Preface to Special Topic: Reactive high power impulse magnetron sputtering

Magnetron sputtering is a widely used physical vapor deposition technique that has been applied successfully by various industries for the deposition of thin films and coatings for over four decades. The high power impulse magnetron sputtering (HiPIMS) discharge is a recent addition to plasma-based sputtering. In HiPIMS, high power is applied to the magnetron target in unipolar pulses at a low duty cycle and low repetition frequency while keeping the average power about two orders of magnitude lower than the peak power. This high discharge power leads to the generation of a very high density plasma and thus a high ionization fraction of sputtered atoms. Thus, for the sputtered material, the ion flux is larger than the neutral flux and HiPIMS can be referred to as ionized physical vapor deposition. Ionized sputtered flux allows for controlled ion bombardment of the growing film by the acceleration of the sputtered material across the plasma sheath created by a negative bias applied to the substrate.

Reactive sputtering, where metal targets are sputtered in a reactive gas atmosphere (e.g., O₂, N₂, CH₄, etc.) to deposit compound materials, including various protective and decorative coatings such as transparent conductive oxides (TCO), permeation barrier coatings, and hard coatings, is of utmost importance in various technologies. The high electron density in high power impulse magnetron sputtering (HiPIMS) discharge is expected to enhance the dissociation of molecular gases, which is sometimes considered to be beneficial for the oxide, nitride, or carbide deposition. Reactive HiPIMS has been successfully used for the deposition of various compound films including various protective metal oxides and nitrides such as Al₂O₃, CrN, and TiN and various transparent metal oxides such as TiO₂, HfO₂, and ZrO₂ and transparent conductive oxides (TCO) including Al-doped ZnO and RuO₂. The films grown by HiPIMS are usually denser, have a higher degree of crystallinity, and have smoother surfaces than films grown by conventional magnetron sputtering (dc or mid-frequency asymmetric bipolar pulsed) at the same average power and pressure.

Despite the success of the reactive HiPIMS technique in growing high quality compound films, the underlying plasma mechanisms are still not well understood. With the Special Topic Section on reactive HiPIMS, we wish to highlight the most recent developments aiming to show the full potential of this technique which is still restricted by a limited understanding of the plasma physics of the pulsed reactive processes. This Special Topic Section on reactive HiPIMS opens with one invited tutorial and one invited perspective, which are followed by 18 invited papers.

The underlying physics of the pulsed reactive magnetron sputtering process is discussed in a tutorial given by Anders.¹ It starts by covering the basics of sputtering and why the magnetron sputtering discharge was developed, followed by discussion on topics such as the reactive sputtering process, the issues of target poisoning, hysteresis, self-sputtering, gas rarefaction, localized ionization zones, and working gas recycling. These topics are discussed as introduction to a more thorough discussion of the reactive HiPIMS process. This tutorial also includes a summary of the material systems that have been successfully deposited by reactive HiPIMS.

The presence of a reactive gas can lead to the formation of compound materials on the target surface, often referred to as target coverage or target poisoning. Due to this target coverage, the reactive sputtering process is inherently unstable and is commonly represented in a familiar hysteresis curve that shows, e.g., the deposition rate or the target voltage versus the flow rate of the reactant molecular gas. There have been somewhat conflicting claims regarding the hysteresis effect in the reactive HiPIMS discharge in the literature, as some report the reduction or elimination of the hysteresis effect, while others claim it to be more significant in the HiPIMS discharge and that a feedback control is essential. This issue is discussed in an invited perspective by Strijckmans et al.,² which uses modeling and simulation to argue that the hysteresis effect is much reduced in reactive HiPIMS compared to dc magnetron sputtering (dcMS). This is mainly due to implantation of ionized metal atoms that return to the target, while the high current pulses and gas rarefaction effects can be ruled out as a source of the hysteresis reduction. The reduced hysteresis in reactive HiPIMS is also explored by an analytical model of reactive pulsed sputtering by Kadlec and Čapek.³ This model combines a Berg-type model of reactive sputtering with the phenomenological HiPIMS model of Christie-Vlček. This model predicts that the reduced hysteresis in HiPIMS is due to the return of target material ions or to the less negative slope of the metal flux to substrates and of reactive gas sorption as functions of reactive gas partial pressure than in dc or mid-frequency pulsed magnetron sputtering. Thus, this model predicts a significantly lower critical pumping speed for the HiPIMS discharge. This work is followed by a systematic study of the reduced hysteresis in reactive HiPIMS where titanium and aluminum targets are reactively sputtered in Ar/O₂ and Ar/N₂ gas mixtures.⁴ In this study, reactive HiPIMS is compared to dcMS and mid-frequency pulsed magnetron sputtering operated at the same average discharge power. The lower critical pumping speed, lowered target coverage at a given partial pressure, and a reduced hysteresis are confirmed for the HiPIMS process. Kubart and Aijaz⁵ explore the dynamics of the formation and the removal of a compound on the titanium surface from the evolution of discharge characteristics in an argon atmosphere with nitrogen and oxygen. They determine that the time response of a reactive process is dominated by surface processes. They also find that the thickness of the compound layer is several nm and its removal by sputtering requires much larger ion fluence than a single HiPIMS pulse provides. Furthermore, they find that the formation of the nitride or oxide layer is significantly slower in HiPIMS than in dcMS under identical conditions. The hysteresis effect is also studied experimentally combining optical emission, optical absorption, and laser-based diagnostic techniques by Britun et al.⁶ They quantify some of the discharge parameters such as the ground state density of the oxygen atoms, the density of the sputtered atoms and ions, and the argon metastables at various process conditions.

