Vacuum Techniques

Introduction
Vacuum technology is necessary to cryogenics. The vacuum techniques are needed, for
example, to eliminate gas convection in cryostats, to open thermal switches, to decrease the
pressure above the surface of a liquid to lower its temperature , to circulate helium in dilution
refrigerators, for the purpose of leak detection . The ultimate performance of any refrigerator
depends on the right choice of pumps and connecting tubes.

A cryogenist must be sufficiently experienced in vacuum physics to be able of projecting

leak tight vacuum equipments, of avoiding gas leaks, or if they occur, of locating and getting
rid of them. Hence, the cryogenist must know how to solder and glue. As a matter of fact,
cryogenic equipments undergo very high mechanical stress due to great thermal gradients and
due to different expansion coefficients of the various materials used. A peculiar characteristic
of vacuum technologies as applied to cryogenics is that vacuum equipments are built and
tested at room temperature, then evacuated and cooled. Defects that may develop cannot be
fixed at low temperature. Moreover, some problems that occur at low temperatures disappear
when the equipment has been warmed up again: hence, they cannot be easily located.
Let us remember also that cold surfaces adsorb gases: if a small leak to atmosphere exists,
air will condense on cooled surfaces. If the amount of gas adsorbed is large, during warm up,
the pressure in the ‘vacuum’ space may become very high. A release valve must be therefore
present in the system.
A container is said to be under vacuum when the inside pressure is lower than the outer
one, usually the atmospheric pressure. If no container exists, vacuum is a space region at a
pressure lower than the atmospheric pressure.
Vacuum may be ‘natural’ or artificially produced. Natural vacuum, for example, occurs on
the lunar surface or in the interstellar space where one should however speak of numerical
density of particles (∼1 particle/cm3) instead of pressure. In the intergalactic space, the
density is around 1 particle/m3. Natural vacuum laboratories in use are, for example, the
Space Shuttle or Space Stations, but we will deal only with artificial vacuum, produced by
pumps inside a container.
Vacuum can be divided into three regions depending on the pressure of the gas:

• low or rough vacuum (from the atmospheric pressure down to about 1 Pa);
• high vacuum (1 Pa–10-4 Pa);
• ultrahigh vacuum (below 10−4 Pa).

Such more or less arbitrary classification reflects three different physical situations.
In the low vacuum range, the number of molecules in the volume is much larger than that
adsorbed on the internal surface of the container. In the high-vacuum range, the mean free
path of the molecules is of the order or larger than the dimensions of the container, and most
of molecules are on the container inner walls. In the ultrahigh vacuum range, the flux of
molecules onto an initially clean surface is so low that there is enough time to carry out
experiments before it is covered by a monomolecular layer of gas.
The simplest vacuum system consists of a vacuum chamber, an interconnecting tube
(Vacuum line or pipe) and a vacuum pump which produces a pressure gradient along the
tube.
To obtain a high or ultrahigh vacuum, both the chamber and the tube must be clean,
since impurities (like water, with its high vapour pressure, and its unavoidable presence
in the air) slacken the reaching of the final vacuum.
Let us also anticipate that, if a vacuum chamber initially contains air at the atmospheric
pressure , during the pumping process the composition remains approximately the same in the
low-vacuum range. Then, the composition changes, becoming usually richer in light
molecules.

Vapour pressure
Any liquid or even solid material always produces a gaseous phase in equilibrium with
the denser phase. The pressure of the gaseous phase is called ‘vapour pressure’. The final
vacuum in an evacuated chamber is often controlled by the vapour pressure of the most
volatile material present in the system.
The dependence on temperature of the vapour pressure p can be approximately
expressed as:
p ∝ exp (−L(T)/RT)
where L(T) is the molar latent heat of evaporation.

Mean free path and viscosity

In the kinetic theory, the gas molecules are represented by hard spheres colliding
elastically with each other and with the container walls.An important parameter that can be
calculated by this model is λ, the mean free path of a molecule between collisions. The mean
free path λ of molecules is:
where d is the equivalent hard-sphere diameter of the molecule (d = 0.22nm for He 4 [6]) and
n is the number of molecule for unit volume.

Flow regimes in vacuum

When a gas is removed from a container through a tube, the type of gas flow depends on
pressure. In the low vacuum range, the gas flow is controlled by the collisions among
molecules (viscous flow). If some molecules are removed from a region (by a pump), other
molecules will refill the ‘empty’ region. In this situation, the diameter of the pumping line is
not crucial. In the high- and ultrahigh-vacuum range (molecular range), the mean free path is
larger than the dimensions of the container, and interactions among molecules are rare. The
pumping tube diameter must be large to increase the probability that molecules reach the
pump. For a tube of radius r, the flow is viscous if /r < 10-2 and is molecular if /r > 1.
Gas flowin vacuum systems may fall into one of the three regimes.
These are:
1.Viscous Flow:
Distance between molecules is small; collisions between molecules dominate; flow
through momentum transfer
2.Transition Flow:
Region between viscous and molecular flow
3.Molecular Flow:
Distance between molecules is large; collisions between molecules and wall dominate;
flow through random motion
Vacuum pumps
The main components of a vacuum system are the pumps. The types of pumps most
commonly used in low-temperature experiments are:

1. Roughing pumps: pressure range 105–1 Pa

• Rotary vane pumps;

• Booster (Roots) pumps; • Scroll pumps;
• Sorption pumps (cryopumps).

