Sputtering Paper

Plasma Sources Science and Technology
TOPICAL REVIEW • OPEN ACCESS You may also like

- Cu/Cu Barrier Interconnect with Low
Foundations of physical vapor deposition with Resistivity for the Application to the Next-
Generation and High-Resolution Display
plasma assistance Fabricated Using Microwave-Assisted
Sputter
Wooseok Jeong, Jae-chul Do, Jeong Rak
Lee et al.
To cite this article: Jon Tomas Gudmundsson et al 2022 Plasma Sources Sci. Technol. 31 083001
- Study of surface evolution via Modified Ion
Beam Sputtering Semi-Empirical Model
(MIBSEM)
Oluwole E. Oyewande, Samuel A.
Awelewa and T.V. Omotoso
View the article online for updates and enhancements.
- Effect of Sputtering Deposition Process on
Magnetic Properties of Magnetic
Multilayers
Yukiharu Maeda, Yusuke Suzuki, Yohei
Sakashita et al.
This content was downloaded from IP address 14.139.158.18 on 18/10/2023 at 16:58

Plasma Sources Science and Technology
Plasma Sources Sci. Technol. 31 (2022) 083001 (33pp) https://doi.org/10.1088/1361-6595/ac7f53
Topical Review
Foundations of physical vapor deposition

with plasma assistance
Jon Tomas Gudmundsson1,2,∗ , André Anders3,4
and Achim von Keudell5
1
Space and Plasma Physics, School of Electrical Engineering and Computer Science,
KTH Royal Institute of Technology, SE-100 44 Stockholm, Sweden
2
Science Institute, University of Iceland, Dunhaga 3, IS-107 Reykjavik, Iceland
3
Leibniz Institute of Surface Engineering (IOM), Permoserstraße 15, 04318 Leipzig, Germany
4
Felix Bloch Institute of Solid State Physics, Leipzig University, Linnéstraße 5, 04103 Leipzig, Germany
5
Institute for Experimental Physics II, Ruhr-University Bochum, Germany
E-mail: [email protected]
Received 29 May 2021, revised 16 June 2022

Accepted for publication 7 July 2022
Published 19 September 2022
Abstract
Physical vapor deposition (PVD) refers to the removal of atoms from a solid or a liquid by
physical means, followed by deposition of those atoms on a nearby surface to form a thin film
or coating. Various approaches and techniques are applied to release the atoms including
thermal evaporation, electron beam evaporation, ion-driven sputtering, laser ablation, and
cathodic arc-based emission. Some of the approaches are based on a plasma discharge, while
in other cases the atoms composing the vapor are ionized either due to the release of the
film-forming species or they are ionized intentionally afterward. Here, a brief overview of the
various PVD techniques is given, while the emphasis is on sputtering, which is dominated by
magnetron sputtering, the most widely used technique for deposition of both metallic and
compound thin films. The advantages and drawbacks of the various techniques are discussed
and compared.
Keywords: physical vapor deposition, magnetron sputtering, cathodic arc deposition, ion
beam deposition, sputtering, pulsed laser deposition
(Some figures may appear in colour only in the online journal)
1. Introduction plasma phase. PVD is a broad field and various processes are
applied to create film-forming material and to achieve thin film
The formation of a condensible vapor by physical mechanisms deposition. Physical vapor can be created by a wide range of
and subsequent deposition of this material onto a substrate as a techniques, which have in common that the atoms are removed
thin film or coating is referred to as physical vapor deposition from a solid or liquid source by physical means, momentum
(PVD) (Mahan 2000, Rossnagel 2003, Thornton 1988). The exchange via thermal evaporation, sublimation, ion sputter-
formation of a vapor refers to a phase transition of the film- ing, electron beam and laser ablation, and/or arc-based emis-
forming material from a solid or liquid phase into a gaseous or sion. Historically, vapor was descriptive for the film-forming
∗
material in evaporation processes, as atoms in a vapor can
Author to whom any correspondence should be addressed.
be characterized by the equilibrium parameter temperature.
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further However, when describing most modern PVD techniques, the
distribution of this work must maintain attribution to the author(s) and the title term ‘vapor’ (the gas phase of a substance at a temperature
of the work, journal citation and DOI. lower than its critical temperature) is somewhat of a misnomer
0963-0252/22/083001+33$33.00 Printed in the UK 1 © 2022 The Author(s). Published by IOP Publishing Ltd
Plasma Sources Sci. Technol. 31 (2022) 083001 Topical Review
since we do not simply deal with the gas phase but with a evaporation provides film-forming species at typical thermal
non-isotropic, directional flux of particles. All forms of sput- energies, significantly below 1 eV, sputter techniques provide
tering, pulsed laser deposition (PLD), and arc processes deliver energetic neutrals of a few eV, as these originate from a colli-
fluxes characterized by a preferred direction and the presence sion cascade in the solid when an ion impinges onto a target
of energetic particles. Often the film-forming material is com- material to be transferred to the substrate. PLD and arc deposi-
posed of atoms or small clusters of atoms of elements that tion create species with even higher energies, which can in both
cannot easily be prepared as a gas phase plasma precursor. The cases exceed 100 eV. Thereby, the various PVD techniques
chemical reactions in the PVD process occur almost entirely provide film-forming species that can span a wide range in
on a surface, preferently the substrate surface. energy. Figure 1(a) shows the typical range of particle energy
PVD processes are often categorized into equilibrium and in the substrate vicinity for a few PVD techniques. The parti-
non-equilibrium processes depending on the ways in which cle energy varies over a wide range both within each technique
the source material is vaporized. The equilibrium processes and between the different techniques. Figure 1(b) shows the
include thermal evaporation, by conventional resistive heating, growth rate range for the different PVD techniques. The result-
as well as electron-beam evaporation. Deposition methods that ing film growth rates also vary over a wide range between
are based on thermal evaporation include molecular beam epi- the different PVD processes. Thermal evaporation deposition
taxy (MBE) and ion plating. Evaporation is easiest to achieve has the highest growth rate but exhibits the lowest energy of
by simply heating the source material, that is placed in an evap- the film-forming species, while the growth rate of ion beam
oration crucible, with a hot filament or conventional resistive and PLD is low, but the energy of the film-forming species is
heating. These methods are often combined with a plasma dis- high. Also, the ionization fraction of the film-forming species
charge, that either provides radicals to the process or ionizes varies between the different methods. In thermal evaporation
the evaporated atoms. In the latter case, a discharge is cre- the film-forming species are mostly neutral, while arc deposi-
ated, located between the evaporant and the substrate, and the tion and PLD produces highly ionized flux of the film-forming
evaporated atoms are ionized as they travel through the dis- species. These process parameters, deposition rate, energy of
charge. In other cases, the evaporation deposition is accompa- the film-forming species and their ionization fraction dictate
nied by simultaneous ion bombardment of the growing film by the microstructure of the deposited film, and therefore the
ions from an ion source. In non-equilibrium evaporation, the properties of the resulting films or coatings. The influence of
film-forming material is fed from an open source, often liquid the ionization fraction in the deposition flux on the film surface
material that evaporates off into a large, low pressure volume. morphology is demonstrated in figure 2, which was obtained
Non-equilibrium processes include sputtering, by ions from a by molecular dynamics simulations of copper film growth,
plasma discharge or by ion beams accelerated from a dedicated comparing fully neutral deposition flux (thermal evaporation),
ion source. 50% ionization and 100% ionization (PLD, cathodic arc) of
Plasma-based PVD processes often use a plasma discharge the deposition flux (Kateb et al 2019). Thermal evaporation
as the source of ions for a sputter process, or the sputter pro- deposition gives films that exhibit very rough surfaces as seen
cess is driven by ions from an ion source. In both cases the in figure 2(a), while with increased ionization of the deposi-
ions are accelerated to a solid target, which is the source of tion flux the films exhibit increasingly smoother surface as
the material to be deposited, the film-forming material. The seen in figure 2(b) for 50% ionization, and for fully ionized
positive ions are preferentially ions of a heavy inert gas such deposition flux in figure 2(c). It is clear from this comparison
as argon. In the former case, sputter deposition is driven by that the ionization fraction has a significant influence on the
positive ions from a plasma discharge that is created and main- microstructure and surface morphology of the resulting film.
tained through electron impact ionization of a low pressure Apart from the growth rate, the energy of the film-forming
inert working gas. As the ions bombard a negatively biased tar- species and the ionization fraction, there are a number of other
get (the cathode) composing the source material, atoms of the issues that have to be considered when choosing a deposi-
film-forming material are ejected and fall onto and coat a sub- tion method for a given application. By application of mag-
strate. The film-forming species released from a solid or liquid netron sputtering, the film properties such as the crystalline
by evaporation or sputtering are generally neutral atoms. Often phase, microstructure, stress, morphology, mechanical prop-
the sputtered species are made to travel through a discharge erties, optical properties, and electrical resistivity can be tuned
where they are ionized, either intentionally or unintentionally. by adjusting the applied power and working gas pressure,
Furthermore, the ion bombardment, in the case of sputtering, while in thermal evaporation deposition there is no control
also induces emission of secondary electrons. In fact the dis- over the energetics of the evaporated species and the resulting
charges are maintained by secondary electron emission from film properties. In sputter deposition, the composition of alloy
the cathode (diode sputter source) or by Ohmic heating, or targets is generally reproduced in the deposited film, while
the combination of the two (magnetron sputtering discharge). in thermal evaporation the composition of the resulting film
Another plasma-based PVD process is cathodic arc (including depends on the relative vapor pressure of the constituent ele-
vacuum arc) deposition, characterized by a high flux of ionized ments, which makes it a challenge to deposit alloys. In recent
cathode material. decades there has been a trend of combining the different depo-
The energy, or more precisely the energy distribution, sition techniques for thin film deposition, surface modification,
of the film-forming species can have a significant influence or surface treatment, in order to optimize throughput, and to
on the properties of the deposited film. Whereas thermal achieve the desired thin film properties.
2
where Th = T/Tm is the film temperature normalized by the

melting temperature Tm of the deposited film, Jα is the ion
current density onto the substrate, Nmoved,α is the number of
atoms that are rearranged on the surface, and the potential
energy is
Epot = Ece + (Eiz − φ), (2)
and Ece is the heat of sublimation, or the cohesive energy, Eiz
the ionization energy, and φ the work function of the electron
for neutralization. The generalized kinetic energy is given as

((Ekin,α Mα )/(Ece Ms ))Jα
E∗ = α (3)
α Jα
where Mα is the mass of the energetic particle, Ms is the mass

of an atom in the deposited film. The structure zone diagram
(figure 3) shows that the film morphology varies from tapered
crystallites separated by voids (zone 1) to an array of fibrous
grains separated by grain boundaries (zone T) to columnar
grains extending to the entire coating thickness (zone 2), and
to large grains with flat tops (zone 3), with increased film tem-
perature. Higher film temperatures lead to increased diffusivity
of the film atoms during growth and therefore increased grain
size.
The dependence of the film properties on the parameters E∗
and T ∗ , as shown in figure 3, illustrates that the temperature T ∗
can be partly replaced by the energy of the incident ions E∗ ,
because similar film structures (as characterized by the differ-
ent deposition zones in figure 3) are obtained at lower temper-
ature T ∗ but higher E∗ . Therefore, as the generalized energy
E∗ is increased the transition temperature between the neigh-
boring zones is shifted to lower values of both t∗ and T ∗ . On
the microscopic level, this is explained by the higher adatom
mobility of energetic species contributing to film growth at
Figure 1. The typical (a) range of particle energy at the substrate higher E∗ . The term adatom is short for adsorbed atom and the
and (b) growth rate for the various PVD methods: thermal diffusion of this atom along the surface is referred to as adatom
evaporation deposition (TED), magnetron sputtering deposition mobility. Adatom mobility can be influenced not only by ther-
(MSD), pulsed laser deposition (PLD), ion beam deposition (IBD) mal energy but also by energetic particle bombardment. From
and cathodic arc deposition (CAD). Based partially on Bundesmann a technological perspective, this has significant advantages. A
and Neumann (2018), while for the cathodic arc the deposition rate
is based on Goldberg et al (2012) and the energy range is based on typical application scenario of PVD process is the deposition
Anders and Yushkov (2002) and Byon and Anders (2003). of hard ceramic coatings on metal work pieces. In a purely
thermal evaporation deposition process, such hardness could
only be reached at very high substrate temperatures, where
Both the substrate temperature (or the film temperature) the hardened metal substrate unfortunately becomes weak-
and the energy of the film-forming species influence the ened again. This is avoided in ion-driven deposition processes,
microstructure of deposited polycrystalline films. This can be where the metal substrates can remain at much lower deposi-
shown by a structure zone diagram that incorporates both the tion temperature during the application of the PVD coating,
deposition temperature and the energetic deposition, which is due to bombardment of energetic species.
typically due to a large flux of ions (Anders 2010). Such a Here, we review the various PVD techniques that have been
structure zone diagram where the variables are a generalized developed for thin film deposition, while the main emphasis is
temperature T ∗ , a normalized energy flux E∗ , and the net film on ion-driven plasma discharge-based approaches, including
thickness t∗ , is shown in figure 3. The generalized temperature diode and magnetron sputtering, as well as cathodic arc, and
T ∗ , which includes both the film temperature, and the potential PLD. In the following, we start with the discussion of simple
energy of the particles that arrive on the substrate, is defined thermal and electron beam evaporation processes in section 2,
as which are often initiated by a plasma or are being combined
with a plasma discharge. This is followed by discussion of the
1 α (Epot Jα /Nmoved,α )
T ∗ = Th + , (1) various plasma-based PVD methods in section 3, where the
k α Jα main emphasis is on the magnetron sputtering discharge and
3
Figure 2. Surface topology of copper films obtained by molecular dynamics simulations assuming (a) fully neutral deposition flux, (b) 50%
ionization of the deposition flux, and (c) full ionization of the deposition flux, while assuming the same deposition time and energy
distribution of the film-forming species. The deep blue indicates substrate surface and red denotes thickness higher than 6 nm. Reprinted
with permission from Kateb et al (2019). Copyright 2019, American Vacuum Society.
2.1. Thermal evaporation
The first commercial thin film deposition technique was ther-

mal evaporation or resistance heated evaporation, where the
evaporants are placed in a boat or a crucible. Typically, the boat
is made of a refractory metal, such as tantalum or tungsten,
which is heated by passing a large current through the band
of metal forming the boat. A crucible is typically a ceramic
cup that is wrapped with a metal wire coil which is resistively
heated by passing a current. The evaporation rate is generally
proportional to the evaporation temperature and consequently
dependent on the power density delivered to the source. The
energy distribution of the evaporant is determined by the ther-
modynamics at the evaporation temperature. For evaporation
Figure 3. A structure zone diagram based on a generalized occurring around 800–1300 K, which is common tempera-
temperature T ∗ , a normalized energy flux E∗ , and net film thickness ture range used, the kinetic energy of the evaporated species is
t∗ . Reprinted from Anders (2010), Copyright (2010), with small (see figure 1(a)) compared to the chemical bond strength
permission from Elsevier.
of a solid film and therefore, it has little effect on the deposi-
tion mechanism, the film growth, and film quality. The typical
energy range of the evaporated atoms is 0.03–0.5 eV. Thermal
its variations, and finally we discuss cathodic arc deposition in evaporation is typically performed at sufficiently low back-
section 4. ground pressure (∼10−4 Pa) so that the evaporated atoms are
transported to the substrate without collision and therefore
thermal evaporation is a ‘line-of-sight’ process. The evapo-
2. Evaporation rated material then condenses on a substrate which is kept
at a suitable temperature. The resulting film microstructure
To deposit thin films by evaporation, in a vacuum environment, is columnar, separated by voids, which is denoted as zone 1
sufficient amount of heat has to be provided to the evaporant. and maybe zone T in the structure zone diagram of figure 3.
The evaporation can be driven by thermal heating (resistance This occurs because of limited surface mobility of the evapo-
or induction heating), electron beam heating or laser ablation rated atoms on the substrate, during growth. Furthermore, the
of a source material. Note that thermal evaporation in high vac- surface morphology is rough as can be seen in figure 2(a).
uum is not a plasma supported deposition process as the film- When both solid and/or liquid states and vapor states exist
forming species are neutral atoms. Evaporation methods based simultaneously at a common temperature in an enclosed cham-
on heating where the evaporation process is nearly in steady ber the equilibrium pressure is referred to as vapor pressure or
state equilibrium with its vapor (effusion or Knudsen cell) are saturation vapor pressure. Then, equal number of atoms of the
the basis for deposition techniques such as MBE, where the solid and/or liquid enters the gaseous phase in comparison to
vapor is composed of neutral atoms. the atoms that condense, the evaporation rate is equal to the
4
unit surface area per unit time called the impingement rate Φ
(molecules cm−2 s−1 ) (Harper 1990)
αv A cos φ cos θ
Φ= √ (pv − p), (8)
πr2 2nMkB T
where φ is the angle from source normal to substrate, θ is the
angle from substrate normal to the source, and r is the source-
to-substrate distance. This leads to a deposition rate DR on the
substrate
MΦ
DR = 107 (nm s−1 ), (9)
ρN0
Figure 4. The vapor pressure versus temperature for various
elemental materials. Reprinted from Gall (2005), Copyright (2005),
where ρ (g cm−3 ) is the deposited film mass density, M
with permission from Elsevier. (g/mole) is the molecular weight of the evaporating species,
and N0 is Avogadro number. However, due to the roughly
exponential dependence of vapor pressure on temperature
condensation rate. The vapor pressure is temperature depen- (equation (6)) it is a challenge to control the flux from
dent. The vapor pressure that develops over a solid or liquid the source. Under typical vacuum deposition conditions, the
can be estimated using the Clausius–Clapeyron equation vaporizing material is not in equilibrium with its vapor, how-
ever it arrives at the substrate at the impingement rate given
dpv ΔHv by equation (8) at the given vapor pressure determined by the
= (4)
dT T(vg − vs ) source temperature. Due to this thermal evaporation deposition
systems have to be feedback controlled based on measuring the
where Hv is the heat of vaporation, v g the molar volume of
deposition rate.
the vapor, v s the molar volume of the sublimating solid and/or
Evaporation is very effective for depositing materials of
the evaporating liquid. The molar volume in the liquid or solid
high vapor pressure (which are often materials of low melt-
state is very small in comparison with that of the vapor phase
ing point). A major drawback of thermal evaporation depo-
so that v g − v s ≈ v g ≈ RT/pv , where R is the universal gas
sition is the difficulty in forming alloy films from a single
constant, and
d(ln pv ) ΔHv crucible (Rossnagel 2003). As each element has its own tem-
=− . (5) perature–vapor pressure relation the evaporation rate of two
d(1/T) R
elements in a common crucible or boat is generally very dif-
If the heat of evaporation is assumed to be a constant we find ferent. The element with higher vapor pressure will evaporate
at a higher rate, leaving behind the lower vapor pressure ele-
pv ∝ exp(−ΔHv /RT). (6) ment. The substrate will therefore receive first the higher vapor
pressure element and at a lower rate the lower vapor pressure
The exponential dependence of the vapor pressure with tem-
element. This can be solved by having a source crucible for
perature indicates that small variations in temperature can
each element and adjust the flux to the desired level. This is
lead to large changes in the vapor pressure and the conden-
referred to as co-evaporation, but it is a challenge to operate in
sation rate. The vapor pressure is shown versus temperature
controlled manner. Therefore, due to these difficulties thermal
for various elemental materials in figure 4.
evaporation is typically not used to deposit alloy films.
The main advantages of thermal evaporation deposition are
The film microstructure can be greatly improved, resulting
high-deposition rates and that the source material can be sup-
in enhanced film properties, if the depositing film is bom-
plied in a simple form. The rate of evaporation is given by the
barded with energetic particles (ions and/or neutrals). One
Hertz–Knudsen equation
approach to improve the properties of evaporated thin films
1 dN αv is ion plating, which is discussed in section 2.3. Another
= √ (pv − p), (7) approach is called metal-vapor plasma deposition. Then, the
A dt 2nMkB T
evaporated flux is directed through an electrodeless discharge
where A is the emitting surface area, N is the number of gas as it travels toward the substrate. This has been demonstrated
atoms or molecules, αv the evaporation coefficient, pv is the using electron cyclotron resonance (ECR) discharge to ion-
equilibrium vapor pressure at the evaporant surface, p is the ize the copper vapor that was thermally evaporated (Holber
hydrostatic pressure acting on the surface, M is the molecular 2000, Holber et al 1993). This setup is shown in figure 5. A
weight of the evaporated atoms, kB is Boltzmann’s constant microwave (2.45 GHz) is fed into a vacuum chamber through
and T is the temperature. The flux of material evaporating from a quartz window. Four electromagnets create the magnetic field
a circular area of uniform temperature exhibits a cosine angu- required for efficient microwave absorption, creation of a dis-
lar distribution, with its maximum normal to the emitting area. charge, and transport of the plasma to the substrate. Copper is
The arrival rate onto a substrate can be calculated assuming evaporated from a resistively heated source, which is continu-
a cosine angular distribution to account for the relative posi- ously replenished through the use of an in-vacuum wire feeder.
tions and orientations of the substrate and emitting areas. This The evaporated copper atoms enter the resonance zone of the
gives an arrival rate or the number of particles incident upon a discharge, the plasma is created from the evaporated metal, and
5
Figure 5. ECR discharge based ionization of evaporated copper

