they remain captured in the surface more or less at room temperature, and it the diffusion rates are low, they will not reappear in the gas phase for a long time.



Pumps utilizing chemical and ionization pumping effects can generally be called ion-getter pumps. Early designs (after 1955) had a variety of arrangements for electron sources and for titanium evaporation. Later designs, which are used almost exclusively now, are called sputter-ion pumps because the supply of a fresh titanium film is produced by a process called sputtering. The basic structure consists al two electrodes, anode and cathode, and a magnet (Figure 8).

The anode is usually cylindrical and made of stainless steel; the cathode plates positioned on both sides of the anode tube are made of titanium, which serves as gettering material. The magnetic field is oriented along the axis of the anode. The basic cell resembles a cold cathode ionization gauge.

Electrons are emitted from the cathode due to the action of an electric field, typically 5000 to 7000 V applied between electrodes. The presence of the magnetic field produces long, more-or-less helical trajectories for the electrons. The long travel path of electrons before reaching the anode improves the chances of collision with the gas molecules inside the pumping module. The common result of a collision with the electron is the creation of a positive ion. In other words, when the electron collides with a gas molecule (with the velocity or energy attained due to the high-voltage field), it tends to knock out another electron from the molecule. The positive ions travel to the cathode due to the action of the same electric field. The collision with the solid surface, when the velocity or energy is adequate, produces a phenomenon called sputtering (i.e., ejection of titanium atoms from the cathode surface).



The fresh titanium film covers various surfaces in the pump, where it reacts with gas molecules as illustrated in Figure 8 (black layer inside the anode). Some of the ionized molecules or atoms impact the cathode surface with high enough force to penetrate the solid and remain buried.

The detail pumping mechanisms in sputter ion pumps are more complex than outlined here. It may be appreciated that different gases are pumped in a different way and at different pumping speeds. Also, there exists the possibility of previously pumped gases to be reemitted: for example, when the cathode surface is eroded, as illustrated in Figure 9. Sputter-ion pumps are generally associated with ultrahigh vacuum, although they can be used successfully in the high-vacuum region. Also. other pumps can produce an ultrahigh vacuum. The unique advantage of sputter-ion pumps, aside from their relative cleanliness, is their complete isolation from the atmospheric environment. Like cryopumps, they do not have an exhaust. They accumulate the pumped gases inside the pump. Unlike cryopumps, they are not regenerated periodically. They are used until the cathode is exhausted. Naturally, their life depends on the amount of pumped gas, and therefore they should not be used at higher pressures for extended periods.


The general pumping speed characteristics of sputter-ion pumps are similar to other vacuum pumps. There is a more-or-less constant-speed region at intermediate pressures, the net pumping speed is reduced to zero at low pressure, and it decreases to very low values at higher pressures. Due to practical engineering considerations sputter-ion pumps are not designed for high continuous-gas-throughput conditions. Compared to other high-vacuum pumps of equal pumping speed, sputter-ion pumps have approximately 100 times lower maximum throughput values. The pumping speed begins to decrease near 1×10-5 torr rather than near 1×10-3 torr.

A typical pumping speed characteristic is shown in Figure 10.


The ultimate pressure of an isolated pump is the 10-12 torr range. There is not single, simple, specific reason for the particular ultimate pressure. Presumably, it is due partly to the equilibrium between the residual gas evolution inside the pump and the pumping speed. This is similar to other pumps.

Partly, however, it may be due to the tendency for the electric discharge in the gas to extinguish at very low pressures.

The pumping speed of ion-getter pumps depends, as in cryopumps. on the arrival rate of the gas molecule (i.e., their molecular weight and temperature) and the sticking coefficient or the probability of capture upon arrival at the surface. The pumping speed per unit area, S (liters per second per cm2), can be expressed as:


where s is the sticking coefficient, T the absolute temperature (K), and M the molecular weight. This equation is not very useful unless the sticking coefficient is known. Unfortunately, it can vary within a factor of 2 or more. The initial sticking coefficient for nitrogen on titanium surface has been quoted in the technical literature at values ranging from 0.1 to 0.7. For sputter-ion pumps an additional variable is the efficiency of the sputtering process or the sputtering yield.