2. High-vacuum pumps: pressure range 100–10−6 Pa

• Oil diffusion pumps;

• Turbomolecular pumps;
• Molecular drag pumps.

To achieve very low pressures (high vacuum), at least two pumps (one roughing pump and
a high-vacuum pump) are connected is series. To get high pumping speed, a parallel of
pumps both in rough and in high vacuum is sometimes used. Nowadays, the vacuum industry
offers integrated vacuum systems containing several vacuum pumps in order to cover an
extended range of operating pressure with a compact apparatus.
In the following sections, the functioning principles of the single pumps used in cryogenics
will be described.

1.Rotary vane oil-sealed mechanical pump

A volume of gas is enclosed in the space bounded by the rotor, the stator and the two vanes
(see Fig. ). The pump removes the gas by compressing it to a pressure slightly higher than the
atmospheric pressure. This overpressure opens the spring-loaded outlet valve, and the gas
escapes to the atmosphere. A thin film of oil makes the final seal; therefore the ultimate
pressure depends also on the oil vapour pressure. With one stage, the lowest attainable
pressure is about 10-2 torr and with two stages in series 10-3 torr.

Due to the friction of the sliding vanes, the biggest pump size available (water cooled) is
about 400 l/s. Smaller pumps (0.5–5 l/s) are air cooled.
Condensation of vapours must be avoided since it causes a deterioration of the oil: water
vapour condense at 234 torr at 70°C, which is the typical working temperature of the pump.
To avoid condensation, most pumps have a ‘ballast’ valve in the high-pressure stage. After
the gas has been closed off from the inlet, some gas is bled in through the ballast valve. This
causes the discharge valve to open sooner, with a reduction of the compression ratio. Rotary
piston pumps have similar characteristics.
Most mechanical pumps exhibit vibration that may represent a crucial drawback for very
low-temperature refrigerators.
Another drawback of oil-sealed pumps is the back-streaming of oil vapour into the
roughing line, which may occur at low pressure. Contamination by back-streaming oil can be
drastically reduced by using proper traps like molecular sieve traps with zeolite
The speed of rotary vane pumps is nearly constant from 1 atm down to 10−2 torr. To obtain
higher speeds in the 10−2–10−3 torr range, booster pumps are used in series with rotary vane
pumps.

2.Booster(Roots ) pump
The booster (blower or Roots) pumps are high-throughput pumps with a low compression
ratio. A booster pump is schematically shown in Fig. below. The gap between the two
counter-rotating rotors is of the order of 0.1 mm. The rotation (about 50 rps) of the rotors,
mounted on parallel shafts, is synchronized by gears to avoid contact. No oil is used to seal
this gap. The starting pressure of booster pumps is about 10 torr. Heating of the pump
becomes excessive when it is operated continuously at high pressure. Therefore, a bypass is
used for high-pressure roughing. Booster pumps with throughput up to 600 l/s are available.
To overcome the problem of the low compression ratio, multistage systems are produced in
which several booster units are cascaded. The system presents both an overall high
compression ratio and a high throughput. Such multistage system is capable of reaching 10−2
torr and works from the atmospheric pressure without the need of rouging pump as in the
case of a single booster pump. The throughput can be up to about 300 l/s.

A system of cascaded booster pumps has the characteristics of a medium-size rotary vane
pump but does not present the drawback of the back-streaming. For this reason, it finds
application in cryogenics, for example, for the circulation of the He mixture in dilution
refrigerators.

3.Scroll pumps
A scroll vacuum pump uses two interleaved Archimedean spiral-shaped scrolls to pump or
compress gases (see Fig. below). One of the scrolls is fixed, while the other orbits
eccentrically without rotating, thereby trapping and compressing gases between the scrolls
and moving it towards the outlet.
The device was first patented in 1905 but did not become efficient for 50 years.
Also for these pumps, multistage systems (e.g. Varian TriScroll Pump) are used which
allows for final pressure around 10−2 torr and pumping speed of about 30m3/h. The typical
application of scroll pumps systems is the backing for turbomolecular pumps.
 There are two main drawbacks in scroll pumps: the production of Polytetrafluoroethylene
(PTFE) powder dust due to the friction between the fixed and the moving part; moreover,
since in the present commercial versions the motor is not in vacuum, the tightness is not the
best.