apparatus. The copper atoms are introduced into the vacuum
chamber from a thermal evaporator located underneath the chamber.
Reprinted with permission from Holber et al (1993). Copyright
1993, American Vacuum Society.
Figure 6. A schematic of an electron beam heated evaporation
source. The electron beam originates from a filament located from
no working gas is needed during the deposition. This approach underneath the hearth, passes through the magnetic lens and is
has been demonstrated to give copper flux at the substrate focused upon the evaporant. Graper (2017), reprinted by permission
of the publisher (Taylor & Francis Ltd, http://tandfonline.com.)
that is nearly 100% ionized (Holber et al 1993). The substrate
itself is not placed line-of-sight of the thermally evaporated
copper.
2.3. Ion plating
2.2. Electron beam evaporation
Ion plating refers to an evaporation deposition process in
The electron beam evaporation sources differ from thermal which the substrate or the growing film experiences in addi-
evaporation sources in two fundamental ways: the heating tion to the flux of evaporated species a continuous or periodic
energy is supplied onto the top of the evaporant as the kinetic flux of energetic species that influences the deposition pro-
energy of a high current electron beam. The incident e-beam cess and consequently the resulting film properties (Mattox
only melts a relatively small area on top of the source mate- 1998, chapter 8). When positively charged ions are accelerated
rial. This small area is heated well above the melting point, toward a negatively biased substrate, in an otherwise thermal
while the temperature of most of the source material is much or electron beam evaporation process, it is coined ion plating
lower, as the hearth is typically water-cooled. Therefore, the (Martin 1990). These can be either ions of the evaporated film-
molten source material is only in contact with a de-facto cru- forming species or more often the ions of an inert working
cible made of the same material, and any chemical reactions gas. Ion plating was initially applied to improve film adhe-
with contaminants, such as the crucible, are eliminated. The sion and the surface coverage of the deposited films, but soon
use of an electron beam therefore eliminates the limitations came the understanding that bombardment by energetic parti-
caused by the need to reach the melting point of materials that cles could enhance film properties, such as film mass density
can be evaporated. and residual film stress. Ion plating also provided the possibil-
The electron beam evaporation source is composed of three ity to sputter clean the substrate with argon ions before deposit-
basic components: the electron gun, the beam deflection mag- ing a film. For a more thorough discussion on ion plating the
netic lens, and the evaporant containing hearth. The electron reader is referred to the reviews by Mattox (1998, chapter 8)
beam typically originates from a filament located from under- and Stelmack et al (1989).
neath the hearth, passes through the magnetic lens and is The ion plating system can be either ‘plasma-based’ or
focused upon the evaporant, so the filament is not in line-of- ‘vacuum-based’. Evaporation can be driven by either resis-
sight of the point of evaporation. This approach has the benefit tive or electron beam heating of the evaporant. Plasma-based
that a very high power density can be provided, and hence a ion plating can include evaporation that is carried out in a
wide range of control over evaporation rates from very low to plasma environment where a fraction of the evaporated atoms
very high, can be achieved. These components and the crucible are ionized in a plasma discharge. Ionization can also be
are typically constructed as an integrated assembly as shown achieved by having the electrons or an electron beam that
in figure 6. Typically the electron beam is accelerated to about drives the evaporation also pass through the vapor of the
10 keV and the electron current can be as high as 1.5 A. This evaporant. These ions can then be accelerated toward the
beam impacts an area of up to 1 cm2 with a power density of substrate by an electric field applied using an external voltage
up to 60 kW cm−2 (Graper 2017). supply to create a potential between the evaporation source
6
Figure 8. A schematic of IBAD system. It consists of evaporation

source for the film-forming material and an ion source that directs
ions onto the substrate. Based on Aisenberg and Chabot (1973).
Figure 7. A schematic of plasma-based ion plating or a

thermionically assisted triode coating system. The source of the arate from the source of energetic ions, while the evaporation
film-forming material is electron beam evaporation. The substrate flux and ion bombardment are simultaneous. This is referred
acts as a cathode with bias voltage of up to 5 kV. Reprinted from to as ion beam assisted deposition (IBAD) and was originally
Matthews and Teer (1980), Copyright (1980), with permission from
Elsevier. introduced by Aisenberg and Chabot (1973). A schematic
of an IBAD system is shown in figure 8. The ion sources
used for IBAD are typically of a broad-beam design (Kauf-
and the substrate. One approach is to generate a dc glow man source) (Kaufman and Robinson 1989), while other ion
discharge by applying a highly negative voltage to the sub- sources are also used. The IBAD technique has been reviewed
strate. The working gas is often argon that is maintained by Rauschenbach (2002).
at pressure of 1–10 Pa and negative voltage of 3–5 kV is
applied to the substrate. The substrate is therefore submersed 2.4. Pulsed laser deposition
in a plasma and is negatively biased, and therefore can act
as a cathode. The sheath that forms is collisional and the PLD or laser ablation utilizes a high-energy pulsed laser beam
positive ions and neutrals that bombard the substrate exhibit as an excitation source and the photon energy is coupled to
a broad energy spectrum. The energy distributions of both a solid bulk target material via electronic processes. As the
ions and neutrals range from zero to the energy correspond- energetic photon beam is directed onto a solid target it can
ing to the cathode potential. Teer (1976) estimates that for induce laser ablation and evaporation of the target material.
every energetic ion arriving at the cathode at least 14 ener- The nature of the interaction of a photon beam with a solid tar-
getic neutrals arrive and that the average energy of an arriving get depends on the power density, pulse duration, wavelength,
ion is about 1/10 of the cathode potential while the average as well as the optical and thermodynamical properties of the
energy of the neutrals is 1/22 of the cathode potential. The solid material.
plasma can also be generated by electron beam (thermionic The setup of a PLD system is shown schematically in
triode), by a hollow cathode source (Morley and Smith figure 9. The laser beam is introduced into a vacuum chamber,
1972) or an inductively coupled discharge (Murayama 1974). through an optical window, where it is focused onto the target
The substrate can be positioned either in the active plasma surface. Upon laser ablation, the target material vaporizes and
region or at a remote or downstream position. This approach is the constituents are transported onto a substrate to be deposited
sometimes referred to as ion assisted deposition or ion vapor as a thin film. A typical target-to-substrate distance is 2–10 cm.
deposition. The chamber background pressure is maintained well below
A schematic of a plasma-based ion plating system with an 10−6 Pa using a vacuum pumping system. Often the chambers
electron beam evaporation source delivering the film-forming hold target manipulators for multiple targets, which makes it
material is shown in figure 7. The system employs a posi- possible to fabricate multilayered structures using different tar-
tive electrode which has the purpose of extracting electrons get materials. An example of the construction of a PLD system
generated in the vapor source region to increase ionization. is given by Ye et al (2021). Various gases such as oxygen, nitro-
A hot filament electron source is incorporated and serves the gen, argon, and hydrogen can be introduced through the gas
role of improving controllability, to increase ionization, and inlet during thin film deposition to promote gas-phase reac-
to permit a smooth transition from the etch to the deposition tions of the ablated species. Reactive background gas is then
mode. introduced into the process chamber to incorporate e.g. oxygen
In ‘vacuum-based’ ion plating, the substrate is deposited in or nitrogen into the growing films. Furthermore, a background
vacuum and the ion bombardment is achieved by a separate gas is also sometimes used to act as moderator of the kinetic
ion source. In that case the source of vaporization can be sep- energy of the arriving species.
7
Figure 9. A schematic of a typical PLD setup. A laser beam is

introduced through an optical window and focused onto a target
surface. Upon laser ablation, the target material vaporizes and the
constituents are transported onto a substrate where they form a thin
film. The inset picture shows an actual photograph of the plume.
Reprinted from Martin et al (2010), Copyright (2010), with
permission from Elsevier.
When the laser power density reaches a certain value, mate-

rial is removed from the solid target in the form of a luminous
plume. Then the absorption of the incident laser beam energy
by the target surface leads to ablation. The threshold power
density needed for ejecting a plume depends on the target
material, its morphology, and the laser pulse wavelength and
duration. The electric field amplitude of an electromagnetic
wave is given by (Willmott and Huber 2000)
1/2 Figure 10. Approximate time scales of laser energy absorption and
2pp ablation process using nanosecond laser pulses, along with the
E= (10)
cn0 various processes that accompany it. Reprinted from Hussein et al
(2013), with the permission of AIP Publishing.
where pp is the power density, c is the velocity of light, 0 is the
permittivity of free space, and n is the refractive index of the
solid target. For a radiation with power density 500 MW cm−2 of the pulse energy reheats the plasma. As the pulse width is
the electric field strength is of the order of ∼107 V m−1 , which reduced to sub-picosecond, i.e. shorter than the time scale of
is sufficient to cause dielectric breakdown in many materi- the electron–phonon coupling (a few picoseconds), the abla-
als. Note that the power density is proportional to the laser tion process becomes a direct solid–vapor (or solid–plasma)
fluence and inversely proportional to the laser pulse duration transition.
τ p . Typical values for the applied laser power densities are The first stage of the laser ablation process (i) is the pho-
50–500 MW cm−2 for ablation using ultraviolet (UV) excimer ton absorption by the target surface, as the photon energy of
laser pulses of 10 nanosecond duration. In principle, any pho- the incident laser beam is converted into electronic excitations
ton beam or laser with sufficient output power can be applied which is subsequently transformed into thermal, chemical, and
to induce evaporation. mechanical energy. The photon beam irradiation is absorbed
The temporal evolution of the photon beam ablation pro- within the surface region of the solid target and the electromag-
cesses is summarized in figure 10 for a nanosecond long pulse. netic energy is immediately converted into electronic excita-
The processes include laser photon energy absorption at the tion as plasmons, unbound electrons and, in the case of insula-
surface and material excitation, temperature rise followed by tors, excitons. The penetration depth depends on the absorption
surface melting, ablation and plasma formation, laser–plasma coefficient for the solid target αs at the given laser wavelength
interaction, shock wave formation, and plume collapse. The as dictated by the Beer–Lambert law, so that small absorption
ablation processes can be broadly classified into three regimes coefficients result in larger penetration depths. Consequently,
separated by different time zones (shown in dotted lines in the temperature in the vicinity of the irradiated area rises up to
figure 10). There are differences between nanosecond and fem- several thousands Kelvin. Often the material removal depends
tosecond laser ablation processes, as in the femtosecond case, on the rate of thermal conduction throughthe lattice. If the
the plasma is formed after the pulse end, while for the nanosec- thermal diffusion length, given by T = 2 Dτp , where D is
ond pulse the plasma is formed during the pulse and portion the thermal diffusion coefficient, is smaller than 1/αs , the bulk
8
will be heated down to 1/αs , independent of pulse duration. radiation is stronger attenuated than UV radiation). Note in
Furthermore, shorter wavelengths also lead to shallower light particular that the absorption coefficient scales as ∝ ne ni and
penetration depths and consequently lower threshold fluences therefore the plasma plume absorbs the incident laser radiation
and ablation rates. Transfer of energy from the electrons to mainly at distances very close to the target surface where the
the lattice occurs within a few picoseconds, and heating of the densities of the charged particles are very high. The ablation
absorption layer will begin. When the laser pulse duration is rate exhibits a strongly decreasing process efficiency as the flu-
shorter than the electron–phonon coupling process, conven- ence increases and this has been related to light attenuation and
tional thermal absorption is significantly limited, and other extinction in the laser-induced plasma/vapor plume. For lasers
thermophysical effects play more important roles. For pulse in the infrared, inverse bremsstrahlung is known to be the
lengths τ p that exceed the electron–phonon coupling time, the dominating absorption mechanism, while the effect of inverse
electron–phonon coupling leads to material ablation that pro- bremsstrahlung is expected to play a minor role for excimer
ceeds via conventional heat absorption. This process is limited laser ablation. Also the wavelength does not only affect the
within a depth of tens to hundreds of nanometers. However, absorption by the target material but also the absorption by the
the thermal diffusion lengths are only of this order when using resulting plasma plume, which has higher absorption at longer
femtosecond pulses. Therefore, only when 1/αs T the pho- wavelengths. The plume ionization toward the end of the laser
ton energy is efficiently thermally transported and deposited pulse is in the range 10%–100%, but then decreases again
into the absorption layer. For metals, the optical absorption due to electron recombination as the plume travels toward the
depth is smaller than the thermal diffusion length. Therefore, substrate.
the energy from the laser pulse is first transferred into the The third stage (iii) is the expansion of the plume as the
absorption layer, followed by a thermal transport of the order ejected species (which may include atoms, molecules, elec-
√
of the thermal diffusion length, which is proportional to τp . trons, ions, clusters, and micron-sized particulates) travel away
For ceramics, the thermal diffusion length is shorter than the from the target surface at high mass transport velocity as the
optical absorption depth and the target will be heated over a ablated species travel toward the substrate. In most cases, PLD
distance similar to the optical absorption depth, regardless of is a strongly forward oriented deposition technique and most
pulse duration. This makes it possible to achieve significant of the deposited material is contained within an angular range
material removal even at low laser fluence and with minimum of ±30◦ (Ojeda-G-P et al 2018). Finally, the material that
ionization. composes the plume condenses on a substrate as a thin film.
During the second stage (ii), the photon absorbtion is fol- Typically, PLD uses nanosecond pulses, since for longer
lowed by target material vaporization and formation of plasma. pulses thermal effects with droplet formation can dominate.
Species in the heated area are ejected from the target creat- For ultrashort (femtosecond) pulse nanoparticles are often
ing a plume while they continue to absorb energy from the produced, which may be desired for some specific applica-
photon beam. This is particularly the case for excimer lasers tions. The common pulsed laser sources include a CO2 laser
that exhibit pulses lasting several tens of nanoseconds. For (λ = 10.6 μm), Nd–YAG laser (1064 nm) and up to the fifth
pulses shorter than 1 picosecond, the plasma forms at the end harmonic outputs (213 nm), femtosecond lasers using hybrid
of a laser pulse. The laser ablation creates a plume that is dye/excimer and Ti:sapphire as well as XeCl (308 nm), KrF
composed of a mix of various species, including electrons, (248 nm), ArF (193 nm), and F2 (157 nm) excimer lasers.
positive and negative ions as well as neutrals, all of which The advantages of photon beam evaporation include (i) the
can be in either excited or ground states. Furthermore, there production of ionized and excited species with high kinetic
can be diatomics and clusters in the plume. The main absorp- energies, (ii) numerous compounds can be evaporated in a
tion process in the plume is inverse bremsstrahlung, which similar way with negligible heating of the target, (iii) instanta-
involves the attenuation of the photon flux by inelastic scat- neous control of the evaporation process, and (iv) high vacuum
tering with free electrons. The heating of the plasma plume compatibility. The PLD technique is limited by the fact that
is determined by the plasma absorption coefficient αp , which the area of the photon beam is fairly small, and consequently
depends on the plasma temperature, wavelength, pulse dura- depositing on large area substrates is challenging. Therefore,
tion and the species density in the plasma plume, which in PLD is almost entirely used for research purposes, that is to
turn depends on the degree of ionization, evaporation rate, develop and study new materials on a small scale. An example
and the plasma expansion velocities. The absorption coeffi- of this is the development of the YBa2 Cu3 O7 high Tc super-
cient for the plasma plume can be expressed as (Ready 1971, conductors which were deposited using pulsed excimer laser
Spitzer 1956) evaporation (Dijkkamp et al 1987), and is currently the dom-
inant commercial application of PLD. The advantage of PLD
1/2 for this application is the possibility of a stoichiometric trans-
4 2π n e n i Z 2 e6
αp = 3/2
[1 − exp(−hν/kB T)], fer of multielement compounds from a single target to the
3 3kB T hcme ν 3 substrate that is achieved for (some) complex materials.
(11)
PLD has been discussed in a number of review articles
where Z is the average charge, ne is the electron density, ni is through the years (Cheung and Horwitz 1992, Cheung and
the ion density, T is the plasma temperature, h Planck’s con- Sankur 1988, Fujioka 2015, Shen et al 2004, Willmott and
stant, and ν is the frequency of the laser light (e.g., infrared Huber 2000) as well as in books such as Eason (2007).
9
2.5. Other evaporation-based PVD techniques
Molecular beam epitaxy (MBE) is a thin film deposition tech-

nique in which thermal beams of atoms or molecules are
directed toward a single crystalline substrate, that is main-
tained at high temperatures, in order to grow epitaxial film,
while the deposition process is performed under ultrahigh
vacuum (UHV) conditions. MBE is based on evaporation of
pure elemental species from resistively heated sources forming
localized beams of atoms or molecules which are the source
of the constituents of the growing film. The elevated substrate
temperature provides sufficient thermal energy to the arriving Figure 11. A schematic of the close-space sublimation apparatus.
atoms to migrate over the surface to lattice sites. The evapora- The source material (here CdTe) can be powdered and placed so that
tion sources are referred to as Knudsen cells. The design and it faces the substrate surface. The substrate and source are separated
by a small distance, typically 1–5 mm. The substrate and the source
operation of the MBE deposition sources are discussed by e.g.
material are typically enclosed in a fused silica tube. Reprinted
Bean (1993) and Knodle and Chow (2002). The clean envi- from Rahman et al (2019), Copyright (2019), with permission from
ronment, the slow growth rate as well as independent control Elsevier.
of the beam sources enables precise fabrication of films and
stacked structures. process parameters are the temperatures of the source and the
The MBE systems are typically ”baked out” (heated) at 150 substrate, the operating pressure in the reaction tube, and the
◦
C–250 ◦ C and therefore the deposition species must have neg- composition of the source material. Figure 12 shows the tem-
ligible vapor pressures at these temperatures, which prevents perature schedule for the source material and the substrate,
the use of unusually high vapor pressure elements. Also, ele- for the deposition of CdTe on single crystalline CdS substrate
ments with very low vapor pressure such as silicon, carbon or (Mitchell et al 1975). The substrate temperature is raised to
boron, may require source temperatures well above 1000 ◦ C. the desired deposition temperature a few min before the source
At these temperatures there is generally significant co- temperature reaches its desired value. The source is maintained
evaporation of heater or insulator elements, contaminating the at a higher temperature than the substrate as demonstrated
intended flux. The evaporation is therefore typically limited in figure 12. After the deposition, both the substrate and the
to those elements that sit within the center region in vapor source are cooled rapidly. To maintain the desired tempera-
pressures (see figure 4). The deposition of low vapor pres- tures, infrared radiation is typically used as a heat source, and
sure elements such as Si and Ge therefore calls for use of thermocouples are inserted into the sample holders to moni-
electron beam evaporation (or e-gun) sources. For the deposi- tor the temperatures. For a given temperature, the sublimation
tion of III–V semiconducting nitrides, a radical plasma source, rate increases rapidly as the operating pressure is reduced from
that provides nitrogen atoms to the process, is needed. This is atmospheric pressure. At pressures of about 100 Pa, the mean
often a dc hollow cathode discharge, an ECR discharge, or an free path of the gaseous species has increased enough, such
inductively coupled discharge. that the condensation process is no longer limited to the space
The MBE technique is, in particular, widely used for the between the substrate and the source material. The deposi-
growth of epitaxial semiconducting films. A typical growth tion rates can be high; for CdTe films it can be as high as
rate for III–V semiconductors by MBE is on the order of 150 nm s−1 (Amin and Rahman 2017). The operating pres-
1–3 ML s−1 (a few monolayers per second), which for GaAs sure is typically in the range 100–4000 Pa and the substrate
corresponds to roughly to 0.3–1 nm s−1 , and is quite low. MBE temperature is in the range 500 ◦ C–600 ◦ C, while the source
deposition is therefore almost entirely applied in research envi- temperature is in the range 700 ◦ C–900 ◦ C (Amin and Rah-
ronment, to explore novel materials and for the development man 2017). Parameters such as the source temperature and
of electronic and opto-electronic devices. More details on the temperature differential are crucial for depositing high-quality
MBE technique can be found in a number of review articles films. A fully automated single vacuum manufacturing tool
(Arthur 2002, Cho 1995, 1971, Knodle and Chow 2002, Ptak that utilizes multiple inline CSS sources with automated sub-
2015) and in books (Herman and Sitter 1996, Parker 1985). strate control is described by Swanson et al (2016). Plasma-
Close space sublimation (CSS) sources are often used to enhanced close-spacing sublimation (PECSS) technique has
deposit thin films, in particular CdTe and CdS thin films for been demonstrated to improve the efficiency of such devices
solar cells (Amin and Rahman 2017, Rahman et al 2019). (Metz 2012, Swanson et al 2013, 2011). This approach uses
Figure 11 shows a schematic view a close-space sublimation dc hollow cathode discharge to clean the substrate prior to
apparatus. The source material can be powdered and placed so deposition of the active layer.
that it faces the substrate surface. The substrate and source are
placed on appropriate holders (often graphite boat), and are 3. Sputter-based techniques
separated by a small distance (often 1–5 mm), and enclosed
in a fused silica tube with gas inlet and outlet tubes to create Sputter deposition has been applied to deposit thin films
a controlled environment. During the deposition period, this for over 140 years. For the first 50 years after its practical
is placed inside a furnace and the pressure lowered. The key implementation (Wright 1877) cathode sputtering was mainly
10
Figure 12. An example of a temperature schedule for the source and Figure 13. A schematic showing selected fundamental processes
the substrate employed in CSS deposition of CdTe on CdS single that occur due to ion bombardment of a solid target. Reprinted from
crystal substrate. Based on data from Mitchell et al (1975). Bundesmann and Neumann (2018), with the permission of AIP
Publishing.
applied for the production of reflecting surfaces on mirrors

