Vacuum Technology

Foreword

Underlying the deposition and etch capabilities is a robust understanding of vacuum technology. Nordiko systems are thoughtfully designed and manufactured and assembled with care. They provide our customers with advanced capabilities and exhibit excellent longevity.

Introduction

With the passing years our modern day lives are impacted by the use of vacuum technology to an ever increasing extent.  From smart phones, computers, the cars we drive, communication systems to health care, many elements of life are touched by the application of the technology.  Most thin film technologies rely on a vacuum environment to facilitate the fabrication process for devices and assemblies. Here we shall concentrate on the generation and measurement of vacuum used for thin film applications.

To create a vacuum we use pumps to remove the air from an enclosure.  One can envisage two types of vacuum system, or vessel.  One that is closed; it is evacuated and sealed.  Good examples of these are: a vacuum Dewar, or thermos flask, a old fashioned cathode ray television screen, or a thermionic valve from an old radio.  The latter are seeing a resurgence in niche tube driven HiFi amplification.  These are generally glass vessels that are evacuated to about 1.3 x 10-4 Pa.  They are sealed containing a vacuum getter to help maintain the vacuum level.  The second type of vessel is usually metal, considerably larger than the sealed units and is continuously pumped.  These pumped systems are tools in which processes are performed.

In general, vacuum technology is not that complicated, but is a rather niche discipline, hidden from the layman.  The needs of the semiconductor sector has driven the advances in this area to benefit itself, affiliated disciplines and other industries.  The requirement for low defect levels in devices has pushed the development of pumps that provide a low level of vacuum, together with very low levels of contamination.  Contamination typically takes the form of particulates and residual vapours, such as hydrocarbons.

Basic Physics

It is worthwhile to summarise some basic physics relating the behaviour of gases.  These are well covered in very good text books and shall simply be stated here.

The gas laws:

Boyle’s law (1662):  In a closed system, the pressure of a given mass of an ideal gas is inversely proportional to its volume at a constant temperature.

This can be expressed as:

Where: p1 is the starting pressure

V1 is the starting volume

p2 is the final pressure

V2 is the final volume

Charles’s law (1787):  In a closed system, the volume of an ideal gas at constant pressure is directly proportional to the absolute temperature.

This can be expressed as:

Where: V1 is the starting volume

T1 is the starting temperature

V2 is the starting volume

T2 is the starting temperature

Gay-Lussac’s law (1809): If the temperature of a container is increased, the pressure increases.

If the temperature of a container is decreased, the pressure decreases.

This can be expressed as:

Where: p1 is the starting pressure

T2 is the final temperature

p2 is the final pressure

T1 is the starting temperature

Avogadro’s law (1811):  Equal volumes of all gases, at the same temperature and pressure, contain the same number of molecules.

This can be expressed as:

Where: V1 is the starting pressure

n1 is the staring number of molecules

V2 is the final pressure

n2 is the final number of molecules

From these the combined ideal gas law can be derived:

Where: p is the volume (in Pa)

V is the pressure (in m3)

n is the number of molecules (just a number)

R is the ideal gas constant (8.31441 J K-1 mol-1)

T is the temperature (in Kelvin)

Avogadro’s number tells us the number of molecules in a mole of gas.  From the idea gas law above we can quickly see that pressure of a volume of gas is related the number density of molecules.  Even if we generate a level of vacuum of 1 x 10-10 Pa (EHV) each litre (US quart) still contains about 3 x 104 molecules; that is 30,000 in a each cubic centimetre.  The lowest vacuum normally achieved in practical process equipment is generally around 1 x 10-6 Pa.

It is also possible to derive these laws using molecular kinetic theory of gases.  The kinetic theory, initially developed by Bernoulli, then later expanded upon by Maxwell-Boltzmann and others is covered in several texts.  The interested reader is referred to The Mathematical Theory of Non-uniform Gases by Chapman and Cowling.