An increase in the discharge current is commonly observed with an increased partial pressure of the reactive gas or decreased repetition pulse frequency. In some cases, the current waveform changes when the discharge enters the poisoned mode. This increase in the discharge current is seen by Keraudy et al.⁷ for Ar/O₂ discharge with a Ni target, by Britun et al.⁶ for Ar/O₂ discharge with a Ti target, and by Kubart and Aijaz⁵ for Ar/O₂ and Ar/N₂ discharge with a Ti target. This increase in the discharge current as the discharge transitions in to the poisoned mode is discussed further by Lundin et al.⁸ They explore the oxygen dynamics in a reactive Ar/O₂ HiPIMS discharge with a Ti target using the reactive ionization region model (R-IRM) which is essentially a volume averaged global model of the plasma chemistry. They identify the dominating physical and chemical reactions in the plasma and on the surfaces of the reactor, which influence the oxygen plasma chemistry by investigating the reaction rates for the gain and the loss of the various species. They find that the density of atomic oxygen increases significantly as the discharge transitions from the metal mode to the transition mode and finally into the compound (poisoned) mode. The main contribution to this increase is sputtering of atomic oxygen from the oxidized target. Similarly, Zheng et al.⁹ present a spatially averaged global model (or R-IRM) of the plasma chemistry in a modulated pulsed power magnetron sputtering (MPPMS) discharge operated in Ar/N₂ mixtures with TiAlSi alloy targets. The MPPMS technique is based on a long pulse (a macropulse) which is a series of micropulses. The macropulse can be arbitrarily tailored by modulating the pulse on/off time of the micropulses. This model predicts time-dependent plasma parameters including number densities of the various species, the electron temperature, and the compound-covered fraction of the target. Furthermore, the model results are compared with the microstructure and the composition of the deposited compound films.

The addition of a reactive gas strongly influences the plasma parameters in the reactive HiPIMS discharge. The electron density $ne$ and the effective electron temperature $Teff$ were measured in an Ar/O₂ discharge with the titanium target using a time-resolved Langmuir probe technique for three different modes of operation including pure argon, metallic, transition, and poisoned modes and compared to the recently developed R-IRM.¹⁰ The results demonstrate that the time evolution of the electron density $ne$ follows the discharge current waveform, while the effective electron temperature $Teff$ increases with the increasing oxygen flow rate. Hippler et al.¹¹ also measured the temporal variation of the electron density and electron energy with a Langmuir probe in Ar/O₂ discharge with a Ti target at various positions within the discharge. The discharge properties were explored during HiPIMS growth of NiO_x using time-resolved optical emission spectroscopy (TR-OES) by Keraudy et al.⁷ Observation from TR-OES suggests that the discharge behavior in the poisoned mode indicates the higher contribution of both oxygen and argon ions in the total ionic current, leading to a change in the effective ion-induced secondary electron emission coefficient. Pajdarová et al.¹² also use TR-OES to investigate the plasma composition both near the target and in the plasma bulk, when depositing ZrO₂ films. They record a decrease in the ground-state densities of argon and oxygen atoms in the target vicinity due to gas rarefaction and intense electron-impact ionization, in the first half of the voltage pulse. Very effective electron-impact ionization of the sputtered Zr atoms was also observed.