accelerated across the cathode sheath, away from the nega-
and prisms (Fruth 1932). With improvements in vacuum tech-
tive cathode. They, therefore, gain sufficient energy to excite
nology, sputter deposition yielded to evaporation deposition
and ionize the atoms of the working gas, which is the main
as the main film deposition technique and it remained so for
mechanism that maintains the dc discharge.
a few decades. However, sputter deposition regained inter-
est in the early 1960s as a technique for the fabrication of
3.1. Ion–solid interaction–sputtering
thin films with increasing utilization in various applications.
This occurred as it was realized that a wider range of mate- When an energetic particle hits a solid target, the incoming pri-
rials, including alloys, could be deposited by sputter depo- mary particle collides with a target atom transferring energy
sition rather than by evaporation deposition. In fact, sputter and momentum. This initiates a number of processes, some
deposition is an effective approach for a controlled deposi- of which are shown schematically in figure 13. The scattered
tion of various materials, including alloys and compounds primary particle and the recoiled target atom can encounter
even with complex composition. The main difference to evap- further collisions, and collision cascades can develop. For dis-
oration is the fact that the film-forming species in a sputter charge voltages typically applied, the incident species has suf-
process are energetic with energies of a few eV. Sputtering ficient energy to break bonds and dislodge atoms within the
originates from collision cascades of ions impinging onto a target. The target species can also leave the target, in particu-
target. The sputtered species can leave the target with substan- lar those close to the surface. It is called sputtering when an
tial energy. The energy distribution of the sputtered species atom is ejected from a solid (or a liquid) due to bombardment
follows a Thompson distribution (Thompson 1968). by energetic particles, often ions. When the target species gain
In sputter deposition, the source of ions for the sputter enough energy in order to overcome the surface binding energy
process is typically a glow discharge, which is a very cost- Esb they can be sputtered, which can be due to a single collision
effective source of ions. The preferred ions for sputtering are (direct sputtering) or as a result of multiple collisions (collision
the ions of a heavy inert gas. The working gas is most often cascade). The primary species can either leave the target (scat-
argon, which is inert, and relatively inexpensive. Such a dis- tered species) or they can become implanted in the target. As
charge is typically formed between two parallel electrodes the sputter process progresses the implanted primary particles
where the cathode and the anode are separated by a distance can be sputtered or they can outgas from the target.
of a few cm. The gap between the electrodes is filled with Lets assume that the incident energetic particles are ions.
the working gas at pressure pg . A voltage is applied between Depending on the energy of the incident ions Ei and the mass
the electrodes. This can be simply a dc voltage or some peri- ratio of the projectile and the target atoms Mi /Mt , the collision
odic waveform. If the voltage VD is maintained above a cer- cascade can proceed as (i) a single-knock-on cascade, (ii) a
tain minimum, the discharge is self-sustaining. In a dc glow linear cascade, and (iii) a spike cascade (Greene and Barnett
discharge, almost the entire applied voltage drops across the 1982, Mahieu et al 2008, Sigmund 1969). In the single-knock-
cathode sheath or the cathode dark space, that develops next to on cascade, the incident species transfers energy to the target
the cathode, and the ion bombarding energy can be compara- atoms, after undergoing a few one-to-one collisions, which are
ble to the value of the applied voltage. The ions created within then ejected from the surface. In the linear and spike cascade
the discharge are accelerated across the cathode dark space, processes, a cascade of recoils is generated, causing ejection
toward the negatively biased cathode. The bombardment of the of atoms from the surface. The typical sputter processes fall
electrode by the ions of the inert working gas releases electrons into category (i), and in particular magnetron sputtering falls
from the cathode, the secondary electrons. These electrons are in the lower energy part of the linear cascade process.
11
The sputter process is described by the sputter yield Y(Ei , θi ) the reviews given by Hofer (1991) and Gnaser (2007) or the
which is defined as the mean number of atoms removed from a original work of Thompson (1968, 1981) and Sigmund (1969).
target surface for each incident ion, and depends on the energy
of the incident ion Ei , the angle of incidence θi , the ion mass
3.2. Diode sputter devices
Mi , and the surface binding energy of the target material Esb .
Various empirical formulas for the sputter yield as a function The dc diode sputter discharge is composed of a cathode target
of ion bombarding energy and data for various combinations of placed within a vacuum chamber connected to an external high
bombarding ions and target materials are given by e.g. Yama- voltage power supply. The negative applied voltage is often in
mura and Tawara (1996) and/or Eckstein (2007). The sput- the range 2000–5000 V. The negatively biased cathode is typi-
ter yield for a given impacting species on a given target, as cally electrically isolated from ground and the grounded cham-
a function of the energy of the incident particle, can also be ber often serves as the anode. The substrate, on which the film
calculated using computer codes such as TRIM (Transport of is deposited, is placed on a substrate holder or substrate table.
Ions in Matter) (Biersack and Haggmark 1980), SRIM (Stop- This discharge arrangement was used as a sputter source for
ping and Range of Ions in Matter) (Ziegler et al 2008, 2010) decades, commonly referred to as diode sputtering or cathodic
and TRIDYN (A TRIM simulation code including dynamic sputtering (Kay 1962, Vossen and Cuomo 1978, Westwood
composition changes) (Möller and Eckstein 1984, Möller et al 1976). Figure 14 shows a cross section of a dc diode sputter
1988). tool that consists of a sputter fixture (the cathode target) and a
The atoms that are ejected from the cathode target have concentric substrate table that sits inside a metal cylinder with
considerable energy, and the sputtered neutrals exhibit a broad a shutter located in between. A bias can be applied to the sub-
energy distribution which is described by the Thompson ran- strate table, but the substrate table can also be electrically float-
dom collision cascade model (Thompson 1968, 1981) and ing. The distance between the cathode and the substrate holder
often referred to as the Sigmund–Thompson distribution func- is generally short and these discharges are often configured as
tion and approximated by low aspect ratio discharges, as the inter-electrode separation
Et is small compared to the size of the cathode. Cathode diam-
f S−T ∝ , (12) eters are typically in the range 10–30 cm while the spacing
(Et + Esb )3−2m
between the cathode and substrate holder is 5 to 10 cm. Almost
where Et is the energy of the target species, Esb is the surface all the applied voltage appears across the cathode sheath (cath-
binding energy of the target material and m is the exponent ode dark space or the cathode fall). Consequently, the power
in the interaction potential applied V(r) ∝ r−m (Hofer 1991). applied to the discharge is almost completely used to acceler-
This model describes an energy distribution that peaks sharply ate the ions across the cathode fall to the cathode target. Ion
at 12 Esb , followed by a gradual decrease to higher energies bombardment of the cathode ejects atoms in addition to emis-
(∝ 1/Et2 ). The Sigmund–Thompson energy distribution func- sion of secondary electrons. The ejected atoms of the cathode
tion, given by equation (12), slightly overestimates the proba- material constitute the film-forming material. Due to the bom-
bility to sputter-eject energetic atoms. A modified distribution bardment by ions the cathode target heats up and therefore it
function was introduced by Stepanova and Dew (2004), where is desirable to cool the cathode with cooling fluid. The cath-
a cutoff energy Emax was added to better reflect experimentally ode fall is followed by the negative glow, a plasma composed
measured profiles of equal numbers of electrons and ions. The negative glow
n extends almost to the anode and the positive column is com-
Et + Esb monly absent. However, a short anode zone where the slightly
f S−D = f S−T 1 − . (13)
Emax + Et positive plasma potential returns back to zero at the anode,
is present. This configuration is referred to as an obstructed
Typical values of the constants are n = 1, m = 0.2 (Stepanova
abnormal dc glow discharge. A detailed and practical descrip-
and Dew 2004), and Emax = 20 eV (Lundin et al 2013). Since
tion of a dc diode sputter device with substrate bias capability
Esb is typically in the range 3–6 eV, most sputtered atoms are
is given by Vossen and O’Neill (1968).
emitted with energy in the range 1.5–3 eV.
The dc sputter source forms a weakly ionized discharge
The ejected atoms exhibit an angular distribution that
that is dominated by collisions between the neutral atoms.
depends on the direction of the incoming particles. Yamamura
Some of the energetic secondary electrons can pass through
et al (Yamamura 1981, Yamamura et al 1991) give equations
the discharge without colliding with an atom and are lost to the
for the angular distribution of sputtered species due to ion
substrate or the chamber walls. Therefore, they do not create
bombardment under normal incidence
new ions nor electrons that maintain the discharge but instead
Θ(Et , θt ) ∝ cos θt (1 + B cos2 θt ), (14) cause substrate heating. Consequently, the working gas pres-
sure must be high enough that the secondary electrons are not
where B is a fitting parameter that defines the shape of the lost to the grounded surfaces before performing ionization.
angular sputter distribution. A cosine distribution corresponds However, at these pressures the sputtered atoms experience
to B = 0, while B > 0 and B < 0 describe angular distribu- scattering by the working gas atoms. Hence, there is a narrow
tions of over-, under-cosine, and heart-shaped types, respec- pressure range around 2–4 Pa for dc glow discharge sputtering
tively. A more thorough discussion on the ion energy and ion to be viable. At this operating pressure, the cathode dark space
angular distribution of the sputtered species can be found in extends about 1–2 cm from the cathode, while the ion-neutral
12
which flows during most of the rf cycle. Therefore, a nearly

steady state saturated ion current flows during most of the rf
cycle except near the positive maximum, when an electron
current replenishes the negative surface charge, that was lost
during the remainder of the rf cycle. The target surface volt-
age is positive for only a very small fraction of the rf cycle
(Logan 1990) while the ion bombardment of the cathode target
is almost continuous. This is illustrated in figure 15(a) where
the target surface voltage becomes positive at t/T = 1/4 and
by figure 15(b) showing the replenishing electron pulse. Con-
sequently, the plasma potential is more positive than the poten-
tial of the electrodes. In addition, the sinusoidal voltage on
the target surface may also be superimposed by a negative dc
value with a mean value known as the dc-offset voltage. This
dc-offset value depends either on the geometric asymmetry of
the reactor, namely the area ratio between the powered elec-
trode to the grounded electrode, or in some cases to the exact
Figure 14. A cross section of a dc sputter diode discharge which
waveform of the applied rf-signal that may induce an electrical
allows for substrate bias. Reprinted with permission from Vossen asymmetry in an otherwise geometrical symmetric electrode
(1971). Copyright 1971, American Vacuum Society. assembly. This dc self-bias is zero for a simple sinusoidal rf
voltage and identical sizes of powered and grounded electrodes
or it can reach up to half of the rf amplitude for a very small
mean free path λi < 1 cm. Consequently, the cathode sheath is powered electrode and a large grounded surface area. In diode
collisional and the ions that impinge on the target surface do sputtering, the latter is typically the case. Figure 16 shows a
not have the full cathode potential and instead exhibit a broad model of a voltage driven geometrically asymmetric capaci-
energy spectrum as they bombard the target surface. In princi- tively coupled rf discharge. The dc self-bias follows from rf
ple, sputtering is a line-of-sight process. However, the atoms current conservation through the plasma that depends on the
that are ejected from the target surface, the sputtered species, charging and discharging of the capacitance of the cathode tar-
typically have energies of a few eV but, due to repeated scat- get (C in figure 16). The dc self-bias voltage builds up such
tering events, they eventually thermalize with the surround- that the voltage V(t) = Vb (t) − Va (t) across the discharge can
ing working gas. Therefore, the sputtered species do not have be written
the excess kinetic energy that is known to be beneficial to the V(t) = Vrf sin ωt − Vbias (15)
growing film.
where Vrf is the amplitude of the applied sinusoidal voltage.
The self-bias that builds up is given by (Chabert et al 2021,
3.2.1. Deposition of dielectrics. For the application of dc Lieberman and Lichtenberg 2005, section 11.5)
diode sputtering the cathode target has to be electrically con-
ductive, and therefore, it is mainly applicable for sputtering 1 M
Vbias = Te ln + ln I0 (Vrf /Te ) (16)
of metals. If the cathode is itself an insulator, or if the cath- 2 2πme
ode is under operation in reactive gas such as oxygen, which
might make the surface of the cathode insulating, an ac volt- where I0 is the modified Bessel function. For the case when
age at high frequency can be applied to the target (Anderson Vrf Te the modified Bessel function can be approximated
by
et al 1962, Davidse and Maissel 1966, Logan 1990). Often, 1/2
these discharges are operated at 13.56 or 27.12 MHz. Due Te
I0 (Vrf /Te ) ≈ exp(Vrf /Te ) (17)
to their lower mass, the electrons are able to respond to the 2πVrf
temporal variation of the electric field, while the more mas- so equation (16) becomes
sive ions remain mostly stationary. The more mobile electrons
are repelled for most of the radio frequency (rf ) cycle leading Te M 2πVrf
Vbias = Vrf + ln − ln (18)
to a positive space charge in front of all surfaces. As there is 2 2πme Te
no dc path through the insulating electrode, the time-averaged
flow of electrons and ions to the electrode during each rf cycle which to a zero-order gives Vbias ≈ Vrf . This value is increased
must be equal to preserve charge neutrality. To preserve overall by the thermal term and is decreased since the electrons reach
charge neutrality, a time-averaged negative charge accumu- the electrode only over a finite time interval. This result applies
lates on the surface of the insulator (the target). The electrons, to the driven cathode sheath for a highly asymmetric discharge
due to their high mobility, can provide enough charge over just with small powered electrode (the target) and large grounded
a fraction of the rf cycle to neutralize the positive ion charge counter electrode (or chamber walls).
13
Figure 15. The (a) voltage waveform across the discharge and (b) the current density at an rf plasma boundary.
The advantage of diode sputtering is efficient use of the

target material as the ion flux is nearly uniform across the
target surface, and the electrode area does not need to be pla-
nar. Sometimes an independent source of electrons is added in
order to sustain the discharge rather than relying entirely on the
generation of secondary electrons by ion bombardment of the
cathode target. This can be achieved by adding a thermionic
emitter (e.g. a heated filament). Then the process is referred
to as triode sputtering. In the triode arrangement, a dc sput-
ter discharge can be operated at working gas pressure as low
as 0.2 Pa. Due to the low deposition rate the dc diode sput-
ter discharge is no longer employed in industrial settings, but
discussed here for educational and historical purposes.
Figure 16. A schematic of the geometrically asymmetric rf diode
sputter tool. Based on Kawamura et al (1999). 3.3. Magnetron sputtering discharges
3.2.2. Deposition rate. The rate at which the film-forming A magnetron sputtering discharge is a magnetically enhanced
species are sputtered off the cathode target depends on the flux, diode sputter tool. It is based on magnetically trapping elec-
type and energy of the incident particles, the working gas ions trons in the cathode vicinity. This is typically achieved by
as well as the target material. The sputter rate is proportional placing permanent magnets near the back of the cathode tar-
to the discharge current when the applied voltage is kept fixed. get. This method is usually referred to as dc magnetron sput-
The maximum achievable discharge current density is roughly tering (dcMS) when driven by dc voltage or current source.
1 mA cm−2 and the deposition rate 0.2 nm s−1 at best. Further- A schematic of a circular planar magnetron magnet assem-
more, the sputter power efficiency (sputtered atoms/ion-volt) bly, including the magnets and the cathode target, is shown in
is low in these discharges. Note that rf sputtering can be per- figure 17. The magnetic field lines arch, from a center mag-
formed at lower working gas pressures (<1 Pa) than is viable net to an outer magnet ring, above the target surface. The
for dc diode sputtering. However, the sputter rate is very low as various magnetron sputtering configurations typically utilize
dielectrics can have sputter yields as low as one tenth of those a static magnetic field in the range ∼20–50 mT. The rela-
of metals. The low deposition rate is a significant drawback to tively weak magnetic field confines the electrons near the cath-
these tools. ode target surface while the ion trajectories are not directly
14
Bogaerts 2009, Kolev et al 2005). The discharge voltage

