A time-dependent plasma discharge model has been developed for the ionization region in a high-power impulse magnetron sputtering (HiPIMS) discharge. It provides a flexible modeling tool to explore, e.g., the temporal variations of the ionized fractions of the working gas and the sputtered vapor, the electron density and temperature, and the gas rarefaction and refill processes. A separation is made between aspects that can be followed with a certain precision, based on known data, such as excitation rates, sputtering and secondary emission yield, and aspects that need to be treated as uncertain and defined by assumptions. The input parameters in the model can be changed to fit different specific applications. Examples of such changes are the gas and target material, the electric pulse forms of current and voltage, and the device geometry. A basic version, ionization region model I, using a thermal electron population, singly charged ions, and ion losses by isotropic diffusion is described here. It is fitted to the experimental data from a HiPIMS discharge in argon operated with 100 µs long pulses and a 15 cm diameter aluminum target. Already this basic version gives a close fit to the experimentally observed current waveform, and values of electron density n e , the electron temperature T e , the degree of gas rarefaction, and the degree of ionization of the sputtered metal that are consistent with experimental data. We take some selected examples to illustrate how the model can be used to throw light on the internal workings of these discharges: the effect of varying power efficiency, the gas rarefaction and refill during a HiPIMS pulse, and the mechanisms determining the electron temperature.

In most models of sputtering magnetrons, the mechanism for energizing the electrons in the discharge is assumed to be sheath energization. In this process, secondary emitted electrons from the cathode surface are accelerated across the cathode sheath into the plasma, where they either ionize directly or transfer energy to the local lower energy electron population that subsequently ionizes the gas. In this work, we present new modeling results in support of an alternative electron energization mechanism. A model is experimentally constrained, by a fitting procedure, to match a set of experimental data taken over a large range in discharge powers in a high-power impulse magnetron sputtering (HiPIMS) device. When the model is matched to real data in this way, one finding is that the discharge can run with high power and large gas rarefaction without involving the mechanism of secondary electron emission by twice-ionized sputtered metal. The reason for this is that direct Ohmic heating of the plasma electrons is found to dominate over sheath energization by typically an order of magnitude. This holds from low power densities, as typical for dc magnetrons, to so high powers that the discharge is close to self-sputtering, i.e. dominated by the ionized vapor of the sputtered gas. The location of Ohmic heating is identified as an extended presheath with a potential drop of typically 100-150 V. Such a feature, here indirectly derived from modeling, is in agreement with probe measurements of the potential profiles in other HiPIMS experiments, as well as in conventional dc magnetrons.

The time-dependent plasma discharge ionization region model (IRM) has been under continuous development during the past seven years and used in several studies of the ionization region of high-power impulse magnetron sputtering (HiPIMS) discharges. In the present work, a complete description of the most recent version of the IRM is given, which includes improvements, such as allowing for returning of the working gas atoms from the target, a separate treatment of hot secondary electrons, addition of doubly charged metal ions, etc. To show the general applicability of the IRM, two different HiPIMS discharges are investigated. The first set concerns 400 µs long discharge pulses applied to an Al target in an Ar atmosphere at 1.8 Pa. The second set focuses on 100 µs long discharge pulses applied to a Ti target in an Ar atmosphere at 0.54 Pa, and explorers the effects when varying the magnetic field strength. The model results show that Al 2+-ions contribute negligibly to the production of secondary electrons, while Ti 2+-ions effectively contribute to the production of secondary electrons. Similarly, the model results show that for an argon discharge with Al target the contribution of Al +-ions to the discharge current is over 90 % at 800 V, while Al +-ions and Ar +-ions contribute roughly equally to the discharge current at 400 V. For high currents the discharge with Al target develops almost pure self-sputter recycling, while the discharge with Ti target exhibits close to a 50/50 combination of self-sputter recycling and working gas-recycling. For a Ti target, a self-sputter yield significantly below unity makes working gas-recycling necessary at high currents. In the discharge with Ti target the B-field was reduced in steps from the nominal value, which resulted in a corresponding stepwise increase in the discharge resistivity.

A new reactive ionization region model (R-IRM) is developed to describe the reactive Ar/O 2 high power impulse magnetron sputtering (HiPIMS) discharge with a titanium target. It is then applied to study the temporal behavior of the discharge plasma parameters such as electron density, the neutral and ion composition, the ionization fraction of the sputtered vapor, the oxygen dissociation fraction, and the composition of the discharge current. We study and compare the discharge properties when the discharge is operated in the two well established operating modes, the metal mode and the poisoned mode. Experimentally, it is found that in the metal mode the discharge current waveform displays a typical non-reactive evolution, while in the poisoned mode the discharge current waveform becomes distinctly triangular and the current increases significantly. Using the R-IRM we explore the current increase and find that when the discharge is operated in the metal mode Ar + and Ti + -ions contribute most significantly (roughly equal amounts) to the discharge current while in the poisoned mode the Ar + -ions contribute most significantly to the discharge current and the contribution of O + -ions, Ti + -ions, and secondary electron emission is much smaller. Furthermore, we find that recycling of atoms coming from the target, that are subsequently ionized, is required for the current generation in both modes of operation. From the R-IRM results it is found that in the metal mode self-sputter recycling dominates and in the poisoned mode working gas recycling dominates. We also show that working gas recycling can lead to very high discharge currents but never to a runaway. It is concluded that the dominating type of recycling determines the discharge current waveform.