4.Sorption pumps
In a sorption pump, the gas is trapped within the adsorbing material (zeolites or active
charcoal) called molecular sieve. Zeolites are porous aluminium silicates which adsorb large
amount of gas when cooled to low temperature (usually 77K). The pump is filled with zeolite
and put in a bucket containing liquid nitrogen.

Gases that are condensable at 77K are trapped by ‘cryocondensation’. (‘Cryosorption’ is

instead the trapping of gas with a lower melting temperature inside the pores of the molecular
sieve that has a huge surface/volume ratio, typically about 700m2/cm3.)
If gases like H2, He and Ne are to be trapped, the pump must be cooled to 4.2 K. Ultimate
pressure of the order of 10−5 torr can be achieved.
When cooled with liquid He, this type of pump can reach pressure below 10−8 torr.
The sorption pumps are ‘clean’ but are ‘one shot’, that is, two pumps in parallel and
connected
by valves alternatively are needed for a continuous pumping. When the first pump
is saturated, the second pump is started, while the first is regenerated: removing the liquid
nitrogen, the trapped gas is expelled through the blow-off valve. The pump (with zeolite) is
heated to 200–300°C to remove water vapour. Charcoal pumps are heated to about 100°C.
Another advantage of this kind of pumps is the very low level of vibration.

5.Oil diffusion pumps

In a diffusion pump, the dense oil vapour produced by the boiler (see Fig. below) is ejected
into the vacuum at high (or even supersonic) speed through the nozzles.
It collides with the gas communicating a large downward momentum to the gas
molecules. The top jet operates at the lowest pressure. The body of the pump is typically
cooled by water or air. When the vapour reaches the cooled walls, it condenses back into a
fluid which drops into the pump boiler for reboiling.

A cold cap is usually mounted on the top of the pump assembly to prevent vapour from
reaching the vacuum chamber. A thermal protection switch is often used. The maximum
working pressure of a diffusion pump is about 10−3 torr. The ultimate pressure of a diffusion
pump can be around 10−9 torr and heavily depends on the oil vapour pressure (p < 10−8 torr at
room temperature for very good oils). Pumps with very large pumping speed (up to 104 l/s)
are commercially available.
A drawback of oil diffusion pumps is the so-called back-steaming. It is the flow of a small
quantity of oil vapour towards the inlet of the pump. A water-cooled baffle like that shown in
Fig. can be put above the inlet. Baffles are made up of arrays of optically dense fins cooled
by a continuous water flow. A baffle always reduces the pumping speed.

To reduce to a minimum the amount of fluid reaching the vacuum chamber, the use of a
liquid N2-cooled baffle is the best choice.
 A diffusion pump needs a forepump (usually a rotary pump). Oil can also migrate from the
forepump into the vacuum chamber through the diffusion pump, when the rotary pump works
in the molecular regime (∼10−2 torr).
If an operating diffusion pump is exposed to atmosphere, even for a short time, a strong
oxidation of the fluid takes place. In extreme cases, combustion or explosion is possible.
The pressure in the foreline must be kept below a value called critical forepressure. If this
pressure is exceeded, oil vapour will be injected into the vacuum chamber in great amounts.
A ‘dumping’ of the pump can occur. For this reason, the size of the rotary pump must be
properly chosen.
Note that a decrease of the pumping speed, due to a reduction of the heater power to the
boiler, decreases the critical forepressure.

6. Turbomolecular pumps
Turbomolecular (turbo) pumps are very clean (especially magnetically levitated version)
mechanical pumps, with pumping speed up to more than 7000 l/s.
The pumping action is due to a high-speed rotating surface that transfers momentum to the
gas molecules (see Fig. below).

With metal gaskets and moderate bakeout, turbo pumps can reach pressure below 10−9 torr
without traps. They can be started at a pressure up to 1 torr. The time required to reach full
pumping speed (∼1 min) is much shorter than for diffusion pumps. Also these pumps must be
backed by a primary pump.
The turbo pumps are made up of 10–40 rotating (rotor) and fixed (stator) disks
alternatively arranged. Each disk has 20–60 blades with proper tilt.
The motor (usually working in vacuum) is moved by a special current power supply. The
rotor turns at 104–105 rpm, usually a multiple of the line frequency. The pumping speed of a
turbo pump unit depends on its rotational speed. High-speed turbo pumps need more frequent
maintenance interventions. In some turbo pumps, a ‘low-speed mode’ allows operation up to
10−1 torr. However, full rotational speed is achieved at pressuresbelow 10−3 torr. Low-level
vibrations in kHz range are always produced due to residual unbalance of the rotor.
The pumping speed of a turbo pump is almost constant over a wide range of pressure and
depends on the gas species.
Light gases with higher thermal velocity are pumped less than heavier ones. This is why
turbo pumps produce an (almost) oil-free vacuum. The lubrication of turbo pumps is made
with a special vacuum grease. Pumps with magnetically levitated rotor are available, but they
are more expensive.