thereby falls over both the cathode sheath and an extended pre-
sheath or the ionization region, i.e., VD = VSH + VIR , where
VSH is the sheath potential and VIR the potential drop across
the ionization region. Unlike non-magnetized dc diode sputter-
ing discharges, which are primarily maintained by ion-induced
emission of secondary electrons accelerated in the cathode
sheath (Gudmundsson 2020, Gudmundsson and Hecimovic
2017), the presence of a magnetic field leads to a potential VIR
across the IR which enables Ohmic heating of the electrons
(Brenning et al 2016), and describes locally absorbed power
by the electrons within the IR. Ohmic heating accounts for a
significant fraction of the electron power absorption in dcMS
discharges, while it is believed to be the dominating electron
power absorption mechanism in high power impulse mag-
netron sputtering (HiPIMS) discharges (Brenning et al 2016,
Huo et al 2017, 2013). For discharges with a higher fraction of
Figure 17. A schematic of a circular planar magnetron magnet Ohmic heating over total electron heating (Ohmic heating plus
assembly, the magnets, the anode, and the cathode target. The
magnetic field lines arch, from the center magnet to the outer sheath energization), the same discharge current can be main-
magnet ring, above the target surface. The magnetic field B is radial tained at a lower discharge voltage, as the discharge becomes
and the electric field E is axial (along the z axis) and the E × B drift more energy efficient (Rudolph et al 2022).
path is azimuthal above the cathode target surface and the resulting The main advantage of magnetron sputtering among the
azimuthal current. Reprinted from Anders (2011), Copyright (2011), PVD techniques is that the microstructure and properties of
with permission from Elsevier.
the deposited film or coating can be controlled by low-energy
influenced (Krüger et al 2018). This creates a situation where ion bombardment during the deposition process. This allows
the electrons are magnetized, while the ions are not magne- one to control the reactivity and kinetics of the film-forming
tized. The electrons follow a bouncing motion along the mag- species and to vary the micro- or nanostructure of the deposited
netic field lines B and also experience the electric field E from materials, their phase composition, grain size and orientation,
the applied target voltage. As a result, electrons move with an density, and internal stress. Magnetron sputter deposition tech-
E × B drift along a torus in azimuthal direction as illustrated niques are currently the most widely used processes for thin
in figure 17. The confinement of electrons and of ions, how- film deposition and surface engineering treatments. The appli-
ever, remains still coupled, because any local deviation of ions cations span from thin metal layers in microelectronic circuits,
with respect to the electrons, being bound to the magnetic field protective layers on cutting tools, to optical films on architec-
lines, induces electric fields that also confine the ions. tural glass that can be few square meters in size. Depending
Due to the magnetic confinement of the electrons, the dis- on the application the magnetron sputtering discharge exists
charge voltage is lowered, the working gas pressure can be in a number of configurations and various power sources are
decreased, and the deposition rate is increased substantially, utilized as discussed in the following subsections. A more
compared to non-magnetized dc diode sputtering (Chapin detailed discussion of the magnetron sputtering discharge is
1974, Waits 1978). The magnetron sputtering discharge is typ- given in a recent review (Gudmundsson 2020) and an earlier
ically operated at working gas pressure in the range 0.1–1.5 Pa review with a more industrial emphasis by Kelly and Arnell
and with dc cathode voltage of 300–700 V. These are signifi- (2000).
cantly lower values for both the working gas pressure and the
discharge voltage than those required to operate a diode sputter 3.3.1. Magnetron sputtering discharge configurations. In the
tool and discussed in section 3.2. When operating with dc volt- 1960s, experiments were conducted where a radially sym-
age or current source the resulting current densities are in the metric quadrupole magnetic field was superimposed onto an
range 4–60 mA cm−2 and power densities are several tens of abnormal glow discharge, which gave a significantly increased
W cm−2 (Waits 1978) and the electron density in the substrate ion current density at the cathode (Kay 1963). The early exper-
vicinity is typically in the range 1015 –1017 m−3 (Gudmunds- iments lead to construction of sputter sources that were com-
son 2020). The static deposition rate can be up to 10 nm s−1 . prised of two coaxial cylinders. These were either cylindrical-
However, in dc operation the film-forming material constitutes post (the cathode is the inner cylinder and the anode the outer
almost only neutral atoms and the degree of ionization of the cylinder) (Wasa and Hayakawa 1967a, 1967b, 1969) or hol-
sputtered material is generally very low, often on the order low cathode or inverted (the cathode is the outer cylinder and
of 1% or less, and the majority of the ions bombarding the the anode the inner cylinder) (Gill and Kay 1965, Thornton
substrate are ions of the working gas. and Penfold 1978) magnetron sputter sources. The static mag-
The presence of a transverse magnetic field enables a netic field was then superimposed along the axis of the dis-
potential drop to exist outside the cathode sheath, the region charge, transversely to the electric field (radial) which main-
of dense plasma in the vicinity of the cathode target, the tains a glow discharge between the electrodes. This creates an
ionization region (IR) (Bradley et al 2001, Bultinck and electron trap where the electron E × B drift currents close on
15
themselves. The magnetic field is made strong enough to con-

fine the electrons but not the ions. Despite being important
for the development of magnetron sputtering, the cylindrical
configurations are not in much use today.
The planar magnetron sputtering configuration was intro-
duced in the early 1970s (Chapin 1974, 1979). The core of
the magnetron sputtering discharge in the planar configuration
is the magnetron magnet assembly, shown in figure 17. The
magnetron magnet assembly is composed of the cathode target
and an array of static magnets or electromagnets. In the planar
configuration the cathode target is a circular disk or rectangu-
lar plate. The magnets are arranged such that a central magnet Figure 18. (a) A three-dimensional rendering of a magnetron
forms one pole and a magnet ring or a ring of magnets placed assembly, and (b) a schematic sketch of the cross-section of the
along the edge of the cathode target forms the second pole, as magnetron assembly. Reproduced with permission from Layes
seen in figures 17 and 18. The cathode target and the magnets (2021)
often sit on a water-cooled block of solid copper to prevent
overheating. The planar configuration is often used in labo-
ratory settings, with a small circular cathode target, typically
5–15 cm in diameter. In this planar circular magnetron con-
figuration, the static magnetic field is arranged in such a way
that the electrons drift azimuthally. Due to the magnetic con-
finement of the electrons, an intense plasma is concentrated as
a donut or a torus-shaped ionization region that sits above the
target. Consequently, the plasma is not uniform over the cath-
ode target. The ions originate from this non-uniform plasma
region and bombard the target, and the ion current density Figure 19. (a) Two cylindrical rotatable cathode targets and (b) the
is peaked and maximum ion bombardment occurs over the ionization region which is static while the target rotates. Reproduced
region where the magnetic field is tangential to the cathode with permission from Vetushka et al (2015) © Copyright 2006-2022,
surface (Wendt 1988, Wendt et al 1988). Therefore, a charac- Society of Vacuum Coaters (SVCTM ).
teristic erosion groove forms in the target surface, referred to
as the racetrack (Chapin 1974, Nakano et al 2017). The erosion due to a larger useful target inventory, the target cleanliness,
groove determines the target utilization, the target lifetime, and an increased target material utilization, improved target cool-
the efficiency of the material usage and is therefore a major ing and increased power density, as well as excellent process
drawback of the planar magnetron arrangement in comparison stability for reactive depositions (Blondeel et al 2006). Other
to dc diode sputtering. Typically, the target utilization is below advantages of using cylindrical rotating targets is that the ther-
30% for planar targets. For industrial applications utilizing the mal load is distributed over the entire circumference of the
planar configuration, the cathode targets are often linear (rect- tubular target instead of a localized heating in the racetrack
angular), and larger, up to meters long. A prominent example region in the case of a stationary target. The longer production
are linear planar magnetron sputtering discharges to deposit runs are the results of more target material being available for
thin ceramic heat insulation layers on architectural glass panes the sputtering process and due to the rotation, the target uti-
with a substrate size of 3 m × 6 m. Here, the glass panes are lization is very high and can be higher than 80%. Also, for the
guided on a conveyor belt underneath a 3 m long rectangular rotating cylindrical targets the arching zone is limited to a ring
magnetron assembly. shaped area at each end of the tube, which remains contami-
For deposition on large area solar panels, architectural nated, as other areas are continuously sputtered and therefore
glass, and display panels, large cylindrical cathode targets, as remain clean. This results in reduced machine down time and
shown in figure 19(a), are commonly utilized. This configu- increased coater throughput. A drawback is that the availabil-
ration is referred to as rotatable cylindrical magnetron sput- ity of the target materials tends to be somewhat limited and the
tering. The magnet assembly is installed inside the cylindrical cost of the target can be high (Vetushka et al 2015).
cathode tube (McKelvey 1982, Wright and Beardow 1986). In
this setup the magnet assembly is stationary with respect to 3.3.2. Magnetic field configuration. As low energy ion bom-
the discharge chamber walls, and so is the ionization region, bardment is beneficial for the film properties, one approach to
while the cylindrical cathode target rotates. The stationary ion- increase the influence of ion bombardment on the film growth
ization region formed over two rotating cylindrical cathodes was the development of unbalanced magnetron assemblies
can be seen in figure 19(b). In operation, as the cathode target (Rohde 1994). It is referred to as a conventional or balanced
rotates, it is continuously exposed to the plasma zone result- magnetron assembly when all the magnetic field lines that cre-
ing in a uniform erosion around 360◦ of the target surface, and ate the magnetic electron trap form closed loops between the
the process is sometimes called fully face erosion. Among the magnetic poles as seen in figure 20(a). In reality a perfectly bal-
benefits of using rotatable cathodes are longer production runs anced magnetron assembly is not easy to achieve. However, a
16
Figure 20. A schematic of the magnet configuration in planar magnetron sputtering discharges. The three cases, (a) all the field lines that
originate from the central magnet enter the annular magnet (balanced), (b) all the field lines originate from the central magnet, while some
do not enter the annular magnet (unbalanced type I), and (c) all the field lines originate from the annular magnet, and some do not enter the
cylindrical central magnet (unbalanced type II). Reprinted from Gudmundsson and Lundin (2020), Copyright (2020), with permission from
Elsevier.
perfectly balanced magnetron assembly is not necessarily the a joining of the magnetic field lines, in a mirrored closed
optimal magnetic field design, because the unbalanced con- arrangement. That way the magnetic field lines are linked and
figuration created by a difference in the strength of the center maximize the trapping of electrons. Then the losses to the
magnet and the magnets along the periphery of the magnetron chamber walls are low and the substrate sits in a high density
assembly allows plasma also to escape toward the substrate. By plasma. It has been demonstrated that operation in the closed
strengthening the central magnet with respect to the outer mag- field configuration results in a significant increase of the ion-
net, some of the field lines are directed to the chamber walls to-atom ratio incident on the substrate as compared to a single
and the plasma density in the substrate vicinity is low. This target unbalanced magnetron, while the deposition rate is not
arrangement is called unbalanced magnetron assembly of type significantly influenced (Kelly and Arnell 1998). On the con-
I (Window and Savvides 1986) and is shown schematically trary, in the mirrored arrangement, the field lines are directed
in figure 20(b). When the magnetic field of the outer pole is toward the chamber walls. In that case some of the energetic
strengthened relative to the central pole some of the field lines secondary electrons follow these field lines and are lost from
are directed toward the substrate and others are closed between the discharge, resulting in a low plasma density in the substrate
the central and outer magnet poles. Consequently, the elec- region.
trons can travel toward the substrate region which increases the
ionization in the substrate vicinity. This allows ions to reach 3.3.3. The discharge voltage and current waveforms.
the substrate, and then by biasing the substrate, the ion bom- Depending on the application, the applied target voltage can
bardment energy can be controlled. This arrangement of the be direct current (dc), radio frequency (rf ) or pulsed. This can
magnetron assembly is referred to as unbalanced magnetron be achieved by a power, current, or voltage source, depending
assembly of type II (Window and Savvides 1986) and is shown on the regulation method applied. A particular waveform is
schematically in figure 20(c). selected to avoid instabilities, to provide a high ionization
In order to exploit the magnetron sputtering technology flux fraction of the sputtered species, to allow sputtering from
commercially, sputter systems with two or more magnetron two targets, or to make it possible to sputter from insulating
assemblies can be arranged to maximize the plasma density targets. A prominent instability is the so called arcing, which
in the substrate vicinity. For two magnetron assemblies facing is essentially the onset of an unwanted cathodic arc mode of
each other it can be either described as ‘mirrored’ (like poles the discharge, detectable by the appearance of cathode spots
face each other) or as ‘closed’ (opposite poles face each other) and a simultaneous sharp reduction in voltage between the
field configuration (Sproul et al 1990). In the latter case the anode and cathode, and sudden rise in the discharge current
closing of the magnetic field lines forms a magnetic electron (Anders 2006). Such instabilities are either initiated by
trap that confines the electrons in the region between the cath- thermionic emission of electrons at hot spots on the cathode
odes, where the substrate is typically placed. The effect can be surface or by the dielectric breakdown of insulating layers.
further amplified by combining several unbalanced magnetron These extreme events are detrimental for the coating being
assemblies into a single system such as the closed-field unbal- deposited as well as for the power supplies and need to be
anced magnetron sputtering (CFUBMS) arrangement, which avoided. In most commercial power supplies, an arc control is
provides a high plasma density in the substrate vicinity (Kelly implemented, which temporarily shuts off the power in case
and Arnell 2000, Monaghan et al 1993). The industrial sys- of current spikes induced by these arcs.
tems often consist of multiple long rectangular cathodes [even The dc magnetron sputtering discharge is ideal for deposit-
numbers of magnetron assemblies (2, 4, 6, 8)] that surround ing thin metallic films from electrically conducting targets.
rotating workpieces. Typically, the magnetic configuration of For the deposition from thick electrically insulating (often
the neighboring magnetron assemblies alternates to achieve compound) target materials, rf power needs to be applied. The
17
discussion in section 3.2.1 on rf diode sputtering, including including the chamber walls become covered with an insulat-
the formation of dc-self-bias, applies to rf magnetron sputtering oxide. This arrangement is often applied when depositing
ing (rfMS) as well. As the cathode target is of different surface on large area architectural and automotive glass with large
area than the anode, a dc self-bias develops and is responsible rotating cathode targets (Brückner et al 2005) as shown in
for the ion acceleration through the sheath and onto the cath- figure 19(a) with two rotating cylindrical targets side by side.
ode target. The dc-self-bias biases the plasma positively with The ac frequency is nominally 40 kHz, however, sometimes
respect to the electrode and appears for all target materials, the frequency is varied between 10 and 100 kHz to adjust
metallic or dielectric. An important example of the applica- the power delivered to the cathode by frequency modulation
tion of rfMS is the deposition of ZnO, a transparent conducting (Scherer et al 1992).
oxide, thin films from a ZnO target (Petrea and Stamate 2021, The waveform can also be unipolar high power pulses and
Stamate 2020). The main advantage of sputtering a ceramic then the process is referred to as high power pulsed mag-
target in this case is the fact that the oxygen atom for film for- netron sputtering (HPPMS). By pulsing the power to the cath-
mation is released from the target surface during sputtering ode target with a high peak power density, a high electron
and therefore it is easier to achieve stable deposition condi- density can be achieved. The unipolar pulse can be a sin-
tions instead of having to deal with the complications that arise gle pulse or be composed of a train of pulses, micropulses,
with feedback control in reactive sputtering (see section 3.3.5). the different approaches all fall under the HPPMS umbrella.
A significant drawback for the application of rfMS is that the When the power density is very high and the duty cycle short
deposition rate is rather low and the costs of manufacturing the process is referred to as high power impulse magnetron
ceramic targets through powder metallurgy can be high. sputtering (HiPIMS) (Anders 2017, Gudmundsson et al 2012,
Pulsed magnetron sputtering is utilized in a number of Kouznetsov et al 1999, Sarakinos et al 2010). In HiPIMS oper-
applications. The pulse can be either an asymmetric bipolar ation the repetition frequency is low, typically in the range
pulse or an unipolar pulse depending on the application. One 50–5000 Hz, the pulses are short 10–400 μs, the duty cycle
of the most important application of magnetron sputtering is is short or 1%–3%, and the peak power density is high or
reactive sputtering to deposit compound films such as oxides, pt > 0.5 kW cm−2 . These discharges reach high electron den-
nitrides or carbides, in which a metal target is sputtered inside sities and exhibit significant ionization of the sputtered species.
a discharge of reactive gas. Reactive sputtering is a topic of When the unipolar pulse shape is modulated, the pulses are
significant importance and will be discussed in section 3.3.5, longer, and the peak power density lower than in HiPIMS oper-
and is typically performed using asymmetric bipolar wave- ation, it is called modulated pulse power magnetron sputtering
form. The asymmetric bipolar mid-frequency magnetron sput- (MPPMS). Pulsing the discharge not only increases the elec-
tering discharge was designed to optimize the deposition of tron density and the ion flux to the substrate, it also increases
insulating films from conductive targets trough reactive sput- the average particle energy. Furthermore, the capabilities of the
tering (Sellers 1998). Then the polarity of the target voltage is sputter process are enhanced through control of pulse duration,
alternated between negative and positive in each period and repetition frequency, and duty parameters.
the negative voltage pulse amplitude is larger than the pos- During the initial stage of the pulse in MPPMS operation (a
itive (roughly 10%–20% of the negative voltage amplitude) few hundred microseconds) the power level is moderate (sim-
voltage pulse amplitude, and there is no off time between ilar to dcMS levels), followed by a high-power pulse (lasting
the different polarities. A significant portion of each cycle is a few hundred microseconds up to a millisecond). The result-
spent in the sputter mode, and the deposition rate from an ing pulse is referred to as a macro-pulse. The macro-pulse can
asymmetric waveform can approach that of a dcMS. There- be up to 3 ms long, and the repetition frequencies are in the
fore, asymmetric bipolar magnetron sputtering is sometimes lower end of HiPIMS operation. The macro-pulse is composed
referred to as pulsed dc magnetron sputtering. The asymmetric of a train of shorter micro-pulses that appear with frequencies
bipolar waveform is applied to prevent arcing on the target sur- in the range of several tens of kHz. The on- and off-times of
face during the deposition of non-conducting films. The rep- these micro-pulses, which are typically up to several tens of
etition frequency is typically in the medium frequency range μs wide, as well as their frequency can be altered within the
(10–250 kHz) when depositing dielectric films (Schiller et al macro-pulses. Using this approach, varying the micro-pulse
1993, Sellers 1998). frequency and the ‘on’- and ‘off’-times, arbitrary tailored cath-
Sometimes symmetric bipolar or mid-frequency alternat- ode voltage and discharge current waveforms can be created,
ing current (ac) waveforms are applied for the reactive sputter including what appears to be a multi-step pulse (Chistyakov
deposition of oxide coatings from two targets, called dual mag- and Abraham 2009, Hála 2011, Hála et al 2012, Liebig et al
netron sputtering (Este and Westwood 1988, Heister et al 2000, 2011, Lin et al 2011). One variation to the power delivery is to
Scherer et al 1992). The two targets, often placed side by side, apply packets (or macro-pulses) that consist of a sequence of
are both connected to the same symmetric bipolar pulser. In tightly packed micro-pulses whose duration is only a few μs,
this arrangement one target serves as an anode for the sys- referred to as deep oscillation magnetron sputtering (DOMS)
tem, while the other serves as the cathode target. The roles (Ferreira et al 2016, 2014). By varying the oscillation pulse
of the targets are switched when the discharge voltage polar- ‘on’ and ‘off’ times, the peak target voltage and discharge cur-
ity changes, and a clean target takes over as an anode in each rent can be tailored. The duration of the macro-pulse in DOMS
cycle. This approach eliminates what is referred to as the dis- is 1–3 ms, and the repetition frequency is typically below
appearing anode problem, which can occur when all surfaces 500 Hz. For easier comparison the power delivery in dcMS,
18
back-attracted by the electric field to the biased target. This is