Model studies of 400 µs long discharge pulses in high-power impulse magnetron sputtering have been made to study the gas dynamics and plasma chemistry in this type of pulsed processing plasma. Data are taken from an experiment using square voltage pulses applied to an Al target in an Ar atmosphere at 1.8 Pa. The study is limited to low power densities, <0.5 kW cm −2 , in which the discharge is far away from the runaway self-sputtering mode. The model used is the ionization region model, a time-dependent plasma chemistry discharge model developed for the ionization region in magnetron sputtering discharges. It gives a close fit to the discharge current during the whole pulse, both an initial high-current transient and a later plateau value of constant lower current. The discharge current peak is found to precede a maximum in gas rarefaction of the order of n Ar /n Ar,0 ≈ 50%. The time durations of the high-current transient, and of the rarefaction maximum, are determined by the time it takes to establish a steady-state diffusional refill of process gas from the surrounding volume. The dominating mechanism for gas rarefaction is ionization losses, with only about 30% due to the sputter wind kick-out process. During the high-current transient, the degree of sputtered metal ionization reaches 65-75%, and then drops to 30-35% in the plateau phase. The degree of self-sputtering (defined here as the metal ion fraction of the total ion current to the target) also varies during the pulse. It grows from zero at pulse start to a maximum of 65-70% coinciding in time with the maximum gas rarefaction, and then stabilizes in the range 40-45% during the plateau phase. The loss in deposition rate that can be attributed to the back-attraction of the ionized sputtered species is also estimated from the model. It is low during the initial 10-20 µs, peaks around 60% during the high-current transient, and finally stabilizes around 30% during the plateau phase.

The onset and development of self-sputtering (SS) in a high power impulse magnetron sputtering (HiPIMS) discharge have been studied using a plasma chemical model and a set of experimental data, taken with an aluminum target and argon gas. The model is tailored to duplicate the discharge in which the data are taken. The pulses are long enough to include both an initial transient and a following steady state. The model is used to unravel how the internal discharge physics evolves with pulse power and time, and how it is related to features in the discharge current-voltage-time characteristics such as current densities, maxima, kinks and slopes. The connection between the self-sputter process and the discharge characteristics is quantified and discussed in terms of three parameters: a critical target current density J crit based on the maximum refill rate of process (argon) gas above the target, an SS recycling factor SS-recycle , and an approximation α of the probabilities of ionization of species that come from the target (both sputtered metal and embedded argon atoms). For low power pulses, discharge voltages U D 380 V with peak current densities below ≈ 0.2 A cm −2 , the discharge is found to be dominated by process gas sputtering. In these pulses there is an initial current peak in time, associated with partial gas rarefaction, which is followed by a steady-state-like plateau in all parameters similar to direct current magnetron sputtering. In contrast, high power pulses, with U D 500 V and peak current densities above J D ≈ 1.6 A cm −2 , make a transition to a discharge mode where SS dominates. The transition is found not to be driven by process gas rarefaction which is only about 10% at this time. Maximum gas rarefaction is found later in time and always after the initial peak in the discharge current. With increasing voltage, and pulse power, the discharge can be described as following a route where the role of SS increases in four steps: process gas sputtering, gas-sustained SS, self-sustained SS and SS runaway. At the highest voltage, 1000 V, the discharge is very close to, but does not go into, the SS runaway mode. This absence of runaway is proposed to be connected to an unexpected finding: that twice ionized ions of the target species play almost no role in this discharge, not even at the highest powers. This reduces ionization by secondary-emitted energetic electrons almost to zero in the highest power range of the discharge.

The lower deposition rate for high power impulse magnetron sputtering (HiPIMS) compared to direct current magnetron sputtering for the same average power is often reported as a drawback. The often invoked reason is back-attraction of ionized sputtered material to the target, due to a substantial negative potential profile from the location of ionization towards the cathode. Emitting and swept Langmuir probes have yielded space-and time resolved electric potential profiles and electron energy distributions, Rogowski coils have been used to obtain current density distributions. Also, space-and time resolved, and fast imaging techniques show how the time evolution of the discharge structure varies with gas pressure and species. This combined data set is here used to benchmark two different types of plasma models for different regions of the HiPIMS discharges, with special focus on the problem of electric fields z E in the high density plasma region and their effect on the transport of ionized sputtered material. We propose two different mechanisms to be dominating in different regions: "ionization driven" , z ioniz E in a rather stable ionization region extending a few cm from the target, and "transport driven" , z trans E in the highly dynamic surrounding bulk plasma.2

The transport of charged particles in a high power impulse magnetron sputtering (HiPIMS) discharge is of considerable interest when optimizing this promising deposition technique with respect to deposition rate and control of the ion acceleration. In this study the internal current densities Jφ (azimuthal direction) and Jz (axial direction) have therefore been spatially and temporally resolved in the bulk plasma region above a cylindrical magnetron using Rogowski coils. From the measurements a phenomenological model has been constructed describing the evolution of the current density in this pulsed plasma. The core of the model is based on three different types of current systems, which characterize the operating transport mechanisms, such as current transport along and across magnetic field lines. There is a gradual change between these current systems during the initiation, build-up and steady state of a HiPIMS plasma. Furthermore, the data also show that there are spatial and temporal variations of the key transport parameter Jφ/Jz, governing the cross-B resistivity and also the energy of the charged particles. The previously reported faster-than-Bohm cross-B electron transport is verified here, but not for all locations. These results on the plasma dynamics are essential input when modeling the axial electric field, governing the back-attraction of ionized sputtered material, and might furthermore provide a link between the different resistivities reported in HiPIMS, pulsed-DC, and DC magnetron discharges.