At the present time there are five varieties of industrial sputter-ion pumps: the diode pump (Figure 8), in which the cathodes are attached to the body of the pump and a high (positive) voltage is applied to the anode; the triode pump, where the cathodes are separated from the pump body and a high (negative) voltage is applied to the cathode (Figure 11); “differential” pumps, where one of the cathodes is made of a metal other than titanium; StarCell pumps (Varian Associates), which are triode pumps with a distinctive star-shaped configuration; and pumps that include a non-evaporative gettering module. The latter provides a high speed for hydrogen and consists of a corrugated strip of constantan into which is sintered a special alloy (zirconium, vanadium, iron).

When heated this alloy undergoes a transformation producing a very large gettering surface.

In summary, the pumping action in sputter-ion pumps is thought to be produced by the following processes:

1.   Ion burial and entrapment within the metal lattice a few atomic layers under the surface of the cathode.

2.   Gettering of chemically active gases at the cathode and elsewhere by sputtered deposits.

3.   Diffusion of hydrogen into the cathode material.

4.   Dissociation of complex molecules into simpler fractions which are then pumped by one of the mechanisms.

5.   Production of neutral atoms of high velocity (or energy) by neutralization of ions and scattering from the cathode surface. The neutral atoms can subsequently be pumped by the burial process in areas of deposited material on the anode, cathode, or pump envelope.

For practical designs of sputter-ion pumps the anode cells are between 15 and 25 mm. This is suitable for operations with magnetic fields of 1 to 1.5 kG.

A voltage of at least 3 kV is necessary to start the pump reliably. The pump current increases with the voltage until 6 or 7 kV is reached, which is a common range of operation. These conditions produce pump currents between 5 and 30 A/torr; for the cell, the pumping speed is between 0.3 to 2 l/s.

Pumps are made normally up to 200 l/s but can be made larger.


Ion-getter pumps are capable of pumping any gas, including noble gases and hydrocarbons. Due to the differences in molecular weight and chemical activity, the pumping mechanisms and the associated pumping speeds are different for different gases. The ionized molecules bombarding the cathode produce different rates of sputtering depending on their molecular weight and the angle of approach. Small angles (grazing the surface) produce a higher sputtering rate. For light gases such as hydrogen the sputtering rate is too low to obtain pumping speeds comparable to those for nitrogen and oxygen.

Titanium is chosen for the chemical gettering action because it is a highly reactive element that forms stable compounds with many different gases. Ironically, titanium in bulk form is very corrosion resistant at normal temperatures because it has a tightly packed oxide layer which protects its surface. Getterable gases are oxygen, nitrogen, CO, CO2, and many others.

In addition to sputtering, a fraction of impinging ions can remain buried in the cathode. However, as noted before, the depth of penetration is only a few atomic layers. Because the cathode is continuously eroded, the pumping speed for getterable gases by this mechanism is negligibly small. Hydrogen, in principle, is also getterable, but by itself, does nor produce a sufficient sputtering yield. When hydrogen is pumped together with other gases, an important fraction of hydrogen is pumped at the anode. But the major mechanism of hydrogen pumping is due to the diffusion into the cathode material. It diffuses into the cathode to depths much beyond the range of ion implantation and therefore can be pumped in a stable manner and high amounts. The pumping speed for hydrogen is typically two times higher than the speed for nitrogen. Cathode structure should be resistant to, or should accommodate, distortions because of the presence of large amounts of hydrogen in the lattice. If the temperature increases, the diffusion rate is increased but a fraction of pumped hydrogen may be released because the binding energy of titanium hydride is lower than that for nitride and oxide.

Inert gases cannot be pumped by chemical reactions. As noted before, they are pumped primarily by burying into surfaces in various parts of the pump after impinging either as ions or as neutral atoms at high velocity (or energy). They can then be covered by layers of sputtering titanium films. The inert gas atoms implanted at the cathode may be released as the cathode erodes due to continuing ion bombardment. Early diode pumps, in association with this, had a rather low speed for inert gases and tended periodically to release previously pumped gas (called an argon instability).

One way to overcome this problem is to increase the portions of the atom that rebound from the cathode surface as neutrals and then are implanted into the anode and the walls of the pump, where they are held more or less  permanently. The probability of this reflection depends on the masses of the impinging ions and the cathode material as well on the incidence angle. The higher the mass of the target material, the more likely it is that the ion will rebound and will also retain more energy after reflection. The use of heavier metals, such as molybdenum or tantalum as an auxiliary cathode material, then provides the necessary improvement. If both cathodes were made of the heavier metal, the sputtering yield would be reduced. Pumps with such differentiated cathodes produce a very substantial improvement in the pumping speed of inert gases and also improve its stability (Table 2), although the speed for getterable gases may be somewhat reduced.