referred to as the return effect (Christie 2005) and it reduces the
growth rate per invested power in comparison to dcMS depo-
sition. The resulting low deposition rate is a major drawback
to the high power pulse magnetron sputtering technology. In
HiPIMS operation, there are two goals: to achieve a high ion-
ized flux fraction of the sputtered target material and to deliver
a high deposition rate. The former always comes at the cost
of the latter, which is referred to as the HiPIMS compromise
(Brenning et al 2020). Over the years there have been signifi-
cant efforts to understand and find ways to alleviate the low
deposition rate in HiPIMS deposition. It is well established
Figure 21. Schematic representation of the power delivery for dcMS that the deposition rate depends on the magnetic field strength,
and the unipolar pulsed magnetron sputtering discharges HiPIMS and increases with decreasing magnetic field strength (Hajiho-
( f = 50 Hz), and MPPMS ( f = 50 Hz) when operated at the same
average power P = 200 W with 5 cm diameter Nb target. seini et al 2019, Mishra et al 2010). A promising approach
Reprinted from Hála et al (2012), Copyright (2012), with has to increase the deposition rate has been to shorten the
permission from Elsevier. pulse length (Rudolph et al 2020, Shimizu et al 2021). It has
also been proposed to optimize the HiPIMS discharge opera-
HiPIMS, and MPPMS discharges is shown schematically in tion through mixing two different power levels (Brenning et al
figure 21. Over the past few decades there has been substantial 2021). Standard HiPIMS pulses create the ions of the film-
development in the design and construction of the pulser units forming material followed by an off-time, during which no
that deliver these unipolar high power pulses as discussed by voltage (or, optionally, a reversed voltage) is applied, letting
Hubička et al (2020). The pulser units are currently based on the remaining ions in the magnetic trap escape toward the sub-
insulated-gate bipolar transistors (IGBT) that provide a good strate. After these off-times, a long second pulse with lower
control over the pulse length, and have large storage capaci- amplitude, operated in the dc magnetron sputtering range, is
tors that maintain the discharge voltage throughout the pulse applied (Brenning et al 2021, Lou et al 2021). During this low
length. power pulse, which is continued up to the following HiPIMS
The goal of the different pulsing schemes is usually to pulse, mainly neutrals of the film-forming material are pro-
achieve better control over the process and therefore the prop- duced. This pulse pattern makes it possible to achieve separate
erties of the deposited films. When the deposition flux contains optimization of the ion production, and of the neutral atom pro-
a high fraction of energetic ions of the sputtered species, it can duction, that constitute the film-forming flux to the substrate.
be utilized to determine the film microstructure. The energy The low-power pulse is a much more efficient way of creating
of these ions is a few eV and originates from the sputter pro- neutral atoms of the sputtered species to maintain a high depo-
cess itself, but the energy is also due to acceleration in the sition rate, while the high power pulse creates ions to a desired
plasma sheath in front of the substrate. Furthermore, the ion predetermined ionized flux fraction. The optimum power split
bombarding energy can be controlled by applying substrate is decided by the lowest ionized flux fraction that gives the
bias (Hubička et al 2020). The control of the ion bombard- desired film properties for a specific application.
ing energy by substrate bias provides control over the E∗ axis The disadvantageous return effect can be mitigated by
of the structure zone diagram in figure 3. Such an energetic applying a positive pulse to the sputter target, right after the
deposition flux can therefore enable superior film properties high power negative sputter pulse, as suggested by Nakano
due to the enhanced adatom mobility of the condensing species et al (2013), (2010), (2014). This raises the plasma potential
even for deposition at relatively low substrate temperature. and accelerates ions out of the ionization region, toward the
Increasing the ionized flux fraction increases the film mass growing film (Keraudy et al 2019). This positive pulse can be
density in the deposited film as was demonstrated by compar- few tens to few hundred volts and this approach is referred to
ing dcMS and HiPIMS depositions for a number of elemental as bipolar HiPIMS or HiPIMS with a positive kick pulses.
metals (Samuelsson et al 2010). Furthermore, increasing frac-
tion of ionization in the deposition flux has influence on the 3.3.4. Instabilities and plasma patterns of magnetron plasmas.
surface morphology, a smoother surface appears as the ionized Magnetron sputtering discharges belong to the group of E × B
flux fraction is increased, as can be clearly seen by compar- discharges which are known to be subjected to various plasma
ing figures 2(a)–(c). This is indeed observed when comparing instabilities. These may be driven by gradients in plasma den-
the surface morphology of thin films deposited by dcMS and sity, magnetic field, and temperature (Boeuf 2014). Such insta-
HiPIMS, where the latter deposition method, providing higher bilities have been observed in magnetron sputtering discharges
ionized flux fraction, results in significantly smoother film sur- both experimentally and through simulations. Some instabil-
faces (see e.g. Sarakinos et al (2007) and/or Hajihoseini and ities may lead to regular moving wave patterns that appear
Gudmundsson (2017)). over a wide range of power delivered to the plasma. Examples
However, due to the high ionization degree of the film- are high frequency instabilities such as the electron cyclotron
forming species in these discharges and as the sputtered drift instability with a traveling pattern with a wavelength in
neutrals are ionised in close vicinity of the target, they can be the mm range (Tsikata and Minea 2015) to low frequency
19
instabilities such as the formation of spokes with a traveling powers (Biskup et al 2018). At the same time, the growth
pattern of a region of enhanced ionization with a wavelength rate per invested power significantly decreases with increas-
of a few centimetres (Brenning et al 2013). From an engineering HiPIMS pulse power due to the return effect close to the
ing perspective, the occurrence of these instabilities or plasma target. However, this decrease in the growth rate was mitigated,
inhomogeneities are an unwanted aspect for any application. when spokes form in the HiPIMS discharges. The modulation
However, the traveling velocity of these patterns is very high, of the electrical potential in front of the target surface by the
so that any inhomogeneity in the plasma discharge has aver- rotating spokes has been corroborated by emissive probe mea-
aged out in the deposition flux. Moreover, the plasma patterns surements by Panjan and Anders (2017) for low power dis-
are also the source of internal electric fields in the plasma charges and by synchronised Langmuir probe measurements
that enhance the transport of charged species from the target (Held et al 2020) for high power cases. The modulation of
to the substrate and thereby mitigate the return effect. Note the plasma potential due to the presence of a spoke largely
that electron transport across magnetic field lines is governed removes the electric field in the magnetic trap region point-
by instabilities as opposed to classical diffusion. It is gener- ing toward the target. This mitigates very efficiently the return
ally accepted that instabilities and anomalous transport across effect in HiPIMS plasmas. In addition, azimuthal electric fields
the magnetic field lines plays a significant role, however there also induce an anomalous electron transport out of the mag-
exist currently no consensus nor real quantification of these netic trap region and thus enhance the transport of plasma to
phenomena in magnetron sputtering discharges. the substrate. Based on particle-in-cell Monte Carlo collision
Most prominent of these instabilities are spokes or bright simulations it has been proposed that the spokes appear due to
plasma emission zones that move azimuthally along the modified Simon–Hoh instability that evolves into ionization
plasma torus with a specific mode number above the cath- instability and is sustained by local electron power absorption
ode (Anders 2012, Ehiasarian et al 2012, Kozyrev et al 2011). induced by ∇B drift along a double layer (Boeuf and Takahashi
These ionization zones rotate in the E × B direction with 2020a, 2020b).
velocities in the range of typically 10 km s−1 at high tar-
get power densities. They were coined spokes due to their 3.3.5. Reactive magnetron sputtering. Reactive sputtering is
similarity with rotating spokes observed in Hall thrusters the process of sputtering an elemental target in a gas mix-
(Janes and Lowder 1966). The velocity of the spokes corre- ture of the inert working gas and a reactive gas (e.g. O2 , N2 ,
sponds to the velocity of a plasma excitation or plasma wave, CH4 , etc) to deposit compound films. Compound films can
because they travel slower than the electrons with a velocity of also be deposited by sputtering a compound target. A com-
100 km s−1 along the plasma torus (Krüger et al 2018, Rauch pound refers to chemical binding between at least two chem-
and Anders 2013), but faster than the ions with a veloc- ical elements where at least one is a metal and the other
ity of a few km s−1 (Poolcharuansin et al 2012). A typi- is non-metal (Strijckmans et al 2018). By reactive sputter-
cal image of a plasma pattern on a rectangular chromium ing a range of compounds can be deposited from a low-cost
target (3.5 × 10 ) taken with a fast intensified charge- metal target by addition of an appropriate reactive gas to the
coupled device (ICCD) camera is shown in figure 22 (Preissing noble working gas. This includes hard transition metal nitride
2016). The rotation direction is in the E × B direction at high based coatings, transparent conductive oxides, dielectric lay-
plasma power and in retrograde E × B direction at low plasma ers and photo-catalytically active layers. Reactive magnetron
powers (Anders and Yang 2017). sputtering is therefore of significant industrial and technologi-
The spoke phenomenon is very dynamic since the dis- cal importance as the majority of commercially important thin
charge current during a HiPIMS pulse strongly varies within films and coatings are compounds (Kelly 2011). Such coat-
each single pulse. In industrial applications, with large rect- ings are typically deposited by reactive magnetron sputtering
angular magnetron assemblies elongated spokes appear on the using asymmetric bipolar mid-frequency, rf or unipolar pulsed
straight parts of the target racetrack, and ‘bunching’ of spokes (HiPIMS) waveforms. Reactive sputter deposition is a very
is observed in the curved parts of the racetrack (Anders and important topic and for further details the readers are referred
Yang 2017, Bobzin et al 2017, Preissing 2016). Such a bunch- to a recent tutorial by Strijckmans et al (2018), a review by
ing can be easily explained since the E × B drift velocity scales Sproul et al (2005), and in the context of HiPIMS (R-HiPIMS),
with 1/B and because the velocity of spokes is proportional to a tutorial by Anders (2017).
this drift velocity they slow down in the corners of a rectan- The sputter yield of the compound material is typically sub-
gular target where the magnetic field is stronger. The plasma stantially lower than the sputter yield of the elemental target
density in a spoke is of the order of 1019 m−3 . Ions exhibit also material. Consequently, the deposition rate decreases as the
very high temperatures in the range of a few eV, due to thermal- flow rate of the reactive gas is increased and compound is
ization of the energetic sputtered metal neutral flux (Held et al formed on the target surface, and fewer atoms of the elemen-
2018). The ion energy distribution is dominated by low energy tal target material are sputtered and fewer reactive gas species
ions, but exhibits a high energy tail. Ions are also ejected side- are consumed. This appears as a sudden and sharp rise in the
ways with an average velocity that is different in or against reactive gas partial pressure and the resulting deposited film
the E × B direction (Franz et al 2016, Panjan et al 2014, Yang becomes rich in reactive gas species. It also indicates that the
et al 2015). The ionized flux fraction continuously increases relationship between the compound film composition and the
with increasing peak power density up to values close to 80%, flow rate of reactive gas is non-linear. Similarly, the deposi-
indicating film growth by incident ions only at very higher tion rate shows a non-linear dependence on the flow rate of the
20
Figure 22. ICCD image of spokes, rotating along the race track above a chromium target (3.5 × 10 ) at different target current densities, as
indicated. Reproduced with permission from Preissing (2016).
reactive gas. The difference in sputter yields causes the depo-

sition rate decrease and increase not to occur at the same value
of the flow rate of the reactive gas. This shows up as a hys-
teresis, and the separation width between the decrease and
increase defines the width of the hysteresis region. This is
shown schematically in figure 23 (from Kubart et al (2020)),
where the three main operating regions of a reactive magnetron
sputtering process are indicated. In the metal mode, all the
supplied reactive gas is incorporated into the deposited metal,
and the target surface is free of compound. The mass deposi-
tion rate increases slightly with increasing reactive gas flowrate Figure 23. A schematic showing the deposition rate evolution with
up to the point where the deposition rate drops abruptly as a reactive gas flow. Pronounced hysteresis (wide transition region)
transition from metal to compound mode takes place (seen as defined by the transitions A–B and C–D is shown. Reprinted from
Kubart et al (2020), Copyright (2020), with permission from
seen as a jump from point A to point B in figure 23). Here,
Elsevier.
all the sputtered metal is converted into compound material,
and the excess reactive gas forms a compound layer on the
target surface. The deposited films are stoichiometric, and fur- reactive gas into the chamber in attempts to achieve a stable
ther addition of reactive gas does not have much influence. operating point. But this approach leads to operational prob-
The compound mode is maintained as the reactive gas flow is lems. At fixed power, as the reactive gas flow of the reactive gas
reduced until it reaches point C in figure 23. At this point the is increased, initially all of the reactive gas will be consumed
reactive gas flow is unable maintain the compound layer on by reaction with metal surfaces, including the cathode target
the sputter target surface, and a transition to the metal mode and the chamber walls. Therefore, a number of process param-
(point D) occurs. The presence of hysteresis implies that the eters change simultaneously when increasing/decreasing the
transition region A–B–C–D is ill defined. For low reactive gas reactive gas flow rate. Furthermore, the precision and speed of
flows, a metal mode is maintained with predominantly metallic the standard pressure gauges (e.g. vacuum ionization gauges)
film deposited at a high rate. This also means that it is possi- is often inadequate for control of reactive sputter deposition,
ble to deposit compound films of different stoichiometries and as their accuracy often depends on the type of gas. It is more
physical properties at two stable operating states, that corre- practical to determine the operating point by active feedback
spond to the same value of the flow rate of the reactive gas, control of the partial pressure of the reactive gas (Sproul et al
within the region of hysteresis. The hysteresis effect is there- 2005). This is possible as there is a monotonic relation between
fore simply due do to a competition between two opposing the partial pressure of the reactive gas pRG and the target com-
processes: the formation of a compound on the target surface pound fraction θ t without the appearance of hysteresis. The
and sputtering of the compound off the target surface. This for- steady state relation between pRG and the target coverage θt is
mation of compound material on the target surface is called Ji YCC θt
target poisoning. p RG = (19)
2κSRG (1 − θt )
Stoichiometric compound films can be deposited at rela-
tively high deposition rates, if one operates in the transition where Ji is the ion current density onto the target, YCC is the
zone between the metallic and poisoned mode (Sproul 1998). partial sputter yield of reactive gas atoms from the compound,
In some cases, a desired stoichiometry at the substrate sur- SRG is sticking (or incorporation) coefficient, which stands for
face requires an explicit operation of the reactive magnetron the probability of a neutral reactive species to bind with a free
process within the transition zone. Therefore, it is important metal site. Furthermore, the reactive gas flux is proportional
to identify a stable operating point within the transition zone. to the partial pressure of the reactive gas, and the constant of
Early on this control was achieved using mass flow control of proportionality is κ. According to equation (19), the partial
21
pressure pRG is a smooth curve, a function of the target com- 1982). Bombarding the growing film with ions of the noble
pound fraction. Therefore, the partial pressure is suitable as a working gas can create residual ion-induced compressive
control parameter, and the transition between the metal mode stress in the film, while bombarding the growing film with
and the compound mode is continuous (Sproul 1998). the ions of the film-forming material has several advantages:
A feedback control loop requires a precise measurement of improvement of the film quality, and improved step coverage
the parameter that characterizes the process state. However, and conformity (Greczynski et al 2019). Furthermore, by
direct measurement of partial pressures calls for specialized ionizing the sputtered species, the ion bombarding energy
instrumentation, which is not always available. This can be at the substrate can be controlled by applying a substrate
realized by quadrupole mass spectrometry of the reactive gas bias. Additionally, a directional deposition and collimation
species (Sproul et al 2005, Sproul and Tomashek 1984), by of these ions with the plasma sheath adjacent to the wafer is
optical emission spectrometry to detect the sputtered metal made possible. A PVD process where the flux of ions Γi is
species (Schiller et al 1987, 1982), or by using the cathode larger than the flux of neutrals Γn for the sputtered species, or
voltage as the feedback signal (Affinito and Parsons 1984). The Γi > Γn , is called ionized physical vapor deposition (IPVD)
mass spectrometer provides a direct reading of the particular (Hopwood 2000). Achieving highly ionized flux of the
reactive gas species while the optical emission spectrometry sputtered material was initially based on the application of
reading and the cathode voltage are indirect measures of the a secondary discharge to create a dense plasma between the
partial pressure. evaporation or sputter source of the film-forming material and
An optical emission spectrometer can detect an optical the substrate so that a large fraction of the sputtered atoms are
emission from metal species that are excited during the reac- ionized (see section 3.3.7).
tive sputter process. This signal can be applied in a feedback The dominating ionization process in IPVD discharges is
control loop which is then used to adjust the reactive gas flow electron impact ionization, and to ionize the sputtered species
rate often through a piezoelectric valve (Schiller et al 1987, the average distance it travels before being ionized has to be
1982), an approach that is referred to as a plasma-emission reasonably short. The ionization mean free path for electron
monitor. A monotonic relation exists between the emission impact ionization is given by
intensity of a characteristic line of the sputtered species and the
vs
degree of oxidation of the deposited layer. In a reactive sput- λiz = , (20)
ter process, the metal line intensity exhibits a pronounced drop kiz ne
with increased compound coverage of the target. The increased where v s is the velocity of the sputtered neutral atoms, kiz is the
target coverage is accompanied by a reduction in the ion cur- electron impact ionization rate coefficient and ne is the electron
rent onto the target which, consequently, causes a reduction in density. In dcMS operation ne ∼ 1015 –1017 m−3 and λiz ∼ 1 m
the metal sputter rate. These optical emission based methods and the fractional ionization of the sputtered atoms is expected
are fast and of relatively low cost. to be low. For ne ∼ 1018 m−3 and λiz is tens of cm. These are
Note that the pumping speed also plays a significant role. the typical parameters for a IPVD systems based on a sec-
At sufficiently high pumping speeds the hysteresis may be ondary inductively coupled plasma (ICP) or ECR discharge.
avoided (Kadlec et al 1986, Okamoto and Serikawa 1986). The For electron density of 1019 m−3 , the densities achieved in the
absolute value of such a critical pumping speed, however, can HiPIMS discharge, the ionization mean free path is of the order
often be unrealistically high and therefore it is often practically of one cm. The high electron density is therefore the key to
unfeasible. Also, it is not economical. achieve a high degree of ionization of the sputtered material.
The ionization of the sputtered species is typically given as the
3.3.6. Ionized physical vapor deposition and ionization ionized flux fraction of species s as (Hopwood 1998)
processes. In a sputter deposition process the sputtered
species are released from the cathode target as atoms or (s) Γ(s)
i
Fflux = , (21)
molecules. For magnetron sputtering discharges driven by dc, Γ(s)
i + Γ(s)
n
rf and asymmetric bipolar waveforms, the film-forming mate-
rial at the substrate consists almost entirely of neutral atoms. where Γ(s) (s)
i and Γn are, respectively, the ion and neutral fluxes
However, it is often desired to have ions of the film-forming of the species s arriving at the substrate or detector. The depo-
species bombarding the substrate, as it is well established sition rates are then recorded by manually recording the film
that low energy ion bombardment (Ei is below the lattice thickness at a chosen time on a readout unit connected to the
displacement threshold, ∼20–50 eV depending upon the ion quartz crystal micro-balance (QCM). The ionized fraction of
and deposited film) in the deposition process, has a significant a metal flux can be determined by measuring the total mass
influence on the deposited film microstructure and the film deposition rate and the mass deposition rate of neutral metal
properties (Petrov et al 2003) (see also figure 3). Ionizing the atoms only, as discussed by Green et al (1997) and Wu et al
sputtered material has several advantages: improvement of the (2010).
film quality, deposition on substrates with complex shapes,
enhancement and control of the reactivity in the growth 3.3.7. Magnetron sputtering with secondary discharge. Var-
process. This influences the crystallite orientation, grain size, ious approaches have been taken over the past few decades
the epitaxial temperature, the film mass density, as well as to increase the ionization of the sputtered species in mag-
stress in the film (Greczynski et al 2019, Greene and Barnett netron sputtering by adding a secondary discharge to create
22
a dense plasma in the region between the cathode target and

the substrate (Helmersson et al 2006). These systems con-
sist of a dcMS source and a secondary discharge which could
be either inductively coupled (Rossnagel 2000, Rossnagel and
Hopwood 1993, 1994, Wang et al 1999) or microwave-driven
(Musil et al 1991, Takahashi et al 1988, Xu et al 2001) dis-
charges. The secondary discharge creates a dense plasma that
ionizes a large fraction of the sputtered species as it passes
through. This technique was initially developed to deposit
metal layers and diffusion barriers into trenches or vias of high
aspect ratio in microelectronic fabrication (Hopwood 1998,
2000, Rossnagel 1999). Other methods of creating highly ion-
ized sputtered material include shaping the cathode target in
a particular way in order to confine the electrons, referred
to as hollow cathode magnetron sputtering discharge (HCM)
(Klawuhn et al 2000, Lai 2000).
The inductively coulped plasma (ICP) discharge is driven
from a non-resonant induction coil that is placed parallel to the
cathode target as shown in figure 24. The magnetron assem- Figure 24. A schematic of an ICP-MS in which a
bly is located on the top of the chamber. The sputtered species radio-frequency-driven inductively coupled discharge is placed
parallel to the cathode target in the region between the cathode and
transit the dense plasma, created by driving rf current through the substrate. Reproduced from Gudmundsson (2020). © The
the inductive coil, where they are ionized (Barnes et al 1993, Author(s). Published by IOP Publishing Ltd. CC BY 4.0.
Rossnagel 2000, Rossnagel and Hopwood 1993, 1994). The
rf coil is typically located a few cm below the cathode tar-
get surface and has diameter that is 20%–40% larger than
the substrate diameter (Rossnagel 2000). The inductive coil is
often driven at 13.56 MHz using a 50 Ω rf generator through
a capacitive matching network. The rf power is typically in
the range 200–1000 W resulting in an electron density in the
range of 1016 –1018 m−3 , that increases linearly with increased
applied rf power (Hopwood 1992, Hopwood et al 1993).
The ICP-assisted magnetron sputtering discharge is typically
operated at rather high working gas pressures, in the range
2–4 Pa, to thermalize the film-forming species which shortens
their ionization mean free path (Rossnagel 2000, 1999). The
inductively coupled plasma assisted magnetron sputtering
(ICP-MS) discharge is still very widely used in the micro-
electronics industry for deposition of metal films such as Figure 25. A schematic of an ECR-MS apparatus, two ECR
conductors, diffusion barriers, and adhesion and seed layers. discharge chambers are located at the opposite sites of the main
A secondary discharge can also be created by adding a processing chamber. A highly ionized plasma is created in the
ECR discharge in the region between the magnetron assem- region between the magnetron sputtering cathode target and the
substrate. Reproduced from Gudmundsson (2008). © IOP Publishing
bly and the substrate (Xu et al 2001), the cathode target can Ltd. CC BY 3.0.
be placed between the ECR discharge and the substrate (Berry
and Gorbatkin 1995, Gorbatkin et al 1996, Musil et al 1991, collisionless heating mechanism. Figure 25 shows a schematic
Ono et al 1984, Takahashi et al 1988) or the microwave power of an ECR-assisted magnetron sputtering discharge, where two
can be introduced to the discharge through coaxial-type cav- ECR discharges are placed between the cathode target and the
ity that surrounds the cathode target (Yoshida 1992). An ECR substrate. Electromagnets are placed around the periphery of
discharge provides high plasma densities (1017 –1018 m−3 ) each of the ECR discharge chamber to create magnetic field of
and the working gas pressures can be low (0.01–2 Pa). The 87.5 mT and a resonance zone within each of the chambers.
ECR discharge is based on wave absorption and requires This IPVD technique has also been demonstrated for ion-
application of a strong stationary magnetic field (87.5 mT at izing thermally evaporated metal by having it pass through
resonance) and microwave power (e.g., 2.45 GHz) which is an ECR discharge (Holber 2000, Holber et al 1993) as dis-
injected as a right-hand circularly polarized wave into a reso- cussed in section 2.1. For sputter deposition systems, based on
nance zone. The introduction of a magnetic field leads to a res- a secondary high density discharge, it has been reported that
onance between the microwave frequency (ω) and the electron a significant fraction of the sputtered species is ionized and
cyclotron frequency ω ce = eB/me within the discharge. Due to that the ionized flux fraction increases with increased power
the cyclotron resonance, the gyrating electrons rotate in phase to the secondary discharge. The ionized flux fraction for Cu
with the polarized wave and the wave energy is absorbed by a has been reported to reach values higher than 80% (Rossnagel
23
and Hopwood 1994). The ions of the sputtered material can

