Proceedings of the 2017 CERN–Accelerator–School course on Vacuum for Particle Accelerators, Glumslov, (Sweden)

Ion getter pumps

C. Maccarrone, P. Manassero, C. Paolini

Agilent Technologies S.p.A., Via Fratelli Varian 54, 10040 Leinì, Italy

Abstract
Ion Getter Pumps (IGP) are used to create ultra-high-vacuum. The IGP
operation is triggered by the Penning cell structure, which uses a combination
of electrical and magnetic fields to confine electrons and start ionization and
getter pumping processes. During the years, more IGP configurations have
been developed to cope with different gases, such as reactive gases, hydrogen
or noble gases. When the IGP is required to reach pressures lower than 10
mbar, it needs to be baked out with the whole vacuum system, in order to
accelerate the release of the gas atoms trapped inside the pump materials, and
the leakage current issue has to be taken into account for a reliable pressure
reading.

Keywords
CERN report; ion pump; Agilent; IGP; vacuum

1 Introduction
Ion getter pumps (IGP) are devices able to create and maintain ultra-high-vacuum, reaching pressures
as low as 10 . Lower pressures are in principle achievable, but their measurement is
particularly challenging and strongly depends on the outgassing of the system on which the ion pump is
mounted.
With respect to other vacuum pumps, such as turbomolecular pumps or primary pumps, IGP
have some characteristics that make them unique.
First, ion pumps are closed pumps: they do not have any foreline. In fact, an IGP keeps inside
all the pumped gas: whatever is pumped by an IGP, will remain in it. This avoids the risk of venting the
system to which an IGP is connected, prevents any contamination which could come from the roughing
line and does not require a backing pump during operation, but only at start-up since IGP cannot be
started at atmospheric pressure.
Furthermore, the ion pump is a static device. It has no moving parts, so it is vibration free and,
consequently, it doesn’t need any lubricant which could be a source of contamination.

Available online at https://cas.web.cern.ch/previous-schools 1

 In this paper, we will describe how an Ion Getter Pump works, pointing out the pumping
mechanisms for different gases and referring to different pump types. Besides, some relevant topics will
be discussed, such as the outgassing process and the leakage current.

Fig. 1: IGP pressure range, compared with turbomolecular and primary pumps

2 IGP structure
– Element is made up of one or more Penning cells. It is the active part of the pump, by which the
gas is pumped. Referring to the standard element named Diode, it is made of cylindrical cells (the
anode) and two titanium plates (the cathode). The cathode is grounded, while the anode cells are
fed with positive high voltage, typically 3 - 5 - 7 kV.
– Magnets are positioned outside the pump body to complete the Penning cell structure, creating
an axial magnetic field (typically in the range 1200 ÷ 1400 Gauss).
– Feedthrough (F/T) is the electrical connector used to feed the pump with high voltage. The
central rod of the feedthrough is connected to the element (either to the anode or the cathode,
depending on the element type) by means of a metal tab or a wire.
– Inlet flange is used to connect the IGP and the vacuum chamber. The inlet flange diameter is one
of the parameters determining the conductance, which affects the pumping speed of an IGP.

Fig. 2: (a) Section of the Diode element, where the anode, the cathode and the magnets are indicated; (b) a
drawing of an IGP, where the F/T and the inlet flange are indicated.

2
3 Principles of operation

3.1 Ionization (of gas molecules)

The operating principle of IGP is based on the Penning cell [1], that is usually made of a stainless-steel
anode and two opposite titanium cathodes, which uses a combination of static electric and magnetic
fields to achieve confinement of electrons. The electrical field traps the electrons in the axial direction,
the magnetic field - in the radial direction.

Fig. 3: Representation of the Penning cell structure into an IGP.

The aim of the Penning cell is to trap electrons and increase as much as possible the probability
to ionize the residual gas, that will be pumped by means of the following mechanism:
– free electrons are produced by applying high voltage between the anode and the cathodes and
here are trapped into the Penning cell. This “trapping” increases the path of electrons from cathode
to anode by many orders of magnitude and hence increases the probability of a collision between
electrons and gas molecules;
– because of collisions, the trapped electrons ionize the background gas, thus creating positive ions
and extracted electrons;
– ions, positively charged, are attracted towards the cathodes;
– extracted electrons are trapped as well.
This mechanism can be considered the starting point for IGP operation and occurs in every IGP
independently from the element type. However, the pumping mechanism in its entirety changes in the
presence of different gases, since the ionization process has different consequences depending on the
properties of the gas atoms.

