Secondary Vacuum Pumps (Momentum Transfer Pumps)
5.3 Turbo-molecular pump
Like the molecular drag pump, the turbomolecular pump exploits the interaction of gas molecules with high-speed surfaces to direct them towards an outlet. But in this case the surfaces are the inclined surfaces created by machining angled slots radially inwards from the edge of a circular disc to form a bladed rotor that resembles the multiple-bladed structure of an axial flow turbine. Between the vanes of the rotor and the vanes of the stator there are very narrow slits through which the gas molecules are driven towards the pre-vacuum flange. In the case of the original molecular pump by Gaede and following models, these slits where so narrow that the pumps were difficult to build and made them sensitive to mechanical contaminations like dust grains etc. Later the distances between rotor and stator vanes were increased to a few millimetres, which increased also the tolerances for the production, and a higher degree of reliability could be reached.
Fig1. Schematic of turbo-molecular pump
The speed of the external parts of the rotor vanes should be of the same order of magnitude as the mean velocity of the molecules to be ex-hausted, otherwise there is no sufficient momentum transfer from the vanes to the molecules. For room temperature and for N2 the mean velocity is about 500 m/s. If we want the speed of the vanes of a rotor radius of e.g. r = 5cm to have this value, then the number of revolutions must be: v = ωr = 2πfr, f = v/2πr = 500/(2π)(0.05) = 1.6x103 s-1 = 90000 rev/min. This is indeed of the same order of magnitude of the usual rotary speed of TMPs, which ranges between about 30000 rev/min and 100000 rev/min. Under these conditions the turbine must of course be balanced as well as possible.

Each rotor is paired with a similar set of stationary stator blades that project inwards from the wall of the cylindrical pump housing, as shown in Fig.1. The flow direction is axial, and although the structure resembles a turbine arrangement, it should be emphasized that the pump operates in the molecular flow regime. The blades are not aerodynamically shaped as they would be for hydrodynamic flow, but are simply flat-sided, as results from the machining process.
The pumping speeds of typical drag pumps are of the order of 10 ls−1, those of comparable turbomolecular pumps are of the order 100 ls−1. Another advantage it has, compared with the drag pump, is that there is no requirement for a very small gap between the high-speed rotor and stationary parts to act as a seal; the gap between the rotor blades and the housing in the turbo pump does not have that role and can be as large as 1mm, eliminating the possibility of mechanical seizure.
The essence of the pumping action is that the slanting blades, in moving at high speed through the gas, intercept more molecules on their downward facing side than if they were stationary and tend to project them in an axial direction. This action is duplicated by the stator blades that receive molecules that have gained a velocity increment in rebounding from the rotor.
Some models of TMPs are equipped with magnetic bearings or air bearings so that there is practically no friction in the bearings. In the case of magnetic bearings, the turbine and the motor are actively held in all three directions by electromagnets. One axial bearing at the top of the stator hovers the rotor, two radial bearings take care of the lateral position and sta-bilisation of the rotor. Each bearing is controlled by highly sensitive sensors. The magnetic suspension does not need any lubrication so that the pump is free of hydrocarbons.
A typical pump speed characteristic of a TMP is shown in Fig.2. The pump speed is constant over a wide range and decreases for exhaust pressures above 10-2 mbar since there we approach the transitions region towards the viscous flow range. There also the friction of the rotor vanes with the air becomes too large. The pump speed also depends on the sort of gas, since the transfer of momentum from the vanes to heavier molecules is larger.
Fig.2 Pump speed characteristics of typical Turbo molecular pump pump

Ultra High Vacuum Pumps (Capture Pumps)
5.4 Sputter Ion Pump
The pump effect of SIPs is based on sorption phenomena which are triggered by ions in a Penning-discharge. On one side, the gas molecules to be exhausted are ionised, on the other side they are adsorbed in Ti-films. This works in the following way:
Between two parallel cathode plates of Ti there are many tightly packed cylindrical hollow anodes of stainless steel, the axes of which are oriented perpendicular to the cathode plates (Fig.1).
Fig1. Principle of operation of diode sputter-ion pump
The voltage between these anodes and the two cathode plates is a few kV. In addition the whole system is embedded in a homogeneous magnetic field of about 0.1T, which is oriented parallel to the anode axes. The magnetic field strength is adjusted in such a way that the gyroradius of the electrons is smaller than the diameter of the anode cylinders so that only a few electrons can reach the anodes. Thereby in each anode cylinder a dense cloud of electrons is formed (ne 1013/cm3), which in axial direction are confined by the electric field between the anodes and the cathode plates. In such an electron cloud the probability for ionising collisions with gas molecules is high even for low gas densities, and enough electron-ion pairs are formed so that the electron density remains constant. The ions are not sufficiently confined by the magnetic field, but are accelerated towards the cathodes and bombard them.
The discharge current I is a function of the neutral density n0 of gas molecules, of the electron density ne, the length l of the discharge length and of the ionisation cross section . Since the discharge current is a function of the neutral density n0 of gas molecules, it can be taken as a measure for the pressure. This is the same as for a Penning vacuum meter.
The ions bombard the cathodes of Ti and sputter them because of their high kinetic energy. Thus on other surfaces of the vacuum chamber Ti-films are formed which act as getter films, binding gas molecules. In addition the kinetic energy of the ions is so high that they penetrate deeply into the cathodes, where they are absorbed by implantation. This sorption effect is effective for all sorts of ions, also for rare gases. Moreover, gas molecules can also be con-fined by coverage by getter films.
Fig2. Electrode configuration in a diode sputter ion pump
For diode sputter pump the electrode arrangement is shown in fig.2. The getter films are formed on the anode surfaces and on the cathode plates among the regions where the sputtering takes place. The implantation of the ions occurs inside the cathode plates. However, when the sputtering of the cathodes continues, the implanted gas molecules are released again. Especially for rare gases, which can only be exhausted by ion confinement underneath getter films, the pump effect will decrease again after a while. In order to prevent this, triode pumps have been developed which have transparent (grid) cathodes. These types of SIPs can also pump rare gases without drop of the pump speed.
The pump speed of a SIP depends on the sort of gas. For air, nitrogen, CO2 and water vapour, the pump speeds are almost identical. With respect to the pump speed of air, the pump speeds of such a pump for other gases are: Hydrogen 150 - 200%, methane 100%, other light hydrocarbons 80 - 120%, oxygen 80% , argon 30%, helium 28% . Argon is exhausted with a stable pump speed still at exhaust pressures of 10-5 mbar. The pumps can be started without problems at pressures of more than 10-2 mbar and can be used permanently at 5x10-5 mbar.
The high magnetic field, which is necessary for these pumps causes inevitable stray fields in the surrounding. If this is perturbing, these can be screened off.