subsequently be accelerated toward the substrate to a desired
ion bombarding energy by applying a dc bias to the substrate.
3.4. Ion beam sputter deposition
Ion beam sputter deposition is a PVD technique that is used to

deposit high-quality thin films. Ion beam techniques are non-
thermal techniques in which the energy of the impinging flux
is controlled by an electric field. Ion beam deposition offers
the opportunity to influence the properties of the film-forming
species and consequently the thin film properties can be var-
ied by changing the ion beam and/or geometrical parameters
over a wide range. There are two approaches to apply ion
beams for thin film deposition (Harper 1978, Rossnagel and Figure 26. A schematic of an ion beam sputter deposition system.
Cuomo 1988). The first approach is primary ion beam depo- Reprinted with permission from Bundesmann et al (2021).
sition where the ion beam is composed of the film-forming Copyright 2021, American Vacuum Society.
material which is deposited at low energy (∼100 eV) directly
onto the substrate. The other approach is secondary ion beam
deposition where the ion beam is typically composed of ions β, can be varied. The scattering angle is
of an inert gas that are accelerated to high energy (hundreds to
thousands of electron volts). This ion beam is then directed at γ = 180◦ − (α + β). (22)
a target that constitutes the film-forming material and the ions
of the inert gas sputter the target. These sputtered film-forming Note that the range of feasible scattering angles is some-
species are then collected onto a nearby substrate. In both the what limited. The sputtering and scattering processes can be
ion beam processes the ions are generated at some distance described by the conservation of momentum and energy in a
from the substrate and the ion beam is directed onto the sub- binary elastic collisions. The ion–solid target interaction and
strate or a target. In the latter process the ion generation and sputtering are discussed in section 3.1.
acceleration (ion beam source), and the generation of the film- The ion source is often a gridded broad-beam ion source,
forming species (target), and thin film deposition (substrate) sometimes referred to as a Kaufman ion source (Kaufman and
are spatially separated. Typically the geometrical parameters Robinson 1989). They are composed of an electrically driven
(ion incidence angle and emission angle) as well as the ion discharge for ion generation and a multi-aperture grid system
beam parameters (ion species and ion energy) can be varied, for ion acceleration and extraction. The discharge can be a rf
which influences the energy and angular distributions of the driven inductively coupled discharge, a capacitive discharge
film-forming species, the sputtered target species, as well as or a microwave driven ECR discharge. One side of the dis-
backscattered primary species. charge volume has a set of grids or screens. As ions impinge
The main advantages of ion beam processing are excellent on the mostly open screens, they are accelerated by the applied
control of the flux and energy of the ions incident on either the voltages and leave the source in the form of a broad beam.
substrate or a target (for sputter deposition) and the absence The grid system is composed of one, two, three, or even four
of the ”arcing” instability, thus absence of arc-generated par- grids. Typically, it is a two- or three-grid system. The first grid
ticulates. The independent control of the ion energy and the (the screen grid) or an auxiliary anode is located inside the dis-
ion flux during the deposition process can be used to modify charge chamber. The ions are extracted by applying a negative
the film grain size, nucleation density, defects, crystal structure potential Vaccel to the second grid (accelerator grid). The abso-
lattice spacing, preferred crystallite orientation, density, and lute value Vaccel is typically a few hundred volts. Often a third
stress. The ion beams applied for thin film materials process- grid (decelerator grid) is added and kept at ground potential. Its
ing are operated at low background pressures, typically below role is to prevent electrons of the space-charge-neutralized ion
10−2 Pa, so that both the incident ion beam and the sputtered beam to be accelerated ‘backwards’ into the ion source; such
atoms are a line-of-sight processes. The pressure is kept low electron current represents an unwanted load on the extrac-
enough that collisions with the background gas are rare, and tor power supply and causes x-rays, which would then require
most of the sputtered atoms pass right through. safety shielding. The resulting ion energy can be in the range
An ion beam sputtering setup is shown schematically in from 200 to 2000 eV, the ion current density up to a few mA
figure 26. It consists of a broad-beam ion source, a target cm−2, and the total beam power up to several hundred watts.
holder, and a substrate holder. The distance between the exit Ion beam deposition has much lower deposition rate than
plane of the ion beam source and the target center and between thermal evaporation or magnetron sputter deposition, as seen
the target center and the substrates is in the range 10–20 cm. in figure 1(b), which is a significant drawback. Furthermore,
The ion beam source and target holder are typically placed this technique is not easy to upscale, and in general the ion
on rotary tables, which have their center of rotation located at beam sources are more complex than magnetron sputter depo-
the center of the target surface plane. This means that the ion sition systems or evaporation sources. Therefore, ion beam
incidence angle α, and consequently the polar emission angle deposition is only applied in niche applications, such as laser
24
mirror coatings based on dielectric multilayers (Gibson 1987, be operated, and more than one reactive gas can be added. It
Stolz and Génin 2003), where the relatively high cost is jus- is thus possible to readily produce compound films containing
tified by the superior quality of the resulting films that can be multiple elements, for example, TiAlON when titanium and
achieved (Bundesmann and Neumann 2018). aluminum are in the cathode material, and nitrogen and oxy-
For further discussion on ion beam deposition the inter- gen are added to the gas. The stoichiometry of the compound
ested reader is directed to the book chapter by McNeil et al can be tuned in a wide range depending on the ion flux ratios
(2002) and/or the recent tutorial by Bundesmann and Neu- arriving at the substrate.
mann (2018).
4.1. Plasma from cathode spots
4. Cathodic arcs
The nonlinear nature of electron emission laws in terms of
An arc is a discharge of relatively high current (several cathode temperature and electric field leads to spot forma-
amperes, often tens or hundreds of amperes, or even higher) at tion: for the arc discharge to occur, it is energetically pre-
relatively low voltage between electrodes (generally less than ferred to concentrate the discharge power to a very small area
40 V). The operation mechanism of the arc discharge differs on the cathode to obtain the greatest emission of electrons.
significantly from glow discharges used for sputtering. It is The current in a vacuum arc discharge is typically concen-
characterized by a collective electron emission mechanism. If trated in a luminous area of a few square micrometers on the
the cathode is sufficiently hot that the collective mechanism cathode surface and referred to as cathode spot (Anders et al
is thermionic emission, one calls this arc a ‘thermionic arc’. 1996, Jüttner 2001). As a ‘side effect’, the cathode material at
In the context of PVD, a different arc mode, the cathodic arc, these concentration points, the cathode spot, is readily melted,
is much more relevant. Here, the emission of electrons occurs evaporated and ionized. The development of such hot spots
at small (typically micrometer), non-stationary cathode spots is extremely fast (nanoseconds) so that models of wire explo-
based on thermo-field emission: a nonlinear combination of sions have been successfully applied to describe the process
thermionic and field emission (Jüttner 2001). The power den- (Mesyats and Proskurovsky 1989). An example of erosion
sity of the cathode spots is extremely high (estimated between traces left on a cadmium cathode target after arc deposition
1012 and 1014 W m−2 ) which leads to evaporation and ion- due to ejection of fast plasma jets and liquid metal droplets is
ization of the cathode material, and thus to the formation of shown in figure 27. The cathode processes lead to rapid phase
plasma of the cathode material. transitions and are therefore sometimes referred to as explo-
The name cathodic arc is due to the fact that the current- sive electron emission (Mesyats 1998). The current density
carrying discharge medium between electrodes is the plasma has been determined to reach peak values of 1012 A m−2 (Jüt-
of the cathode material even when gas is present between the tner et al 1984) while the plasma density can temporarily reach
electrodes (Anders 2008). A cathodic arc may also burn in vac- peak values exceeding 1026 m−3 (Anders et al 1992). At such
uum, i.e., without any gas present, and in this case, the arc is high densities, the plasma is dense and ‘nonideal’ which, by
known as a vacuum arc (Boxman et al 1995). Vacuum arcs definition, means that the potential energy between charged
are cathodic arcs, but cathodic arcs are not always vacuum particles is not small compared to their kinetic energy. One
arcs since a process gas (or liquid) can be present between the of the consequences is a reduction of the ionization energy
electrodes. leading to an enhancement of ion charge states (Ebeling and
Historically, the name metal vapor vacuum arc was occa- Kilimann 1989).
sionally used, especially in the context of vacuum arc ion Arcs in air were studied in the 18th century (Anders 2003,
sources (Brown 1994). This name is less suitable since it Priestley 1775), while more ‘recent’ studies of their optical
(falsely) suggests the emission of neutral metal vapor from the spectra, even in the vacuum ultraviolet spectral region, date
cathode. In fact, the flux expanding from the cathode spots con- back ‘only’ a hundred years (McLennan et al 1919). The prop-
sists of a fully ionized metal plasma with multiply charged ions erties of the plasma flux from cathode spots have been exten-
(Davis and Miller 1969), a property that will be considered in sively studied for decades (see e.g., Boxman et al (1995),
greater detail below. Davis and Miller (1969), Lafferty (1980), Plyutto et al (1965),
Cathodic arcs are prolific sources of plasma of any cathode Tanberg (1930)). Most remarkably, the plasma contains multi-
material. However, in order to serve as a cathode, the mate- ply charged ions (most often 2+ and 3+ ) that have supersonic
rial has to be sufficiently conductive to carry the arc current. speed (relative to the ion sound speed), with Mach numbers
Pure metals, alloys and sintered metal composites, graphite between 3 and 5 (Anders and Yushkov 2002). The most likely
and highly doped or heated semiconductors have been used ion velocity is typically 1–3 × 104 m s−1 . Ion charge state
as cathode. For example, a titanium cathode delivers titanium and ion velocity scale with the cohesive energy of the cathode
plasma. If the operation is in vacuum or in a noble gas like (cohesive energy rule, see Anders (2008) (chapter 3)) which is
argon, the titanium plasma condenses on any surface, thereby based on energy considerations of the phase transitions and
forming a titanium coating. In the context of PVD, one is often ionization processes. Also, the discharge voltage is linearly
interested in compound coatings, such as TiN. In this case, dependent on the cohesive energy Ece of the cathode material
nitrogen is added to the discharge region. To illustrate the ver- (Anders 2002b, Anders and Yushkov 2002).
satility of the cathodic arc deposition process: more than one The acceleration of positively charged ions away from
metal can be used in the cathode, or more than one cathode can the (negative) cathode is non-trivial. A combined mechanism
25
Figure 27. Erosion traces left by a 10 A vacuum arc on a cadmium

cathode. Reproduced from Daalder (1979). © IOP Publishing Ltd. Figure 28. Streak camera image of cathode spots on an aluminum
All rights reserved. cathode in 30 mPa of nitrogen (light intensity in false colors). The
spot appears to move, on average, from left to right in irregular
steps. While the on-average direction is affected by the presence of a
based on strong pressure gradients and the local electric field of magnetic field, individual plasma-producing microscopic explosions
expanding plasma is responsible. One should keep in mind that are random and can occur opposite to the macroscopic apparent
even for dc (continuous) cathodic arcs, the cathode processes motion of the spot. Reprinted from Oh et al (2021b), with the
permission of AIP Publishing.
are essentially pulsed due to the non-stationary, even explosive
processes. A good picture is to consider the plasma flux as the
result of a rapid sequence of microexplosions, thus the plasma scale heating (Anders 2002a). The role of kinetic and potential
flux is pulsating or fluctuating in flux density, ion charge state energies has also been investigated by atomistic simulations of
and velocity (Anders et al 2005). Detailed analysis showed energetic film growth (Kateb et al 2021, 2020). The condens-
that the fluctuations do not show characteristic frequencies able nature of the metal plasma implies that the plasma is not
but a broad spectrum. This applies to vacuum arcs (Oh et al just assisting but the plasma itself is deposited. The sticking
2021a, Smeets and Schulpen 1988) and also to cathodic arcs coefficient of metal ions is close to unity, however, as stud-
in the presence of reactive gases (Oh et al 2021b). Figure 28 ies have shown, the small fraction of non-sticking ions, likely
illustrates the non-stationary, fluctuating character of cathode associated with atomic scale heating, represent a flux of neutral
spot activity and related cathode plasma production. In fact, atoms leaving the substrate, which has a notable effect on the
many of the fluctuating properties (voltage, current, ion cur- incoming ion flux due to charge exchange collisions (Anders
rent, charge states, etc) follow a power law with respect to et al 2007). Charge exchange collisions lead to a reduction
frequency, indicative of their fractal character. Simply said: of the contribution of higher charge states and a correspond-
cathode spots, and thus cathodic arcs as a whole, show fractal ing increase of the flux of singly charged ions. The boundary
(self-similar) properties (Anders 2005). between any plasma and a surface, the sheath, is a space charge
layer associated with a voltage drop. The region between a sub-
4.2. Plasma deposition strate and a cathodic arc plasma also presents a sheath. Even
Coming back to the statements that ions are supersonic and in the absence of an intentional substrate bias, or in the case of
often multiply charged: this implies that the plasma flux a floating substrate, a sheath will form. Even a modest sheath
impinging on a substrate delivers substantially more kinetic voltage has a substantial effect on the energy of the arriving
and potential energies to a growing film than with other pro- ions:
cesses such as deposition by evaporation or sputtering. In the Ei = Ei0 + QeVsheath , (23)
case of cathodic arc deposition, the plasma itself condenses
and therefore we deal with plasma deposition. The kinetic where Ei0 is the energy an ion had already gained in the cath-
energy delivered to the growing film can be several 10 eV for ode spot region before acceleration in the substrate’s sheath,
each ion, even of order of 100 eV per ion for refractory met- Q is the charge state number (often 2 or 3 for cathodic arcs), e
als, and typically more than 10 eV of potential energy for each is the elementary charge, and Vsheath = Vplasma − Vsurface is the
incoming ion (Anders and Yushkov 2002). This is a very large substrate sheath voltage, i.e. the difference between the plasma
amount of energy deposited locally, sometimes called atomic potential and the substrate surface potential.
26
4.3. Substrate bias effects
With intentional (negative) bias, the substrate surface poten-

tial is lowered and the sheath voltage thereby increased. For
cathodic arcs, bias is generally more effective than with other
techniques (such as sputtering) due to the high degree of ion-
ization and the presence of multiply charged ions. At moderate
bias of a few 10 V, the deposition process can change its char-
acter since the ion energy leads to insertion of the ion under
the surface. This therefore results in a subplanation growth
process, as opposed to classic film growth processes which
assume adding atoms on the surface. Subplantation leads gen-
erally to densification of the film, but also to defect formation Figure 29. Scanning electron microscope (SEM) image of a
and enhancement of compressive stress (Bilek and McKen- cross-section of a film deposited from a Ti0.3 Al0.7 cathode of
zie 2006). The latter can be excessive as obvious when the unfiltered dc arc source at a N2 pressure of 0.67 Pa. Reprinted from
Zhirkov et al (2015), Copyright (2015), with permission from
deposited film catastrophically break away from the substrate. Elsevier.
Indeed, it is generally difficult to grow thick films using a
subplanation process. For thin to moderately thick films (say properties. Since the size of a macroparticle can be microme-
100 nm or less), subplanation is the preferred method to pro- ters in diameter, i.e., larger than the wavelength of visible light,
duce tetrahedral amorphous carbon (ta-C) films, a hydrogen- the surface can appear hazy or dull to the eye: upon deposi-
free from of diamondlike carbon (DLC) films (Abadias et al tion, the substrate may lose its mirror-like shininess if it was
2018, Pharr et al 1996, Uhlmann et al 1998). At even higher mirror-like before the coating. Macroparticles are the greatest
applied bias, typically of several 100 V, the incoming ions impediment for the even wider use of cathodic arc deposition
cause substantial sputtering from the growing film. It is pos- since the requirements on the quality of coatings is expected to
sible to increase the bias to a point where more material is be excellent. Figure 29 illustrates an example when macropar-
removed than deposited: in this case the process becomes ion ticles are incorporated into a coating thereby greatly affecting
etching. Combining ion etching and deposition in a two-step (usually in a deleterious way) the properties of the coating
process was the underlying idea of ‘arc bond sputtering’ or (Zhirkov et al 2015).
ABS, i.e., a process starting with high bias etching followed by For some applications, the incorporation of macroparticles
deposition at lower bias (Münz et al 1991). The second (depo- can be tolerated since other coating qualities are decisive such
sition) step was preferably done by sputtering to reduce the as hardness, toughness, and of course, low cost. Also, the size
‘macroparticle issue’, a quality reduction caused by the deposi- of the macroparticles can be reduced by a number of measures:
tion of microscopic cathode droplets or ‘macroparticles’. This among them is the effect of the reactive gas, the use of ‘spot
issue will require a separate consideration. steering’ and, last but not least, macroparticle filtering. Each
of those effects is briefly discussed below.
4.4. Macroparticles 4.5. Effect of the reactive gas
The high power density at cathode spots implies that there Similar to magnetron sputtering, the cathode of the arc pro-
are always regions between the spots and the generally much cess reacts with the reactive gas (such as nitrogen or oxygen)
colder (near room temperature) cathode body. These micro- when added to the process. The cathode processes are sensi-
scopic regions of melted cathode material are subject to the tive to the chemical state of the cathode (Jakubka and Jüttner
pressure of the spot plasma: the liquid is pushed from its 1981, Oh et al 2021b), which is already clear when considering
location and travels from the cathode as microdroplets or the change in work function (potential barrier) for the electron
macroparticles. The name ‘macroparticle’ has been used in this emission processes. When the cathode is covered with a thin
context for decades (see e.g. Plyutto et al (1965)) to express the compound layer (‘poisoned cathode’), the ignition of an emis-
fact that those particles are orders of magnitude heavier than sion site is often faster (easier), which may be associated with
plasma particles like electrons and ions. the changed potential structure and/or charge-up in case of an
Macroparticles have a velocity of typically a few 10 m s−1 insulating compound. As a result, more cathode spots cover the
to some 100 m s−1 , they are much slower than the plasma flow cathode but each of shorter duration. The current carried by a
(Schülke and Anders 1999). They cool down during flight and spot can be reduced from 10 s of amperes to less than 1 ampere.
arrive on the substrate in the liquid or solid phase, depending The volume of melted cathode material is much reduced com-
on the material. Low melting point metals form flat ‘pancake’ pared to the pure metal case, and, indeed, one finds smaller
particles on the surface upon impact (Daalder 1976). Refrac- macroparticles. Moreover, the macroparticles emitted from the
tory materials, in contrast, are often solidified before impact. cathode also react with the reactive gas, forming a compound,
They may or may not stick to the substrate: many ‘bounce off’ which generally has higher melting point than the cathode
the surface but some stick and become part of the coating. metal. When such compound particle arrives at the substrate it
Macroparticle incorporation into the coatings leads to much is very likely solid and tend (with a high probability) to bounce
enhanced roughness and the deterioration of many other film off and is thus not incorporated in the coating (Karpov 1997).
27
Cathode poisoning is therefore one of the strategies to reduce

macroparticles when depositing compounds.
4.6. Steered arc sources
Another strategy to reduce macroparticles is to make use of the