Practical Pumping

To remove the air from a system we use pumps.  These pumps are connected by a series of ports and pipes.  Commonly we use two types of pumps; compression pumps and capture pumps.  In either instance a single pump is not capable of evacuating a chamber from atmosphere to the desired level of vacuum.  Typically two types of pump are used in a complementary fashion to achieve the whole evacuation process.

In order to take advantage of high vacuum pumps a separate means of reducing the pressure from atmosphere (105 Pa) to 10 Pa, or thereabouts is needed.  To achieve this a mechanical displacement pump is used.  For a detailed analysis of the various types of pumps in this category the reader is referred to a text by Bello.

Two approaches to achieving high vacuum in process systems are in common usage.  The first uses a dry mechanical pump to achieve the initial evacuation.  This stage is called roughing.  A turbomolecular pump is then used to continue the process.  This second pump cannot exhaust directly to atmosphere and so the mechanical pump is used in series, as a backing pump.  Figure 1 illustrates the pumping schematic of two such arrangements.  In A, the high vacuum pump is coupled directly to the chamber.  Between it and the mechanical pump there is an isolation valve.  In this arrangement, to evacuate the chamber from atmosphere, the isolation valve is opened and the two pumps are turned on together.

Where the chamber is large and the mechanical pump cannot evacuate it quickly enough to prevent the turbomolecular pump from stalling, then a delay is introduced before it is started.  In B, there are two additional valves.  Between the chamber and the turbomolecular pump there is a high vacuum isolation valve.  There is also a second pipeline valve.  This is a bypass arrangement where the valve may be switched to allow the mechanical pump to back the turbomolecular pump, or directly pump the chamber.  The advantage of A is that it provides a higher effective pumping speed for less capital outlay than for B.  The advantage of B is that is allows the turbomolecular pump to remain on when the chamber is vented.  An example of a Nordiko pumping system is shown in figure 2.

 

The high vacuum pump compresses gas from  1 x 10-6 Pa to about 1 Pa and the roughing pump from 1 Pa to 1 x105 Pa.  In combination they achieve a differential pressure of eleven orders of magnitude, 1011.

The second approach still requires a mechanical pump, but the high vacuum pumping element is provided by a capture pump.  Unlike the case where a compression pump is used and the gas is progressively swept out of the system, the capture pump retains the gas through sorption phenomena.  The most common capture pumps used are cryogenic pumps.  These are typically two stage pumps, cooled using a Helium compressor.  The first stage is refrigerated to about 80 K and the second stage to about 15 K.  The first stage is used to condense non-permanent gases (principally water vapour) and the second pumps the permanent gases.  While Nitrogen and Oxygen are frozen, Hydrogen and Helium are trapped within a cold molecular matrix (normally a carbon absorber).

In order to cool a cryogenic pump, it first needs to be evacuated.  The thermal conductively of the air within the pump causes too much heat transfer from the outer case.  As with a Dewar a vacuum provides very good thermal insulation.  In figure 3 a schematic diagram of cryogenically pumped chamber is shown.  The bypass arrangement is used, otherwise the cryogenic pump would need to be turned off whenever the chamber is vented.  The cycle time from cold to warm (room temperature) and then warm to cold (80 and 15 K) is about 150 minutes.

 

An advantage of adopting cryogenic pumping is that it provides very clean vacuum with a high water vapour pumping speed.  There are other factors that need to be considered.  The pump needs to be periodically regenerated.  This can render the system unavailable while the regeneration is performed (about 2½ hours).  They are also unsuitable for use in many reactive environments.   In figure 4 a vacuum cassette load lock is illustrated fitted with a two stage cryogenic pump.

 

There is also the opportunity to combine these pumping techniques into a hybrid arrangement.  For many circumstances, including reactive deposition systems, this can be very attractive.  When using RF discharge excitation in the presence of Oxygen, Ozone (O3) will be formed.  Using a two stage cryogenic pump this gas will be condensed.  Upon regeneration of the pump the liquid ozone produced can be detonated.  Other than in extreme cases this is unlikely to be a direct risk to operators of the equipment.  The noise can startle people and the percussive event damages the absorber array within the pump.