3.2 Pumping of chemically active (getterable) gases

Getterable gases, such as nitrogen ( ), oxygen ( ), carbon monoxide ( ) or carbon dioxide ( ),
are those gases which form chemical compounds with the surface of getter materials, such as titanium
(the material the cathode is made of).

3
When gas atoms are ionized and electrically attracted by cathodes, some are neutralized and buried into
the titanium cathodes. This mechanism, used by IGP to pump the gas, is called pumping at the cathode
or ionization pumping.

Fig. 4: Representation of the pumping mechanism for getterable gases

The ions, hitting the cathodes, cause the sputtering of the titanium atoms: they are emitted from
the cathodes and cover the anode surfaces, creating an active titanium film which chemically traps the
background gas molecules. This mechanism is named pumping at the anode or getter pumping.

3.2.1 Saturation of an IGP

While pumping at the anode mechanism is permanent, since it works forming chemical bonds, pumping
at the cathode is not stable: as the sputtering of the titanium atoms from the cathodes goes on, the
cathodes are eroded and the previously implanted atoms can be released.
So, the net pumping speed decreases until an equilibrium condition between ion implantation and
gas re-emission is reached. At equilibrium, the IGP is called saturated. This term is sometimes
misleading, since it can be confused with the end of the pump life.
Hence, when starting to use an IGP, its pumping speed will be higher than the nominal one,
because the pump is still unsaturated and the pumping at the cathode mechanism is still actively working.
The term saturated indicates the stable operation mode of an IGP, in which it reaches its nominal
pumping speed, and not the end of the pump life. The time the pump needs to reach the equilibrium
depends on the pressure: the higher is the pressure, the faster the equilibrium is reached (see Fig. 5).

Fig. 5: Time needed to saturate an IGP depends on the pressure

4
 The cathode erosion occurs not uniformly on the cathode plates, since the electric field lines carry
most of the ions in front of the anode cells centre, as shown in Fig. 6

Fig. 6: Titanium erosion under microscope (a) and a cathode plate after IGP long term use: the erosion
drills the cathode plate in correspondence of the centre of the anode cells (b)

3.3 Pumping of hydrogen

Since hydrogen is chemically reactive, it can be chemisorbed by the sputtered titanium film (pumping
at the anode).

Fig. 7: Representation of the pumping mechanism for hydrogen

Besides, hydrogen has a high solubility in titanium, so it diffuses in the cathode after implantation
and forms a solid solution with it. This improves the pumping at the cathode mechanism and leads a
decrease of the titanium sputtering yield; the erosion of the cathode is negligible and only a small
quantity of hydrogen is re-emitted into the pump. As a consequence, the pump remains almost
unsaturated for hydrogen and the pumping speed is roughly 50% to 100% higher than for nitrogen.

3.4 Pumping of noble gases

Noble gases are not chemically active, so they are not chemisorbed by the titanium film which covers
the anode.
They can be implanted in the cathode, but this pumping mechanism is not stable: after saturation,
noble gases are released back into the pump and a phenomenon usually called noble gas instability
occurs. Because of the gas release, the pressure increases of one or few decades. Then the gas is re-
implanted in the cathode and the pressure decreases without need of intervention on the system; the

5
situation is stable until the cathode erosion reaches again the implanted gas and another re-emission
happens, leading to another pressure increase.

Fig. 8: An example of noble gas instability

Therefore, a key indicator of noble gas instability occurrence is the periodicity of pressure peaks,
and the time needed to see such a behaviour depends on the pressure (the amount of pumped gas).
The most stable and efficient mechanism to pump noble gases occurs when gas ions are reflected
back from the cathodes as high energy neutrals. Hence, noble gases can be implanted (not chemisorbed)
permanently into the titanium that covers the anode. This titanium is not subject to sputtering
phenomena; the noble gases buried into it will therefore not be released again in the system.

Fig. 9: Representation of the pumping mechanism for noble gases

4 Agilent IGP elements

As discussed above, the IGP operating principle depends on the gas type. For this reason, different types
of IGP were developed to optimize the pump performances when pumping a particular gas type.
So, an IGP optimized for noble gases can be different from an IGP optimized for getterable gases.
In particular, ion pump types differ in their geometry, the shape of the element and the materials
used for the cathode.