A more effective approach is to provide a more favorable incidence angle for the ions. One way to achieve this is to slot the cathode surface, providing shallow angles for incident ions, and to increase both the reflection probability and the sputtering yield. Such pumps produce inert gas speeds that are 5 to 10% of the pumping speed for air, and the pumping speed for the inert gases is stable.

The next logical development is to make the cathode in the form of thin parallel strips with some space between them. This grid structure is partly transparent for ions and allows the pump wall (behind the cathode) to be available as a pumping surface. The wall can be considered as an auxiliary anode. This type of pump is called a “triode pump” (Figure 11), as expected, the pump walls are kept at a ground potential and the high negative voltage is applied to the cathodes. In a triode pump the ions that retain their energy after being neutralized at the cathode may travel to the pump wall and be buried there or reflect and be pump at the anode. The pumping speed for the inert gases for triode pumps is 20 to 25% of speed for air and it is very stable even after a long period of pumping. The pumping speed for hydrogen is somewhat lower than that of diode pumps with titanium cathodes. Also, distortion of cathode strips can occur after pumping large amounts of hydrogen or when pumping at the high end of the pressure range.

To overcome problems with cathode distortions the cathode has been designed with radial strips or spokes located in alignment with each anode cell (Figure 12).


This avoids the tendency for distortions because the spokes are attached only on one end. They are free at the center arid are symmetric relative to the ion activity in the pumping cell. Because of the shape these pumps are known under the trademark StarCell (Varian Associates). An additional important advantage is obtained from the symmetric pattern of cathode erosion, which begins at the center and progresses toward the periphery. In this way more titanium can be sputtered before the cathode structure loses its integrity. In a diode pump, the ion bombardment eventually produces a hole in the cathode plate after which the pumping action is nearly lost. In a triode pump the end of the cathode life occurs when a strip brakes, which often causes a short circuit. Due to the gradual symmetric erosion in the StarCell pump its life is increases by nearly a factor of 2 compared to a standard triode pump and by 60% compared to a diode pump, so that the pump can operate continuously at 1×10-5 torr for a period of 1 year. Pumping speeds of StarCell pumps for different gases compare as follows: for oxygen, nitrogen, water vapor, and methane, nearly the same; for carbon dioxide, 10% higher; for hydrogen, twice as high as nitrogen; for argon, 22%; and for helium, 30% of nitrogen. The most recent development in ion pump design is the addition of a nonevaporating getter strip inside the pump. This increases the pumping speed and the maximum throughput capability of the pump.


Sputter-ion pumps are normally shipped to the user after evacuation, degassing, and sealing under vacuum. This keeps the pump surfaces clean, unsaturated with adsorbed gases, and makes it easier to start the pump. The pump inlet is covered and sealed by a flanged plate, which includes an evacuation tube made of’ copper. After the final processing this tube is sealed usually by a pinch-off tool that produces a leaktight cold weld as it cuts off the tube. When constructing a new vacuum system the pump should be opened only after everything is completed and the system is ready to be evacuated. The exposure of the pump to atmosphere should be minimized.

Ion pump systems normally have a high-vacuum valve which can isolate the pump when the system is exposed to air. If ultimate pressure in the 10-9 torr range is adequate, baking may not be mandatory and the system may have two Viton O-rings, one under the bell jar (Figure 13) and one in the valve plate. For lower pressures the chamber and all seals must be made of metal and the entire system surrounded by a bake-out oven.


When pumping speeds of ion pumps are measured or specified it should be clearly understood whether the measurements are made with a new, processed pump or after a certain amount of gas has been pumped. The pumping speed of new ion pumps decreases during operation until a stable level is reached. The initial pumping speed for nonsaturated pumps (StarCell types) is shown in Figure 14. This should be compared with Figure 10, which shows the saturated pumping speeds for the same pumps. The higher pumping speeds at pressures above 10-7 torr (Figure 14) may or may not be utilized, depending on the operation of the system. For example, to saturate a 60 l/s StarCell pump 2.5 torr l of gas is normally required.