‘spot steering’ effect on the cathode. Namely, the ignition of
cathode spots is affected by the presence of plasma from pre-
vious ignitions. When a transverse magnetic field is applied to
the cathode (magnetic vector component parallel to the cath-
ode surface), the spot ignition probability is not isotropic but
there is a preferred direction due to the magnetic field effect on
the plasma of previous spots (Robson and von Engel 1956).
This can be used to ‘steer’ the arc in a preferred direction.
The path can be closed–similar to an erosion ‘racetrack’ that
is formed in the target surface of a magnetron sputtering dis-
charge. The ‘speed’ of spot motion can be enhanced when the Figure 30. Example of an open-coil 90-degree filter, here to filter
plasma from an aluminum-doped zinc cathode in the presence of
magnetic field is strong. As a side effect, the spot emission time
oxygen to deposit high-quality AZO (aluminum-doped zinc oxide).
is reduced, leading to less cathode melting and less macropar- Reprinted from Anders, (2014), Copyright (2014), with permission
ticles. Of course, the steering effect can be combined with the from Elsevier.
strategy of cathode poisoning. However, even in the best cases,
the coatings still contain macroparticles.
arc discharge is a prolific, efficient and economical generator
4.7. Macroparticle filtering of plasma of the cathode material. As a low voltage, high cur-
rent process, arcs produce a high flux of condensable and ion-
For applications, where macroparticles are highly detrimental, ized material and deliver it onto a substrate with high energy
like in optical coatings applications, the previous measures are efficiency. High deposition rate, low cost as well as being capa-
not sufficient, and one can apply a filtering process that sep- ble of delivering desirable film properties made cathodic arc
arates plasma (ions, electrons) from macroparticles based on deposition a preferred choice for some coatings applications
the fact that macroparticles are so much heavier than plasma over the last decades. The very high degree of ionization of
particles. The plasma is guided to a substrate that is not in line- the flux allows producers of coatings to effectively use bias,
of-sight with the cathode (Aksenov et al 1978). Guiding the thereby controlling the energy of deposited particles, which
plasma is generally accomplished by using a curved magnetic in turn affects the microstructure and related properties (see
field that connects the region of plasma generation (cathode discussion in section 1 and around figure 3). Arc deposition is
surface where spots burn) with the substrate where the coat- therefore both an economical and flexible approach to produce
ings should be deposited. Various geometries of macroparticle coatings in a wide range of composition and microstructure.
filters have been invented and tested (Anders 1999). One such This is especially established for hard, tough, and oxidation-
macroparticle filter is shown in figure 30. The filters are based resistant tool coatings like TiAlN (see e.g. Hans et al (2019)
on the general principle that electrons are easily magnetized and/or Schultrich (2022)) and diamondlike carbon coatings
and, on average, effectively follow magnetic field lines. The (Vetter 2014).
motion of ions, in turn, is affected by the motion of electrons:
the plasma tends to stay quasi-neutral due to strong electric
5. Conclusion
fields that immediately arise when electrons and ions sepa-
rate. Plasma transport in macroparticle filters is thus facili-
The different PVD techniques are discussed and shown that
tated by a combined magnetic and electric field mechanism.
they provide depositing fluxes that vary in absolute value, in
While macroparticle filters are effective, there is also a sub-
ionization degree, and in species composition and energy. This
stantial plasma loss in the transport, which implies a loss
includes deposition by thermal evaporation, electron beam
of deposition rate (Aksenov et al 2003). The largest advan-
evaporation, magnetron sputtering, ion beam sputtering, laser
tage of cathodic arc plasma deposition, its high deposition
ablation, and cathodic arc deposition (while this list contains
rate, is therefore diminished. Filtered arc deposition has there-
the most important methods, no claim of completeness is made
fore only be applied in few applications. The vast majority of
since a variety of other techniques exist, including hybrid tech-
industrial cathodic arc deposition systems limit themselves to
niques of the above). The variations in the deposition flux
utilizing the cathode poisoning and spot steering approaches.
have influence on the film microstructure and morphology
and therefore on the film structural properties, composition,
4.8. Using the advantages of cathodic arc plasma for PVD
mass density, optical properties, the stress in the film, electri-
After elaborating at length about the macroparticle issue and cal resistivity and surface roughness. In fact, PVD stands for a
macroparticle mitigation strategies such as macroparticle fil- collection of versatile deposition techniques, that have a vari-
tering, we ought to also address why cathodic arc deposition is ety of applications. Most importantly, these techniques are the
a widely used PVD technique. Most importantly, the cathodic basis for many approaches applied in material science, where
28
the synthesis of a material with superior quality or of films Anders A 2002b Cohesive energy rule for vacuum arcs Emerging
from a metastable compound can only be realised by PVD. Applications of Vacuum-Arc-Produced Plasma, Ion and Elec-
tron Beams (NATO Science Series II: Mathematics, Physics and
Chemistry vol 88) ed E Oks and I Brown (Dordrecht: Kluwer)
pp 1–14
Acknowledgments Anders A and Yang Y 2017 Appl. Phys. Lett. 111 064103
Anders A and Yushkov G Y 2002 J. Appl. Phys. 91 4824–32
This work was partially supported by the Icelandic Research Anderson G S, Mayer W N and Wehner G K 1962 J. Appl. Phys. 33
2991–2
Fund Grant No. 196141, the German Science Foundation Arthur J R 2002 Surf. Sci. 500 189–217
under the umbrella of the SFB TR 87, and the Leibniz Associ- Barnes M S, Forster J C and Keller J H 1993 Apparatus for depositing
ation via the Collaborative Excellence Project K128/2018. material into high aspect ratio holes US Patent 5,178,739 (New
York: International Business Machines Corporation, Armonk)
Bean J C 1993 Techniques for the growth of crystalline films by
Data availability statement molecular beam deposition Multicomponent and Multilayered
Thin Films for Advanced Microtechnologies: Techniques, Fun-
damentals and Devices (NATO ASI Series E: Applied Sciences
The data that support the findings of this study are available
vol 234) ed O Auciello and J Engemann (Dordrecht: Kluwer) pp
upon reasonable request from the authors. 87–107
Berry L A and Gorbatkin S M 1995 J. Vac. Sci. Technol. A 13
343–8
ORCID iDs Biersack J P and Haggmark L G 1980 Nucl. Instrum. Methods 174
257–69
Bilek M M M and McKenzie D R 2006 Surf. Coat. Technol. 200
Jon Tomas Gudmundsson https://orcid.org/0000-0002- 4345–54
8153-3209 Biskup B, Maszl C, Breilmann W, Held J, Böke M, Benedikt J and
André Anders https://orcid.org/0000-0002-5313-6505 von Keudell A 2018 J. Phys. D: Appl. Phys. 51 115201
Achim von Keudell https://orcid.org/0000-0003-3887- Blondeel A, Delrue H, Matthews S, Van Holsbeke J and De Bosscher
9359 W 2006 Advantages of using rotating cylindrical ITO targets
for large area coating applications Proc. 49th Annual Technical
Conf. Proc. (Washington DC 22–27 April 2006) (Albuquerque,
New Mexico: Society of Vacuum Coaters) pp 187–91
References Bobzin K, Brögelmann T, Kruppe N C, Engels M, von Keudell A,
Hecimovic A, Ludwig A, Grochla D and Banko L 2017 J. Appl.
Phys. 122 015302
Abadias G et al 2018 J. Vac. Sci. Technol. A 36 020801 Boeuf J P and Takahashi M 2020a Phys. Plasmas 27 083520
Affinito J and Parsons R R 1984 J. Vac. Sci. Technol. A 2 1275–84 Boeuf J-P 2014 Front. Phys. 2 74
Aisenberg S and Chabot R W 1973 J. Vac. Sci. Technol. A 10 104–7 Boeuf J-P and Takahashi M 2020b Phys. Rev. Lett. 124 185005
Aksenov I I, Belous V A, Padalka V G and Khoroshikh V M 1978 Boxman R L, Sanders D M and Martin P J (ed) 1995 Handbook
Instrum. Exp. Tech. 21 1416–8 of Vacuum Arc Science and Technology: Fundamentals and
Aksenov I I, Strel’nitskij V E, Vasilyev V V and Zaleskij D Y 2003 Applications (Park Ridge, New Jersey: Noyes Publications)
Surf. Coat. Technol. 163–164 118–27 Bradley J W, Thompson S and Gonzalvo Y A 2001 Plasma Sources
Amin N and Rahman K S 2017 Close-spaced sublimation (CSS): Sci. Technol. 10 490–501
a low-cost, high-yield deposition system for cadmium telluride Brenning N, Butler A, Hajihoseini H, Rudolph M, Raadu M A,
(CdTe) thin film solar cells Modern Technologies for Creating Gudmundsson J T, Minea T and Lundin D 2020 J. Vac. Sci.
the Thin-Film Systems and Coatings ed N Nikitenkov (London: Technol. A 38 033008
InTechOpen) ch 18 pp 361–79 Brenning N, Gudmundsson J T, Lundin D, Minea T, Raadu M A and
Anders A 1999 Surf. Coat. Technol. 120–121 319–30 Helmersson U 2016 Plasma Sources Sci. Technol. 25 065024
Anders A 2002a Appl. Phys. Lett. 80 1100–2 Brenning N, Hajihoseini H, Rudolph M, Raadu M A, Gudmundsson J
Anders A 2003 IEEE Trans. Plasma Sci. 31 1052–9 T, Minea T M and Lundin D 2021 Plasma Sources Sci. Technol.
Anders A 2005 IEEE Trans. Plasma Sci. 33 1456–64 30 015015
Anders A 2006 Thin Solid Films 502 22–8 Brenning N, Lundin D, Minea T, Costin C and Vitelaru C 2013 J.
Anders A 2008 Cathodic Arcs: From Fractal Spots to Energetic Phys. D: Appl. Phys. 46 084005
Condensation (Springer Series on Atomic, Optical, and Plasma Brown I G 1994 Rev. Sci. Instrum. 65 3061–81
Physics vol 50) (New York: Springer) Brückner J, Teschner G, Milde F and Krause J 2005 Advanced rotat-
Anders A 2010 Thin Solid Films 518 4087–90 able magnetron module designed for large area glass coaters
Anders A 2012 Appl. Phys. Lett. 100 224104 Proc. 48th Society of Vacuum Coaters Annual Technical Conf.
Anders A 2017 J. Appl. Phys. 121 171101 (Denver, Colorado 23–28 April 2005) (Albuquerque, New Mex-
Anders A, Anders S, Jüttner B, Botticher W, Luck H and Schroder G ico: Society of Vacuum Coaters) pp 265–9
1992 IEEE Trans. Plasma Sci. 20 466–72 Bultinck E and Bogaerts A 2009 New J. Phys. 11 103010
Anders A, Anders S, Jüttner B and Luck H 1996 IEEE Trans. Plasma Bundesmann C and Neumann H 2018 J. Appl. Phys. 124 231102
Sci. 24 69–70 Byon E and Anders A 2003 J. Appl. Phys. 93 1899–906
Anders A, Fukuda K and Yushkov G Y 2005 J. Phys. D: Appl. Phys. Chabert P, Tsankov T V and Czarnetzki U 2021 Plasma Sources Sci.
38 1021–8 Technol. 30 024001
Anders A, Oks E M and Yushkov G Y 2007 J. Appl. Phys. 102 043303 Chapin J S 1974 Res. Dev. 25(1) 37–40
29
Chapin J S 1979 Sputtering process and apparatus US Patent Gudmundsson J T 2020 Plasma Sources Sci. Technol. 29 113001
4166018A (Montvale, New Jersey, United States of America: Gudmundsson J T, Brenning N, Lundin D and Helmersson U 2012 J.
Airco, Inc.) Vac. Sci. Technol. A 30 030801
Cheung J T and Sankur H 1988 Crit. Rev. Solid State Mater. Sci. 15 Gudmundsson J T and Hecimovic A 2017 Plasma Sources Sci.
63–109 Technol. 26 123001
Cheung J and Horwitz J 1992 MRS Bull. 17 30–6 Gudmundsson J T and Lundin D 2020 Introduction to magnetron
Chistyakov R and Abraham B 2009 Advanced pulsed DC technology sputtering High Power Impulse Magnetron Sputtering: Fun-
for material processing applications: II. Plasma generator for damentals, Technologies, Challenges and Applications ed D
arbitrary voltage pulse shape SVC Bulletin (Summer) pp 32–6 Lundin, T Minea and J T Gudmundsson (Amsterdam, The
Cho A Y 1971 J. Vac. Sci. Technol. 8 S31–8 Netherlands: Elsevier) pp 1–48
Cho A Y 1995 MRS Bull. 20 21–8 Hajihoseini H, Čada M, Hubička Z, Ünaldi S, Raadu M A,
Christie D J 2005 J. Vac. Sci. Technol. A 23 330–5 Brenning N, Gudmundsson J T and Lundin D 2019 Plasma 2
Daalder J E 1976 J. Phys. D: Appl. Phys. 9 2379–95 201–21
Daalder J E 1979 J. Phys. D: Appl. Phys. 12 1769–79 Hajihoseini H and Gudmundsson J T 2017 J. Phys. D: Appl. Phys. 50
Davidse P D and Maissel L I 1966 J. Appl. Phys. 37 574–9 505302
Davis W D and Miller H C 1969 J. Appl. Phys. 40 2212–21 Hála M 2011 Characterization of high power impulse magnetron
Dijkkamp D, Venkatesan T, Wu X D, Shaheen S A, Jisrawi N, sputtering discharges PhD Thesis École Polytechnique de
Min-Lee Y H, McLean W L and Croft M 1987 Appl. Phys. Lett. Montréal, Montréal, Canada
51 619–21 Hála M, Čapek J, Zabeida O, Klemberg-Sapieha J E and Martinu L
Eason R 2007 Pulsed Laser Deposition of Thin Films: Applications- 2012 Surf. Coat. Technol. 206 4186–93
Led Growth of Functional Materials (New York: Wiley) Hans M et al 2019 Coatings 9 24
Ebeling W and Kilimann K 1989 Z. Naturforsch. A 44 519–23 Harper J M E 1990 Particle bombardment effects in thin film depo-
Eckstein W 2007 Sputtering yields Sputtering by Particle Bombard- sition Plasma-Surface Interactions and Processing of Materials
ment: Experiments and Computer Calculations from Thresh- (NATO ASI Series E: Applied Sciences vol 176) ed O Auciello,
old to MeV Energies (Topics in Applied Physics vol 110) ed R A Gras-Marti, J A Valles-Abarca and D L Flamm (Dordrecht:
Behrisch and W Eckstein (Berlin: Springer) pp 33–187 Kluwer) pp 251–80
Ehiasarian A P, Hecimovic A, de los Arcos T, New R, Harper J M E 1978 Ion beam deposition Thin Film Processes ed J L
Schulz-von der Gathen V, Böke M and Winter J 2012 Vossen and J J Cuomo (New York: Academic) pp 175–206
Appl. Phys. Lett. 100 114101 Heister U, Krempel-Hesse J, Szczyrbowski J, Teschner G, Bruch J
Este G and Westwood W D 1988 J. Vac. Sci. Technol. A 6 1845–8 and Bräuer G 2000 Vacuum 59 424–30
Ferreira F, Oliveira J C and Cavaleiro A 2016 Surf. Coat. Technol. Held J, Hecimovic A, von Keudell A and Schulz-von der Gathen V
291 365–75 2018 Plasma Sources Sci. Technol. 27 105012
Ferreira F, Serra R, Oliveira J C and Cavaleiro A 2014 Surf. Coat. Held J, Maaß P A, Schulz-von der Gathen V and von Keudell A 2020
Technol. 258 249–56 Plasma Sources Sci. Technol. 29 025006
Franz R, Clavero C, Kolbeck J and Anders A 2016 Plasma Sources Helmersson U, Lattemann M, Bohlmark J, Ehiasarian A P and
Sci. Technol. 25 015022 Gudmundsson J T 2006 Thin Solid Films 513 1–24
Fruth H F 1932 Physics 2 280–6 Herman M A and Sitter H 1996 Molecular Beam Epitaxy: Fundamen-
Fujioka H 2015 Pulsed laser deposition (PLD) Thin Films and Epi- tals and Current Status (Springer Series in Materials Science
taxy: Basic Techniques (Part A of Handbook of Crystal Growth vol 7) 2nd edn (Berlin: Springer)
vol 3) 2nd edn ed T Kuech (Amsterdam, The Netherlands: Hofer W O 1991 Angular, energy, and mass distribution of sputtered
Elsevier) ch 4 pp 365–97 particles Sputtering by Particle Bombardment III: Character-
Gall D 2005 Vapor transport processing: mechanisms Encyclopedia istics of Sputtered Particles, Technical Applications (Topics in
of Condensed Matter Physics ed F Bassani, G L Liedl and P Applied Physics vol 64) ed R Behrisch and K Wittmaack (Berlin:
Wyder (New York: Academic) pp 291–6 Springer) pp 15–90
Gibson U J 1987 Ion-beam processing of optical thin films Advances Holber W M 2000 Ionization by microwave electron cyclotron res-
in Research and Development (Physics of Thin Films vol 13) onance plasma Ionized Physical Vapor Deposition (Thin Films
ed M H Francombe and J L Vossen (New York: Academic) pp vol 27) ed J A Hopwood (San Diego, California: Academic) pp
109–50 67–94
Gill W D and Kay E 1965 Rev. Sci. Instrum. 36 277–82 Holber W M, Logan J S, Grabarz H J, Yeh J T C, Caughman J B
Gnaser H 2007 Energy and angular distributions of sputtered species O, Sugerman A and Turene F E 1993 J. Vac. Sci. Technol. A 11
Sputtering by Particle Bombardment: Experiments and Com- 2903–10
puter Calculations from Threshold to MeV Energies (Topics in Hopwood J A 2000 The role of ionized physical vapor deposition in
Applied Physics vol 110) ed R Behrisch and W Eckstein (Berlin: integrated circuit fabrication Ionized Physical Vapor Deposition
Springer) ch 5 pp 231–328 (Thin Films vol 27) ed J A Hopwood (San Diego, California:
Goldberg O, Goldenberg E, Zhitomirsky V N, Cohen S R and Academic) pp 1–7
Boxman R L 2012 Surf. Coat. Technol. 206 4417–24 Hopwood J 1992 Plasma Sources Sci. Technol. 1 109–16
Gorbatkin S M, Poker D B, Rhoades R L, Doughty C, Berry L A and Hopwood J 1998 Phys. Plasmas 5 1624–31
Rossnagel S M 1996 J. Vac. Sci. Technol. B 14 1853–9 Hopwood J, Guarnieri C R, Whitehair S J and Cuomo J J 1993 J. Vac.
Graper E B 2017 Electron beam evaportation Handbook of Thin Film Sci. Technol. A 11 152–6
Process Technology ed D A Glocker and S I Shah (Boca Raton, Hubička Z, Gudmundsson J T, Larsson P and Lundin D 2020 Hard-
Florida: CRC Press) p A1.2 ware and power management for high power impulse mag-
Greczynski G, Petrov I, Greene J E and Hultman L 2019 J. Vac. Sci. netron sputtering High Power Impulse Magnetron Sputtering:
Technol. A 37 060801 Fundamentals, Technologies, Challenges and Applications ed
Green K M, Hayden D B, Juliano D R and Ruzic D N 1997 Rev. Sci. D Lundin, T Minea and J T Gudmundsson (Amsterdam, The
Instrum. 68 4555–60 Netherlands: Elsevier) pp 49–80
Greene J E and Barnett S A 1982 J. Vac. Sci. Technol. 21 285–302 Huo C, Lundin D, Gudmundsson J T, Raadu M A, Bradley J W and
Gudmundsson J T 2008 J. Phys.: Conf. Ser. 100 082002 Brenning N 2017 J. Phys. D: Appl. Phys. 50 354003
30
Huo C, Lundin D, Raadu M A, Anders A, Gudmundsson J T and Westwood and J J Cuomo (Park Ridge, NJew Jersey: Noyes
Brenning N 2013 Plasma Sources Sci. Technol. 22 045005 Publications) pp 140–59
Hussein A E, Diwakar P K, Harilal S S and Hassanein A 2013 J. Appl. Lou B-S, Chen W-T, Diyatmika W, Lu J-H, Chang C-T, Chen P-W
Phys. 113 143305 and Lee J-W 2021 Surf. Coat. Technol. 421 127430
Jakubka K and Jüttner B 1981 J. Nucl. Mater. 102 259–66 Lundin D, Vitelaru C, de Poucques L, Brenning N and Minea T 2013
Janes G S and Lowder R S 1966 Phys. Fluids 9 1115–23 J. Phys. D: Appl. Phys. 46 175201
Jüttner B 2001 J. Phys. D: Appl. Phys. 34 R103–23 Mahan J E 2000 Physical Vapor Deposition of Thin Films (New York:
Jüttner B, Pursch H and Anders S 1984 J. Phys. D: Appl. Phys. 17 Wiley)
L111–4 Mahieu S, Van Aeken K and Depla D 2008 Transport of sputtered
Kadlec S, Musil J and Vyskočil H 1986 J. Phys. D: Appl. Phys. 19 particles through the gas phase Reactive Sputter Deposition ed
L187–90 D Depla and S Mahieu (Springer Series in Materials Science
Karpov D A 1997 Surf. Coat. Technol. 96 22–33 vol 109) ch 6 (Berlin: Springer) pp 199–227
Kateb M, Gudmundsson J T, Brault P, Manolescu A and Ingvarsson Martin L W, Chu Y-H and Ramesh R 2010 Mater. Sci. Eng. R 68
S 2021 Surf. Coat. Technol. 426 127726 89–133
Kateb M, Gudmundsson J T and Ingvarsson S 2020 J. Vac. Sci. Martin P 1990 IEEE Trans. Plasma Sci. 18 855–68
Technol. A 38 043006 Matthews A and Teer D G 1980 Thin Solid Films 72 541–9
Kateb M, Hajihoseini H, Gudmundsson J T and Ingvarsson S 2019 J. Mattox D M 1998 Handbook of Physical Vapor Deposition (PVD)
Vac. Sci. Technol. A 37 031306 Processing: Film Formation, Adhesion, Surface Preparation
Kaufman H R and Robinson R S 1989 Vacuum 39 1175–80 and Contamination Control (Westwood, New Jersey: Noyes
Kawamura E, Vahedi V, Lieberman M A and Birdsall C K 1999 Publications)
Plasma Sources Sci. Technol. 8 R45–64 McKelvey H E 1982 Magnetron cathode sputtering apparatus
Kay E 1963 J. Appl. Phys. 34 760–8 US Patent 4356073 (Detroit, Michigan: Shatterproof Glass
Kay E 1962 Impact evaporation (sputtering) and thin film growth in Corporation)
a glow discharge Advances of Electronics and Electron Physics McLennan J C, Ainslie D S and Fuller D S 1919 Proc. R. Soc. A 95
vol 17 ed L L Marton (New York: Academic) pp 245–322 316–32
Kelly P J and Arnell R D 1998 Surf. Coat. Technol. 108–109 317–22 McNeil J R, McNally J J and Reader P D 2002 Ion beam deposition
Kelly P J and Arnell R D 2000 Vacuum 56 159–72 Handbook of Thin-Film Deposition Processes and Techniques:
Kelly P 2011 Mater. Technol. 26 1–2 Principles, Methods, Equipment and Applications 2nd edn ed K
Keraudy J, Viloan R P B, Raadu M A, Brenning N, Lundin D and Seshan (Norwich, New York: Noyes Publications) pp 463–99
Helmersson U 2019 Surf. Coat. Technol. 359 433–7 Mesyats G A 1998 Explosive Electron Emission (Ekaterinburg: URO
Klawuhn E, D’Couto G C, Ashtiani K A, Rymer P, Biberger M A and Press)
Levy K B 2000 J. Vac. Sci. Technol. A 18 1546–9 Mesyats G A and Proskurovsky D I 1989 Pulsed Electrical Discharge
Knodle W S and Chow R 2002 Molecular beam epitaxy: equipment in Vacuum (Springer Series on Atomic, Optical, and Plasma
and practice Deposition Processes and Techniques: Principles, Physics vol 5) (Berlin: Springer)
Methods, Equipment and Applications 2nd edn ed K Seshan Metz G E 2012 Characterization of a plasma reactor device for photo-
(Norwich, New York: Noyes Publications) pp 381–461 voltaic applications Master’s Thesis Colorado State University,
Kolev I, Bogaerts A and Gijbels R 2005 Phys. Rev. E 72 056402 Fort Collins, Colorado
Kouznetsov V, Macák K, Schneider J M, Helmersson U and Petrov I Mishra A, Kelly P J and Bradley J W 2010 Plasma Sources Sci.
1999 Surf. Coat. Technol. 122 290–3 Technol. 19 045014
Kozyrev A V, Sochugov N S, Oskomov K V, Zakharov A N and Mitchell K, Fahrenbruch A L and Bube R H 1975 J. Vac. Sci. Technol.
Odivanova A N 2011 Plasma Phys. Rep. 37 621–7 12 909–11
Krüger D, Köhn K, Gallian S and Brinkmann R P 2018 Phys. Plasmas Möller W and Eckstein W 1984 Nucl. Instrum. Methods Phys. Res. B
25 061207 2 814–8
Kubart T, Gudmundsson J T and Lundin D 2020 Reactive high Möller W, Eckstein W and Biersack J P 1988 Comput. Phys. Com-
power impulse magnetron sputtering High Power Impulse Mag- mun. 51 355–68
netron Sputtering: Fundamentals, Technologies, Challenges and Monaghan D P, Teer D G, Laing K C, Efeoglu I and Arnell R D 1993
Applications ed D Lundin, T Minea and J T Gudmundsson Surf. Coat. Technol. 59 21–5
(Amsterdam, The Netherlands: Elsevier) pp 223–63 Morley J R and Smith H R 1972 J. Vac. Sci. Technol. 9 1377–8
Lafferty J M 1980 Vacuum Arcs: Theory and Application (New York: Münz W-D, Hauzer F J M, Schulze D and Buil B 1991 Surf. Coat.
Wiley) Technol. 49 161–7
Lai K F 2000 Ionized hollow cathode magnetron sputtering Ionized Murayama Y 1974 Japan. J. Appl. Phys. 13 459–62
Physical Vapor Deposition (Thin Films vol 27) ed J A Hopwood Musil J, Kadlec S and Münz W D 1991 J. Vac. Sci. Technol. A 9
(San Diego, California: Academic) pp 95–139 1171–7
Layes V 2021 Fundamental surface processes in HiPIMS plasmas: Nakano T, Hirukawa N, Saeki S and Baba S 2013 Vacuum 87 109–13
physics and chemistry at the plasma magnetron target interface Nakano T, Murata C and Baba S 2010 Vacuum 84 1368–71
PhD Thesis Ruhr-Universität Bochum, Germany Nakano T, Saitou Y and Oya K 2017 Surf. Coat. Technol. 326 436–42
Lieberman M A and Lichtenberg A J 2005 Principles of Plasma Dis- Nakano T, Umahashi T and Baba S 2014 Japan. J. Appl. Phys. 53
charges and Materials Processing 2nd edn (New York: Wiley) 028001
Liebig B, Braithwaite N S J, Kelly P J, Chistyakov R, Abraham B and Oh K, Kalanov D and Anders A 2021a Plasma Sources Sci. Technol.
Bradley J W 2011 Surf. Coat. Technol. 205 S312–6 30 095005
Lin J, Sproul W D, Moore J J, Wu Z, Lee S, Chistyakov R and Oh K, Kalanov D, Birtel P and Anders A 2021b J. Appl. Phys. 130
Abraham B 2011 JOM 63 48–58 183304
Logan J S 1990 RF sputter etching and deposition Handbook of Ojeda-G-P A, Döbeli M and Lippert T 2018 Adv. Mater. Interfaces 5
Plasma Processing Technology: Fundamentals, Etching, Depo- 1701062
sition, and Surface Engineering ed S M Rossnagel, W D Okamoto A and Serikawa T 1986 Thin Solid Films 137 143–51
31
Ono T, Takahashi C and Matsuo S 1984 Japan. J. Appl. Phys. 23 Schülke T and Anders A 1999 Plasma Sources Sci. Technol. 8 567–71
L534–6 Schultrich B 2022 Coatings 12 109
Panjan M and Anders A 2017 J. Appl. Phys. 121 063302 Sellers J 1998 Surf. Coat. Technol. 98 1245–50
Panjan M, Franz R and Anders A 2014 Plasma Sources Sci. Technol. Shen J, Gai Z and Kirschner J 2004 Surf. Sci. Rep. 52 163–218
23 025007 Shimizu T, Zanáška M, Villoan R P, Brenning N, Helmersson U and
Parker E H C (ed) 1985 The Technology and Physics of Molecular Lundin D 2021 Plasma Sources Sci. Technol. 30 045006
Beam Epitaxy (New York: Plenum) Sigmund P 1969 Phys. Rev. 184 383–416
Petrea M-A and Stamate E 2021 Plasma Sources Sci. Technol. 30 Smeets R P P and Schulpen F J H 1988 J. Phys. D: Appl. Phys. 21
045002 301–10
Petrov I, Barna P B, Hultman L and Greene J E 2003 J. Vac. Sci. Spitzer L 1956 Physics of Fully Ionized Gases (New York:
Technol. A 21 S117–28 Interscience)
Pharr G M et al 1996 Appl. Phys. Lett. 68 779–81 Sproul W D 1998 Vacuum 51 641–6
Plyutto A A, Ryzhkov V N and Kapin A T 1965 Sov. Phys - JETP 20 Sproul W D, Christie D J and Carter D C 2005 Thin Solid Films 491
328–37 http://jetp.ras.ru/cgi-bin/dn/e_020_02_0328.pdf 1–17
Poolcharuansin P, Liebig B and Bradley J W 2012 Plasma Sources Sproul W D, Rudnik P J, Graham M E and Rohde S L 1990 Surf.
Sci. Technol. 21 015001 Coat. Technol. 43–44 270–8
Preissing S 2016 Spectroscopic investigation of an extensive mag- Sproul W D and Tomashek J R 1984 Rapid rate reactive sputtering
netron plasma Master’s Thesis Ruhr Universität Bochum, Ger- of a group IVB metal US Patent 4,428,811 (Chicago, Illinois,
many United States of America: Borg-Warner Corporation)
Priestley J 1775 The History and Present State of Electricity 3rd edn Stamate E 2020 Surf. Coat. Technol. 402 126306
(London, England) Stelmack L A, Thurman C T and Thompson G R 1989 Nucl. Instrum.
Ptak A J 2015 Principles of molecular beam epitaxy Thin Films Methods Phys. Res. B 37–38 787–93
and Epitaxy: Basic Techniques (Part A of Handbook of Crys- Stepanova M and Dew S K 2004 Nucl. Instrum. Methods Phys. Res.
tal Growth vol 3) 2nd edn, ed T Kuech (Amsterdam: Elsevier) B 215 357–65
ch 4 pp 161–92 Stolz C J and Génin F Y 2003 Laser resistant coatings Optical Inter-
Rahman K S, Harif M N, Rosly H N, Kamaruzzaman M I B, ference Coatings (Springer Series in Optical Sciences vol 88) ed
Akhtaruzzaman M, Alghoul M, Misran H and Amin N 2019 N Kaiser and H K Pulker (Berlin: Springer) pp 309–33
Results Phys. 14 102371 Strijckmans K, Schelfhout R and Depla D 2018 J. Appl. Phys. 124
Rauch A and Anders A 2013 Vacuum 89 53–6 241101
Rauschenbach B 2002 Vacuum 69 3–10 Swanson D E, Geisthardt R M, McGoffin J T, Williams J D and Sites
Ready J 1971 Effects of High-Power Laser Radiation (New York: J R 2013 IEEE J. Photovoltaics 3 838–42
Academic) Swanson D E, Kephart J M, Kobyakov P S, Walters K, Cameron K C,
Robson A E and von Engel A 1956 Phys. Rev. 104 15–6 Barth K L, Sampath W S, Drayton J and Sites J R 2016 J. Vac.
Rohde S L 1994 Unbalanced magnetron sputtering Plasma Sources Sci. Technol. A 34 021202
for Thin Film Deposition and Etching (Physics of Thin Films Swanson D E, Lutze R M, Sampath W S and Williams J D 2011
vol 18) ed M H Francombe and J L Vossen (New York: Plasma cleaning of TCO surfaces prior to CdS/CdTe deposition
Academic) pp 235–88 38th IEEE Photovoltaic Specialists Conf. (Austin, Texas 3–8
Rossnagel S M 1999 IBM J. Res. Dev. 43 163–79 June 2012) (IEEE) pp 859–63
Rossnagel S M 2003 J. Vac. Sci. Technol. A 21 S74–87 Takahashi C, Kiuchi M, Ono T and Matsuo S 1988 J. Vac. Sci.
Rossnagel S M and Cuomo J J 1988 MRS Bull. 13 40–5 Technol. A 6 2348–52
Rossnagel S M and Hopwood J 1993 Appl. Phys. Lett. 63 3285–7 Tanberg R 1930 Phys. Rev. 35 1080–9
Rossnagel S M and Hopwood J 1994 J. Vac. Sci. Technol. B 12 Teer D G 1976 J. Phys. D: Appl. Phys. 9 L187–9
449–53 Thompson M W 1968 Phil. Mag. 18 377–414
Rossnagel S 2000 Ionization by radio frequency inductively coupled Thompson M W 1981 Phys. Rep. 69 335–71
plasma Ionized Physical Vapor Deposition (Thin Films vol 27) Thornton J A 1988 Physical vapor deposition Semiconductor Mate-
ed J A Hopwood (New York: Academic) pp 37–66 rials and Process Technology Handbook for Very Ultra Large
Rudolph M, Brenning N, Hajihoseini H, Raadu M A, Minea T M, Scale Integration (VLSI) and Ultra Large Scale Integra-
Anders A, Gudmundsson J T and Lundin D 2022 J. Phys. D: tion (ULSI) ed G F McGuire (Norwich, New York: Noyes
Appl. Phys. 55 015202 Publications) pp 329–454
Rudolph M, Brenning N, Raadu M A, Hajihoseini H, Gudmundsson Thornton J A and Penfold A S 1978 Cylindrical magnetron sputtering
J T, Anders A and Lundin D 2020 Plasma Sources Sci. Technol. Thin Film Processes vol 4 ed J L Vossen and W Kern (New York:
29 05LT01 Academic) pp 75–113
Samuelsson M, Lundin D, Jensen J, Raadu M A, Gudmundsson J T Tsikata S and Minea T 2015 Phys. Rev. Lett. 114 185001
and Helmersson U 2010 Surf. Coat. Technol. 205 591–6 Uhlmann S, Frauenheim T and Lifshitz Y 1998 Phys. Rev. Lett. 81
Sarakinos K, Alami J and Konstantinidis S 2010 Surf. Coat. Technol. 641–4
204 1661–84 Vetter J 2014 Surf. Coat. Technol. 257 213–40
Sarakinos K, Alami J and Wuttig M 2007 J. Phys. D: Appl. Phys. 40 Vetushka A, Brown R, Azzopardi A, Bellido-Gonzalez V, Li H and
2108–14 Papa F 2015 Full face erosion planar cathodes as a low cost
Scherer M, Schmitt J, Latz R and Schanz M 1992 J. Vac. Sci. Technol. cylindrical rotatable R & D tool Proc. 58th Society of Vacuum
A 10 1772–6 Coaters Annual Technical Conf. (Santa Clara, California 25-30
Schiller S, Goedicke K, Reschke J, Kirchhoff V, Schneider S and April 2015) (Albuquerque, New Mexico: Society of Vacuum
Milde F 1993 Surf. Coat. Technol. 61 331–7 Coaters) pp 566–70
Schiller S, Heisig U, Korndörfer C, Beister G, Reschke J, Vossen J L 1971 J. Vac. Sci. Technol. 8 S12–30
Steinfelder K and Strümpfel J 1987 Surf. Coat. Technol. 33 Vossen J L and Cuomo J J 1978 Glow discharge sputter deposition
405–23 Thin Film Processes vol 4 ed J L Vossen and W Kern (New York:
Schiller S, Heisig U, Steinfelder K, Strümpfel J, Voigt R, Fendler R Academic) pp 11–73
and Teschner G 1982 Thin Solid Films 96 235–40 Vossen J L and O’Neill J J 1968 RCA Rev. 29 566–81
32
Waits R K 1978 J. Vac. Sci. Technol. 15 179–87 Xu J, Deng X, Zhang J, Lu W and Ma T 2001 Thin Solid Films 390
Wang W, Foster J, Snodgrass T, Wendt A E and Booske J H 1999 J. 107–12
Appl. Phys. 85 7556–61 Yamamura Y 1981 Radiat. Eff. Defects Solids 55 49–55
Wasa K and Hayakawa S 1967a Proc. IEEE 55 2179–80 Yamamura Y, Takiguchi T and Ishida M 1991 Radiat. Eff. Defects
Wasa K and Hayakawa S 1967b Microelectron. Reliab. 6 213–21 Solids 118 237–61
Wasa K and Hayakawa S 1969 Rev. Sci. Instrum. 40 693–7 Yamamura Y and Tawara H 1996 At. Data Nucl. Data Tables 62
Wendt A E 1988 Dynamics of a planar magnetron discharge PhD 149–253
Thesis University of California at Berkeley, Berkeley, California Yang Y, Tanaka K, Liu J and Anders A 2015 Appl. Phys. Lett. 106
Wendt A E, Lieberman M A and Meuth H 1988 J. Vac. Sci. Technol. 124102
A 6 1827–31 Ye B et al 2021 Rev. Sci. Instrum. 92 113906
Westwood W D 1976 Prog. Surf. Sci. 7 71–111 Yoshida Y 1992 Appl. Phys. Lett. 61 1733–4
Willmott P R and Huber J R 2000 Rev. Mod. Phys. 72 315–28 Zhirkov I, Petruhins A and Rosen J 2015 Surf. Coat. Technol. 281
Window B and Savvides N 1986 J. Vac. Sci. Technol. A 4 196–202 20–6
Wright A W 1877 Am. J. Sci. s3-14 169–78 Ziegler J F, Biersack J P and Ziegler M D 2008 SRIM—The Stopping
Wright M and Beardow T 1986 J. Vac. Sci. Technol. A 4 388–92 and Range of Ions in Matter (Chester, Maryland: SRIM Co.)
Wu L, Ko E, Dulkin A, Park K J, Fields S, Leeser K, Meng L and Ziegler J F, Ziegler M D and Biersack J P 2010 Nucl. Instrum.
Ruzic D N 2010 Rev. Sci. Instrum. 81 123502 Methods Phys. Res. B 268 1818–23
33