6
4.1 Noble Diode
As discussed in Sec. 2.4, it is required to increase the probability of reflecting ions as high energy
neutrals to better pump noble gases. This goal can be achieved by choosing a different material for the
cathodes, with a higher atomic number, to increase the probability that this neutralization and reflection
occur for the noble gas ions. For this reason, Noble Diode IGPs have been developed. Their elements
contain one cathode made of tantalum (atomic number Z = 73) instead of titanium (Z = 22).
This solution increases both pumping speed and stability when pumping noble gases with respect
to a standard Diode pump. In fact, by increasing the probability of reflecting ions as neutrals, the most
efficient pumping mechanism is enhanced and noble gas instability is prevented. Otherwise, the
modified Noble Diode element leads to lower performances for hydrogen with respect to a Diode IGP,
since the solubility in tantalum is lower than in titanium.

Fig. 10: Schematic representation of the Noble Diode element.

4.2 Triode and StarCell®

While the Noble Diode element is conceived to better pump noble gases by changing the cathode
material, another approach is to change its geometry. Thus, the Triode IGP has been developed. Its
element has a cathode made of strips that are separated from the pump body.
For this element configuration, negative voltage is applied to the cathode while the anode is
grounded and the Penning cell ionization mechanism is still the same as in Diode ion pumps.
Because of the Triode element geometry the probability to pump the gas at the cathode, that leads
to instability for noble gases (see Sec. 2.4), is greatly reduced. On the contrary, ions hitting the cathode
with a glancing angle have an increased probability to be emitted as neutrals and, then, to be implanted
in the titanium sputtered on the anode and the pump walls, which become active components in pumping
the gas.

7
 Fig. 11: Schematic representation of the Triode element.

An improved version of the Triode element is the StarCell element [2], shown in Fig. 12, conceived and
developed by Agilent (formerly Varian) [1].
The shape of the cathode, which is made of stars with small bended wings, is optimized to
maximize the reflection of neutrals and to ensure a longer lifetime with respect to a Diode pump, since
titanium erosion is not centred in a few points (see Par. 2.2.1) but it is distributed more evenly.
With respect to the Noble Diode IGP, the StarCell IGP better pumps hydrogen since both cathodes
are made of titanium.

Fig. 12: Photo of the StarCell element

Table 1: Comparison of different Agilent IGPs performances for different gases

Noble
Gas StarCell a Diode a
Diode a
Hydrogen 2 3 1
Argon (or Helium) 4 1 3
Water 2 3 2
Methane 3 2 3
Nitrogen 2 3 3
O , CO, CO 2 3 3
a 1: poor; 2: good; 3: excellent; 4: outstanding

8
5 Lifetime and leakage current
The principal causes of the end of an IGP life are summarized in the following:
1) the erosion completely drills the cathode along its whole thickness, so the sputtering is no
more effective and can’t refresh the titanium film on the anode. This event is a direct result
of the IGP operating mechanism (discussed in Sec. 2).
The time needed to wear out the titanium depends on the cathode thickness and the working
pressure. It can be assumed that ten times higher is the pressure, ten times higher is the
sputtering rate and so ten times shorter is the IGP life.
For Agilent IGP, a rough evaluation of the operating life of a pump working at 1 ∙ 10 mbar
(nitrogen) is:
- 50000 hours for Diode and Noble Diode
- 80000 hours for StarCell pumps
2) A vacuum leak: if this does not happen at the beginning of the pump life, it is very unlikely
to occur on a long-term time scale.
3) Low magnetic field: if the magnetic field is not strong enough, the Penning cell is not able to
trap electrons to ionize gas atoms. It may occur if the magnets are overheated (e.g. during the
bakeout) and reach a temperature higher than their corresponding Curie temperature.
4) Noble gas instability: see Sec. 2.4. This phenomenon may prevent an efficient use of the ion
pump on the system.
5) Short circuit: it may occur in the case of complete metallization of the ceramics, which work
as electrical insulators through the anode and the parts of the pump at ground potential, or in
the case of a remarkable cathode deformation, that can shorten its distance from the anode.
The latter, for example, can be due to the absorption of a huge amount of hydrogen.
6) The leakage current, as discussed in the following section. Again, as in the case of the noble
gas instability, the IGP will continue pumping even if affected by a leakage current, but the
leakage current can make it less efficient as a pressure indicator and, in the most severe cases,
can cause the overheating of the pump and consequently increase its outgassing.