Sputter-ion pumps are sometimes difficult to start if the pressure is too high or too low when the high voltage is applied. At the higher pressures near 10-2 torr, particularly if the pump has been exposed to the atmosphere for a few days, the initiation of’ the electric discharge may produce enough current drain from the power supply to increase the temperature of the internal electrodes. This, in turn, produces increased outgassing, which may cause a runaway condition. At the same time the voltage may drop, which reduces the pumping speed, making the starting condition even more difficult. If the pump is not clean, more gas may be driven from it than is pumped. Sometimes it is necessary to start and stop the pump a few times before it finally begins to reduce the pressure in a stable manner. Triode and diode ion pumps exhibit different behavior during starting. These differences depend on the design of the power supply and its polarity as well as the cathode shape and material.

At very low pressures the initiation of the pump may take a long time. If a system had been evacuated to 10-9 torr by some other means, the ion pump will start in a few seconds; at 10-12 torr it may take half an hour. As mentioned before, the performance of the pump at the higher pressures depends to a great degree on the power supply characteristics. One of the important aspects is the requirement for maintaining a current sufficient to obtain higher throughput without overheating. The best performance is obtained when a pump is properly matched to a power unit. A power supply that is too large for a given pump will cause overheating and may damage the pump at high starting pressures. A power unit that is too small will reduce the maximum operating pressure of the pump or, in other words, reduce its maximum-throughput capability. The power unit must supply voltage and current required under different working conditions, automatically limiting the delivered power at high pressures to facilitate starting of the pump under different conditions. The power consumption of the pump strongly depends on pressure and becomes essentially negligible at pressures below 10-6 torr (Figure 15).


One interesting and useful characteristic of sputter-ion pumps is the approximately linear relationship between the electrical power used by the pump and the pressure. The pump automatically adjusts the intensity of the pumping activity to the number of molecules present in the system, This should not be surprising because the basic pumping mechanisms of the pump depend on the number of available molecules. The electron “cloud” rotating in the cell of a sputter-ion pump under the influence of a magnetic field (somewhat analogous to a rotor of an electric motor) is relatively stable and is not strongly dependent on pressure. Therefore, the current of the pump. similar to cold cathode gauges can be used as a measure of pressure, as shown, for example, in Figure 25.

It should be understood that this is only an approximate indication and should not be thought of as a substitute for an ultra-high-vacuum gauge of good quality. The measurement becomes more uncertain as the pressure is reduced below 10-8 – 10-9 torr. This is due to leakage currents and other effects that are independent from pressure.


There is a class of gettering pumps in which the active material is not evaporated or sputtered on a pumping surface but remains in a specially prepared very porous intermetallic compound. Such pumps were pioneered the by SAES Getters company in Milan, Italy and are often called NEG pumps. The pumping action initially involves an adsorption event followed by diffusion inside the material. (It is ironic that these pumps can be called “diffusion pumps” more than the vapor jet pumps, for which the common name is a misnomer). The most comprehensive treatment (in high-vacuum textbooks) of the physical basis of the pumping action and properties of various materials was presented by O. Saksaganskii, who devotes 47 pages to this subject in his book.

Usually the pumping material is in the form of thick films or layers deposited on a substrate strip or ribbon, typically a nickel-iron alloy such as constantan, nichrome, or stainless steel and others. The compound has a porosity of 50 to 70%, may be as thick as a few millimeters, and can have a real internal physical area 100 times higher than the projected area (100 to 1000 m2/kg). The active metallic compound is usually composed of finely dispersed powders of transition metals, which may be sintered, plasma-sprayed, or rolled into the substrate. A common composition has 84% zirconium and 16% aluminum. Other compounds include titanium, vanadium, iron, nickel, and also tantalum, molybdenum, tungsten, and rhenium.

Because of the progressive saturation of the exposed surfaces, the getter must be reactivated by heating to enhance the diffusion process and clean the surface. This activation is achieved by raising the temperature to about 750 °C for several minutes in vacuum. Subsequently, the getter is held usually at about 400 °C. The performance of the pump is normally judged by its pumping speed for hydrogen. Most common gases present in high-vacuum systems are pumped permanently, but hydrogen can be reemitted if the temperature is raised above 1000 °C. As in other gettering pumps, the total capacity for pumped gas is limited and after a number of activations, the pumping speed declines and the pump saturates. Pump specifications provide information about total accumulation capacity before the getter strip or ribbon becomes brittle and loses its mechanical integrity.