Sputtering Paper

Uploaded by

Copyright:

Available Formats

Sputtering Paper

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sputtering Paper

Uploaded by

Copyright:

Available Formats

Plasma Sources Science and Technology

TOPICAL REVIEW • OPEN ACCESS You may also like

This content was downloaded from IP address 14.139.158.18 on 18/10/2023 at 16:58

Plasma Sources Sci. Technol. 31 (2022) 083001 (33pp) https://doi.org/10.1088/1361-6595/ac7f53

Foundations of physical vapor deposition

Received 29 May 2021, revised 16 June 2022

where Th = T/Tm is the film temperature normalized by the

where Mα is the mass of the energetic particle, Ms is the mass

2.1. Thermal evaporation

The first commercial thin film deposition technique was ther-

Figure 5. ECR discharge based ionization of evaporated copper

Figure 8. A schematic of IBAD system. It consists of evaporation

Figure 7. A schematic of plasma-based ion plating or a

Figure 9. A schematic of a typical PLD setup. A laser beam is

When the laser power density reaches a certain value, mate-

2.5. Other evaporation-based PVD techniques

Molecular beam epitaxy (MBE) is a thin film deposition tech-

applied for the production of reflecting surfaces on mirrors

which flows during most of the rf cycle. Therefore, a nearly

The advantage of diode sputtering is efficient use of the

Bogaerts 2009, Kolev et al 2005). The discharge voltage

themselves. The magnetic field is made strong enough to con-

back-attracted by the electric field to the biased target. This is

reactive gas. The difference in sputter yields causes the depo-

a dense plasma in the region between the cathode target and

and Hopwood 1994). The ions of the sputtered material can

Ion beam sputter deposition is a PVD technique that is used to

Figure 27. Erosion traces left by a 10 A vacuum arc on a cadmium

4.3. Substrate bias effects

With intentional (negative) bias, the substrate surface poten-

4.4. Macroparticles 4.5. Effect of the reactive gas

Cathode poisoning is therefore one of the strategies to reduce

4.6. Steered arc sources

Another strategy to reduce macroparticles is to make use of the

You might also like