5.1 Leakage current and SEM element

Ion pumps are often used as pressure indicators, thanks to the well-known almost linear dependence of
pressure on the ion current, as long as the cell geometry, the number of cells, the voltage and the
magnetic field are taken as fixed parameters [3]. Unfortunately, some spurious currents, independent of
pressure, can arise in the ion pump; at low pressures, such currents can be comparable to or much higher
than the pressure-dependent ion current.
The leakage current does not actually affect the pumping efficiency at all, but when this
phenomenon occurs, the ion pump becomes useless in pressure reading; in the most dramatic case,
pressures below 10 mbar cannot be read.
The spurious current can arise from both external sources, such as power supply, the connecting
cable or the high-voltage feedthrough, and internal sources. Among internal sources, the metallization
of ceramic insulators during operation is potentially responsible for this phenomenon.
However, the main cause of leakage current occurrence is the Field Electron Emission (FEE)
from cathodes, by which free electrons are emitted from the titanium surface when strong electric field
is applied.

9
 Therefore, the leakage current can limit the use of an ion pump as a pressure indicator, especially
in low pressure applications (lower than 10 mbar). In this pressure range, the pumping current,
proportional to the pressure, is comparable to or, in some cases, can be even lower than the leakage
current due to FEE.
In order to reduce leakage current, a patented [4] anode structure was developed by
Agilent, in which interstitial cavities between contiguous cells were eliminated, as shown in Fig 13.

Fig. 13: Patented SEM element

Studies have shown that a greater number of dendrites (whiskers), which are sources of field
emission, are created in the interstitial areas between the cells. This optimized element is usually called
“SEM element” since it is very often employed in scanning electron microscope applications, where the
ion pump current is used as a pressure set point to protect the microscope gun (filament or microtip
electron emitters) in case of overpressure. It allows a reliable pressure indication, converted from the
ion pump current reading, down to the UHV pressure range.

6 Outgassing
In solids, about 1 atom on 1000 is a gas atom. Part of this gas is trapped in the atomic structure of the
matter, while the other part is on the surface. This last amount of gas, which has been adsorbed at
atmospheric pressure, can be released by desorption process when the surface is exposed to vacuum.
Hence, the outgassing is an infinite source of gas which limits the base pressure achievable by a
vacuum system. The first way to cope with outgassing is to choose low-outgassing materials, but this
solution is not sufficient to reach ultra-high-vacuum level. Since we know that outgassing quickly
increases with temperature and slowly decreases in time, vacuum systems and vacuum components in
general, and ion pumps among them, are usually baked at ~250 , in order to accelerate the release of
adsorbed and trapped gas atoms. It is worth noticing that the bakeout can be effective only if performed
on all the vacuum system components. In fact, it is of course not sufficient to bake only an IGP.

10
 As shown in Fig. 14, during the bakeout the pressure increases as a consequence of the enhanced
outgassing process, but once the system and the pump are cooled to room temperature, an improvement
of the reached base pressure is observed with respect to the pressure that would be achieved by the
system in the same time without the bakeout.

Fig. 14: Plot representing the pressure as a function of time when the pump is backed out

Another way to limit the outgassing rate during the pump operation is to provide preliminary
thermal treatments to the pump components. In particular, the Vacuum Firing process, during which the
material is baked at high temperature (e.g. 950 °C in the case of stainless steel) in a vacuum furnace, is
used to eliminate hydrogen atoms from cathode and anode surfaces (in some cases from the whole pump
body) before IGP is assembled and finally baked out.
For Agilent VIP 200, which was launched in the market in 2016, the Vacuum Firing process has
been applied to all the surfaces exposed to vacuum, in order to further reduce the hydrogen outgassing
rate and to accelerate the pumpdown to ultimate pressure.

7 Conclusions
Since their invention in 1957, ion pumps have continuously moved towards lower pressures and are still
today the technology of choice for most ultra-high vacuum (UHV) applications, in both research and
industry. This includes a large variety of vacuum systems including particle accelerators, synchrotron
light sources, gravitational wave detectors, electron microscopes, surface analysis and medical
equipment.

References
[1] Kimo M. Welch, Capture pumping technology, Brookhaven National Laboratory (1991)
[2] M. Pierini, L.Dolcino, A new sputter ion-pump element, Varian Associates (1982)
[3] M. Audi, C.Paolini, C. Maccarrone, The phenomenon of leakage current in ion pumps, Agilent
Technical Note (2017)
[4] J. B. McGinn and M. Spagnol, U.S. Patent No. 6,388,385 B1 (May 14, 2002).

11