NEG pumps are available as nude cartridges, as glass-encased appendage capsules, or as autonomous flanged pumps that can be attached to any vacuum system. To provide a higher pumping speed in a smaller package, the getter ribbon is often pleated or corrugated (shown in Figure 16) and provided with a heater. Water cooling or even air cooling is not necessary for small pumps. A small pump (2.5 cm diameter, 7.5 cm long), if properly thermally shielded, can be heated in steady-state operation with a 10 watt heating source, Also, for pure hydrogen, the pump can be operated at a reduced speed at room temperature. Pumps are made in many sizes and a nude cartridge may produce 2000 l/s pumping speed for hydrogen. Typical pumping speeds are 65% for oxygen, 50% for carbon monoxide, and 17% for nitrogen as compared to hydrogen. The hydrocarbons are pumped with a speed that is one hundredth of the speed for carbon monoxide. Such ratios, however, can be misleading because they depend on the quantity of previously pumped gas. For example, Figures 17, 18, and 19 show the pumping speeds as a function of the amount of previously pumped gas.




Inert gases and methane are not pumped in significant amounts. Therefore, usually a small ion-getter pump is required to provide the pumping speed for these gases In this manner, ultrahigh-vacuum can be obtained in ultra-high-vacuum systems with the primary pumping done by NEG pumps, especially, when, after baking. the main residual gas is hydrogen. NEG pump strips are sometimes included for miles, along the entire length of tubing in particle accelerators, and as auxiliary pumping modules for enhance hydrogen pumping speed in autonomous ion-getter (sputter-ion) pumps, which may be called combination pumps.


Non evaporated getter pumps can be incorporated into ion-getter pumps to provide a higher pumping speed for hydrogen. For this purpose, it is convenient to select a NEG composition that can be activated at baking temperatures normally used in the process of obtaining ultra high-vacuum (300 – 400 °C). An example of such an integrated pump, using the NEG composition St707, is shown in Figure 20.


The normal range at the activation temperature for this compound is specified as 600 – 800 °C, but after 12 h bake at 350 °C, sufficient pumping action is obtained (after cooling in room temperature) as shown in Figure 21.


The combination pump performance, as expected, provides higher pumping speed for hydrogen, which is usually the main gas in an ion-getter pump system. The pumping speeds for hydrogen, for three different sized pumps, are shown in figure 22, where dashed curves represent pumps without NEG addition and the solid curves represent combination pumps.


The corresponding residual gas measurements are shown in Figure 23 and demonstrate a sharp reduction of the partial pressure of hydrogen.



The initial evacuation of high- and ultra high-vacuum systems that are subsequently pumped by ion pumps has two basic requirements: the ultimate pressure of the roughing pumps should be low (near 1×10-3 torr), and the rough pumps should not contaminate the system and then the ion pump which can cause starting and operating difficulties. An additional requirement is to use an ion pump with high-enough pumping speed or a system with low-enough outgassing rate to be able to reduce the pressure quickly after the high-vacuum valve is opened. It should be recalled that ion pumps have approximately one hundredth of the maximum-throughput capability of other high-vacuum pumps. System design and operation must reflect this characteristic.

With regard to the maximum practically permissible steady-state gas flow into the pump, the ion-getter pumps must be considered as a different class of equipment. Compared to throughput-type pumps of the same nominal pumping speed, they have nearly one thousand times lower throughput capability. This observation immediately suggests that the starting conditions, after the initial evacuation, must be different. It is common in textbooks and operating instructions to specify a pressure level where the pump can be started. This is an unfortunate choice that leads to errors in system design such as the incorrect selection of rough pump type or size, or the size of the high-vacuum pump. There are three items to consider and are illustrated in Figure 24.


The amount of gas initially entering the pump (after the high-vacuum valve is opened), depends on the volume of the chamber and the pressure (at the end of the roughing cycle), not the pressure alone. The bypass line shown in Figure 24 can help in softening the sudden rush of the pumped gas into the pump, but it does not change the two components of the gas contained and emanating from it: the gas in space and the gas emitted from the surface. The selection of the high-vacuum pump size will also depend on the frequency of evacuation events. With all this in mind, a simple gas accumulation estimate should be made before deciding on a particular roughing system or the size of the high-vacuum pump.