It has been recognized that discharges operated with crossed electric and magnetic fields develop instabilities that play a crucial role in the transport processes of charged particles. In magnetron sputtering discharges, ionization and particle transport are governed by a host of waves and instabilities. There have been a number of reports on instabilities in magnetron sputtering discharges, pronounced plasma non-uniformities drifting along the racetrack, referred to as bunches, ionization zones, spokes, or emission structures. These localized ionization zones, which are commonly observed in magnetron sputtering plasmas, appear either in a triangular shape or with a diffuse shape, exhibiting self-organization patterns.¹ Hecimovic et al.¹³ explore the spoke properties (shape and emission) in a reactive HiPIMS discharge with N₂ or O₂ added to the argon working gas, for three target materials: Al, Cr and Ti. For all the investigated materials, the spoke shape changes towards a diffuse spoke shape and an increase in the secondary electron emission is observed as decreased discharge voltage, as the discharge transitions into the poisoned mode. The spokes are also explored when the HiPIMS discharge is operated with Ar-Cr composite targets along with in-situ X-ray photoelectron spectroscopy (XPS) characterization of the target surface.¹⁴ This allowed for exploring the target redeposition of the sputtered species and oxidation at the target racetrack.

The main purpose of developing the reactive HiPIMS discharge and the ionized physical vapor deposition (IPVD) process is to enable the growth of high quality thin films with various properties. The pulsed reactive HiPIMS process provides a number of control parameters to optimize the film properties. Process control and optimization along with novel approaches and applications of reactive HiPIMS is discussed in a number of contributions. The coating properties of NiO_x deposited using reactive HiPIMS are explored by Keraudy et al.⁷ They find that by tuning the oxygen partial pressure, a large range of chemical compositions from pure nickel to nickel-deficient NiO_x (x > 1) can be achieved while operating in the poisoned mode. Furthermore, the optoelectronic properties of NiO_x films can be easily tuned by adjusting the O/Ni ratio and thus through control of the oxygen flowrate. Ganesan et al.¹⁵ demonstrate that it is possible to tune the optical and electrical properties of hafnium oxide, tungsten oxide, and tungsten oxynitride coatings by varying the HiPIMS pulse length and therefore the duty cycle. The deposition process parameters that can be used to tailor the residual coating stress in order to improve the adhesion of SiN_x coatings deposited by reactive HiPIMS are investigated by Schmidt et al.¹⁶ The angular distribution of plasma parameters and the resulting thin film characteristics were explored for Ar/O₂ HiPIMS discharge while varying the under growth angles with respect to the target surface when growing the TiO₂ film.¹¹ The deposition rates were found to be the largest in the forward directions and decrease towards large inclination angles and thus a pronounced anisotropy in the film deposition process and of the deposited layer structure. They suggest that this can be exploited in the manufacture of films with particular properties. Magnfält et al.¹⁷ demonstrate a single-step synthesis process that allows for tuning the plasmonic response of Ag/AlO_xN_y nanocomposite thin films with high precision in the range from green (∼2.4 eV) to violet (∼2.8 eV). They achieve this by employing Modulated Impulse Magnetron Sputtering Interplay (MIMSI) which is essentially well-defined trains of electrical pulses applied to the cathode target, where each single pulse in a pulse train has a width of the order of a few tens of microseconds.

Reactive HiPIMS can also be combined with other discharge systems in a hybrid configuration. Greczynski et al.¹⁸ use VAlN thin films as a model system to illustrate the synthesis of a metastable single-phase NaCl-structure thin film with the Al content far beyond the solubility limits obtained using conventional plasma processes. This supersaturation is achieved by separating the film-forming species in time and energy domains through synchronization of 70 μs long substrate bias pulses with intense periodic fluxes of energetic Al⁺ metal ions during reactive hybrid HiPIMS of an Al target and dcMS of a V target in an Ar/N₂ gas mixture. Similarly, Fager et al.¹⁹ use a hybrid high-power pulsed and dc magnetron co-sputtering (HiPIMS/dcMS) technique to grow fully dense and hard Ti_0.41Al_0.51Ta_0.08N alloys without external heating. The separate Ti and Al targets are operated as dcMS, while the heavy Ta⁺/Ta²⁺-ions are provided by HiPIMS from a Ta target, using pulsed bias synchronized to the metal-ion-rich part of each HIPIMS pulse. Finally, Straňák et al.²⁰ combine a microwave surfatron Ar/O₂ plasma discharge with the reactive HiPIMS discharge in order to activate the oxygen gas in the substrate vicinity. This allows the reduction of the total oxygen flow rate and thus a suppression of the target poisoning effect.

As Guest Editors, we hope that this Special Topic Section will help the readers to gain a better understanding of the reactive HiPIMS process and its opportunities. We also hope it will inspire the readers to generate new research ideas to broaden the understanding of this highly complex yet very much relevant topic. We would like to thank all the authors for their interesting contributions, as well as the reviewers, for providing useful and constructive feedback on the submitted papers. Finally, we would like to thank the Editor-in-Chief, André Anders, as well as the Journal Manager, Benedetta Camarota, and the Peer Review Manager, Ania Bukowski, for coordinating this Special Topic Section.