To avoid these risks and to attain the benefit of a high water vapour pumping speed a single stage cryogenic pump, with a controlled temperature baffle, can be used in concert with a turbomolecular pump.

In the section on sputtering, we noted that in order to perform thin film deposition processes under vacuum, we need to reduce the concentration of the gas molecules (number density) to reduce the mean free path.  This is the mean (average) distance a molecule travels between two successive collisions with other molecules.

It is expressed as:     

Where: l = mean free path

k = Boltzmann’s constant (= R/NA, the gas constant divided by Avogadro’s number)

T = temperature

p = pressure

d = molecular diameter

As a chamber is evacuated from atmosphere and the pressure drops, several different types of flow through the connecting pipework are encountered.  The differing flow types are determined by a combination of pressure, mean free path and component dimensions.

Between 105 and 102 Pa gas collision with the walls of the vessel, or tube, are less frequent than between the molecules.  This is because the mean free path is significantly shorter than the enclosure dimensions.  Within this range both viscous and turbulent flow can result.  The turbulent flow region is significant since eliminating it can reduce the movement of particles around the system.  The transition is determined by what is known as the Reynolds number(Re).

 

Where: ρ = gas density (Kg.m-3)

η = viscosity (Pa.s)

v = flow velocity (m.s-1)

d = tube diameter (m)

When Re is greater than 4000 the flow will be turbulent.  Between this and about 2500 the flow is in transition and below 2300 it becomes purely laminar.  Systems are designed so that the turbulent flow region is only encountered briefly at relatively high pressure.  It can be further reduced, or eliminated, by reducing the flow rate until a lower Re value is be achieved, then allowing the flow rate to rise.  This can be achieved by reducing the flow path with a narrow aperture valve for a period while the pressure drops, then opening the constriction to allow faster pumping.  Positive displacement pumps, of the type commonly used for roughing, have an almost constant volume flow rate over a fairly wide inlet pressure range.  Multiplying the volume flow rate by the inlet pressure gives the throughput of the pump.

As the pressure falls further into the range from 102 to 10-1 Pa a regime is encountered where the the mean free path becomes comparable with the dimensions of the enclosure.  In this region the collisions between the molecules and the walls of the enclosure both influence the movement of the gas.  This region is often called the Knudsen flow region.  It is also the region where many vacuum processes are carried out.

Finally we reach the molecular flow region, where the gas collisions between molecules become rare and the flow is dominated by the frequent interaction with the walls of the enclosure.  This generally occurs from 10-1 Pa and below.

Contamination, Condensation and Outgassing

To achieve the best base pressures and a fast evacuation time, vacuum chambers should be clean and dry.  Typical contaminants from the manufacturing and assembly processes might include: oil and grease, condensation of vapours (water, cleaning fluids).  All parts used on a vacuum system should be clean and dry, and clean gloves should be worn during assembly.

Permanent gases pump out of the system more readily than non-permanent gases.  Generally the key to achieving a good base pressure in the high vacuum region is the removal of water vapour.  Water is a polar molecule and adsorbs readily on metal surfaces.  Many working environments target a comfortable relative humidity level of 40 to 50%.  At these levels and at 20℃ the air can hold a partial pressure of about 1000 Pa of water vapour.  In the molecular flow region any water vapour molecule that leaves the chamber surface is much more likely to hit the wall again and be re-adsorbed than reach the pump port.  This is further complicated by the presence of deposition shields protecting the walls of the chamber.

All materials degas in vacuum.   A large part of this is the slow release of water vapour from the surface.  Once water is fully removed other factors start to become noticeable.  At the vacuum levels encountered down to  1 x 10-6 Pa these tend not to be significant.  For UHV and EHV the evolution of hydrogen from the metal surface and ultimately it’s diffusion though the chamber walls becomes significant.

For high vacuum and very high vacuum systems we can benefit from the simplicity of using elastomer demountable seals.  For the ultra-high vacuum region it is necessary to adopt metal seals.  The elastomer demountable seals take the form of o-rings.  These also outgas and have finite gas permeation rates.  They are generally synthetic rubbers and have a Poisson’s ratio approaching 0.5.