Even if some ion pumps can be started at pressures as high as 10 to 20 mtorr, it is not a good idea to do this too often because of the accelerated exhaustion of the cathodes. The cathodes of an ion pump, which are eroded by sputtering, can not be regenerated but must be replaced. This replacement is not a simple matter because the pump will has to be cut, open and then re-welded. The cathode life depends on the amount of gas pumped, on the integral of pressure and time. Typical values for diode pumps are 200 h at a constant pressure of 10-4 torr, 2000 h at 10-5 torr, 20,000 h at 10-6 torr, and so on. Note that at 10-6 torr the life is over 2 years and at 10-3 torr only 20 h! For pumping nitrogen only the life of a diode pump at 10-6 torr will be about 50,000 h (5.7 years). Corresponding values for a triode pump are 35,000 h, and for a StarCell pump, 80,000 h.

Ideally, ion pumps should be started at pressures near 1×10-5 torr, although in practice starting pressures near 10-3 torr are more common. There are several ways to accomplish this. It is possible to use oil-sealed mechanical pumps with well-designed and carefully operated trapping arrangement. Such pumps can produce pressures below 10-4 torr with a liquid nitrogen trap or a carefully degassed zeolite trap. However, oil-sealed mechanical pumps are rarely used for the initial evacuation of ion pump systems. Fear of accidental contamination of the system and of the ion pump is sufficient to discourage their use.


Sorption pumps are commonly used as roughing systems for ion pumps (Figure 26). To prolong the saturation period, air ejectors (Venturi pumps) or coarse oil-free mechanical pumps can be used at the beginning. They can remove about 90% of the air before sorption pumps are engaged. Recently developed oil-free high-vacuum mechanical pumps can be very useful. They can produce pressures below 20 mtorr and they can be left pumping on the system for longer periods of time without any possibility of contamination, allowing the system to degass. If the ion pump is large and the evacuation object small (e.g., a microwave tube), the oil-free mechanical pump can be used alone. In general, they should be used with a single stage of sorption pumping and/or with a titanium getter pump before the ion pump power is started.

In large, expensive installations it may be useful to evacuate the vacuum system to 10-5 torr or below using oil-free mechanical pump followed by cryopumps or turbodrag, or turbomolecular pumps before the system is switched to ion-getter pumps. One example of successful employment of these pumps in large vacuum systems is in the field of high-energy physics, where many of the modern particle accelerators are constructed using ultrahigh-vacuum techniques and are pumped by ion pumps.


1.  Y. Z. Hu and J. H. Leck, Vacuum, 37(10), 757 (1987).

2, D. J. Harra, J. Vac. Sci. Technol. (Jan./Feb. 1976).

3.  G. F. Weston, Ultrahigh Vacuum Practice, Butterworth & Company (Publishers) Ltd., London, 1985.

4.  P. A. Redhead, J. P. Hobson. and E. V. Kornelson, The Physical Basis of Ultrahigh Vacuum, Chapman & Hall, Ltd., London, 1968; reprinted by AIP, New York, 1993,

5, G. F. Weston, Vacuum, 34(6), 619 (1984).

6.  M. Audi and M. deSimon, Vacuum, 37(8/9), 629 (1987).

7.  T. Snouse, J. Vac. Sci. Technol., 8(1), 283 (1971).

8.  M. Audi and M. DeSimon, J. Vac, Sci. Technol, A6(3), 1205 (May/June, 1988).

9.  G.L. Saksaganskii, Getter and Getter-ion Vacuum Pumps, Harwood Academic Publications, Switzerland, 1994.

10.  K. M. Welch, Capture Pumping Technology, Pergamon Press Ltd., Oxford, 1991.

11. R. W. Roberts and T. A. Vanderslice, Ultrahigh Vacuum and its Applications, Prentice Hall, Englewood Cliffs, N.J., 1964.

12.  M. Audi et al., J. Vac. Sci. Technl, (A5) 4, 2587, (1987).

13.  P. Della Porta and B. Ferrario, Proc. 4th Int. Vac. Congr, AIP. 369, (1968).

14.  P. Della Porta a al., Trans. 8th Natl. Vac. Symp., 1961, Pergamon Press Ltd., Oxford, 229, 1962.