Pressure gauges
The precise measurement of the total or partial pressure of gas is an extremely arduous
task. Pumps, pressure gauges and vacuum chamber walls deeply influence the total pressure
and composition of gas under measure; selective pumping, chemical reactions and physical
processes continuously change the gas parameters, and in most cases only an approximate
knowledge of the gas pressure is possible. Fortunately, the order of magnitude of pressure is
often an adequate information in most problems. An accuracy of ±10% is usually considered
quite good. A further difficulty occurs when the pressure of a gas at low temperature is to be
known: the gage is usually at room temperature, and gradients of temperature and pressure
occur along the tube connecting the measuring head to the low-temperature gas.

1. Total-pressure gages
We shall describe only a few of the most commonly used total-pressure gages. All gages
described, except the McLeod and the diaphragm gage, measure density rather than pressure.

2. McLeod gage
The McLeod gage is shown in Fig.below. By rising the mercury reservoir, the gas in the
volume V at the pressure p to be measured is trapped in the bulb B.
A further rising of the reservoir causes a compression of the gas in the capillary C (closed).
Capillary D is open and connected to the vacuum system. The difference dh between the two
mercury heights corresponds to a pressure difference dp = µg·dh (dh in mm gives
numerically dp in torr); µ is the density of mercury. If the compression of the gas in B and C
is isothermal, we can write:

The McLeod gage is a primary vacuometer; moreover its readings are independent of the
type of gas, except condensable vapours. It covers a wide range of pressures with a good
accuracy (±10−4 torr around 0.1 torr, ±2×10−7 torr around 10−6 torr) .
The main drawback is represented by the fact that a continuous monitoring of the pressure
is impossible. Bakeable McLeod gages have been built .
The main application of this vacuometers is for calibration of other gages.
 Scheme of a McLeod pressure gauge

3. Bourdon gage
It is a simple and reliable gage covering six decades of pressure (not with one gage).
Below 1 torr, its sensitivity is very low. The gage is usually made up of a bended flexible
metallic tube (see Fig.below) of elliptic cross-section. When pressure inside the tube is
different for the outside (atmospheric) pressure, the tube bends. A gear and lever system
moves the needle.

Bourdon-type pressure gauge

4. Diaphragm gage
It measures differential pressures. A thin diaphragm separates two chambers (see
Fig. below). One of the two chambers is kept at a constant pressure. A pressure difference
causes the bending of the diaphragm which forms, together with a fixed electrode, a
capacitor. The change in capacitance is measured by a very sensitive capacitance meter
(measurement of _C ≈ 10−19 Farad are possible).
Pressure down to 10−4 torr can be measured. The accuracy of the diaphragm gage is
limited by the dependence of the capacitance on temperature (typical 1% of full scale).
The time response can be around 10−3 s. Bakeable units are commercially available.
 Diaphragm capacitance pressure gauge

5. Thermal conductivity gages

In these gages, a wire inside the gas whose pressure is to be measured, is electrically
heated by a constant power (see Fig. below). As the gas density decreases, the heat loss from
the filament to the envelope walls decreases and hence the filament temperature increases
(not linearly). The temperature (200–300 C) is read by a thermocouple in thermal contact
with the wire.

Thermal Conductivity gauge

Pirani gage is similar: here the change in temperature of the wire is measured by the
variation of the wire electrical resistance. The lowest measurable pressure is around 10−3 torr.
The calibration (usually for dry air) is a function of the heat conductivity of the gas.

6. Hot cathode ionization gage

This is a high-vacuum gage. Figure shows the ‘triode’ and inverted configuration gage
with indicative biasing. As shown in Fig, electrons emitted by the filament are accelerated
towards the grid; they collide with the gas molecules and ionize them. The ions are attracted
by the collector. The ion current is a measure of the gas density. The ratio of the ion current
to the pressure is called sensitivity of the gage (ampere/torr). It depends primarily on the
geometry of the gage, biasing and ionization cross-sections of the gas. The ionization
efficiency of electrons for some gases as a function of energy is shown in Fig.
 When the grid is bombarded by electrons, soft X-rays are emitted that, in turn, produce
photoelectrons at the collector. This phenomenon limits the minimum pressure that can be
measured to about 10−8 torr.
The inverted configuration of Fig. (b) reduces the contribution of the photoelectron
current, extending the range to about 10−10 torr.
Due to the presence of a high-temperature filament, frequent recalibrations are necessary.

Ion Gauges:(a)convenyional ion gauge: (b)inverted ion gauge(Baayard-Allpert gauge)