Measurement of Vacuum

In the field of vacuum processing for deposition and surface modification of thin films the measurement of pressure can be divided into;

1      The measurement of total pressure.

2      The measurement of partial pressure.

In the measurement of total pressure for practical vacuum tools it is common for the precision of the measurement to be more important than the accuracy.  In fact when taking into account the placement of a measuring gauge within the system (which is seldom consistent between equipment vendors), for the high vacuum region the reported figure is really an indication rather than an accurate measurement of the pressure.

Measurement of Total Pressure

Of the wide range of techniques that can be used to measure the pressure in a rarefied gas and within vacuum only a few are in common use industrially.  There are techniques that can be used to directly measure the pressure.  These are termed force effect gauges.  Of these only two are popular; the capacitance manometer  and the piezoelectric gauge.  In the true high vacuum range indirect measurement gauges are used.  Again a wide variety of techniques can be used to achieve this and only a few have become widely adopted.

Piezoelectric:  This gauge use a piezoresistive device to detect the deflection of a diaphragm. It is used over the range from atmosphere (105 Pa) down to about 100 Pa.  They are very good for reproducibly measuring the achievement of atmospheric pressure when a system is vented.

Capacitance manometer:  these gauges capacitively measure the deflection of a diaphragm and can be made to cover a range of about four orders of magnitude.  To measure into the 10-3 Pa range these become expensive.  They can be made using materials that are very corrosion resistant and are used widely in RIE and PECVD machines.

Thermal conductivity/convection:  the most common is the Pirani gauge which is used to measure over the range  from 102 to 10-2 Pa.  The range can be extended towards atmospheric pressure using convection enhanced designs.  This type of gauge is popular to monitor the transition region from vacuum roughing to high vacuum pumping.

Hot cathode ionisation:  these use a heated filament and measure electric current as a function of molecular gas density, which is proportional to pressure.  Their measurement is very dependent on the nature of the gas/es present.  Their measurement range extends from 1 Pa through to 10-7 Pa.

Cold cathode ionisation:  these gauges use magnetically enhanced DC excited gas discharges and have a useful range from  1 to 10-6 Pa.  They are very robust, but are not very accurate.

Spinning rotor:  these gauges exploit the fictional forces between gas molecules and a spinning ball-rotor (≈ 25,000 rpm).  They are used as transfer standards over the range from 100 to 5 x 10-5 Pa.

In recent years combination gauge packages have become available that provide three, sometime four gauges, into a compact space with onboard electronics.  These gauges are provided with power and can communicate digitally with the machine supervisory system.

Measurement of Partial Pressure

In a clean system at high vacuum there are several gas species.  These include water vapour.  The presence and quantitive measurement of water vapour can be of key interest to the thin film technologist.  The presence of trace hydrocarbon impurities can also be of significance.  An instrument that can measure the various contributions to the total pressure of the various constituents present can be valuable.  As with the measurement of total pressure there are a variety of different instruments that can be used for this role.  The interested reader is referred to Bello for further information.  One type is far more important than the others and is widely used.  This is a type of resonance high-frequency spectrometer, known as a quadrupole mass spectrometer.  Such instruments are available with mass ranges from 1 to 100 amu through  to 1 to 500 amu.  Generally these are low resolution instruments with a resolving capability of about 1 amu.  They can measure pressure over the range from 10-3 to 5 x 10-9 Pa with a Faraday detector.  Equipped with an electron multiplier or a channeltron the detection limit can be extended as low as 10-12 Pa.

A quadrupole mass spectrometer can be used to fingerprint the residual vacuum within a system.  This is why often they are marketed as residual gas analysers.  With fast electronics they can also be used to achieve closed-loop partial pressure control within systems used for reactive processing.  Depending upon the process pressure it may be necessary to differentially pump the spectrometer.  There are also instruments that are specially designed for operation at higher pressure, but the upper limit for this type of unit remains in the region of 1 Pa.