ICMCTF2015 Session TS5: Plasma Diagnostics and Modeling

The quality of the films produced in plasma materials processing ultimately depends on the fluxes of ions, radicals and photons incident onto the surface. The ability to controlling these fluxes at a desired power or pressure is constrained by the system parameters required to sustain the plasma. That is, using continuous excitation, the plasma sources and losses must always be in balance. Pulsed power processing widens the parameter space available to control reactive fluxes since maintenance of the plasma is only required averaged over the pulsed period. Pulsed plasmas are now widely used in microelectronics fabrication, particularly for etching, however pulsed plasmas are only beginning to be broadly adopted in deposition. Low pressure examples of such successes are HIPIMS (high power impulse magnetron sputtering), ALD (atomic-layer-deposition) and pulsed-plasma-ion-implantation. The majority of high pressure surface modification techniques are by necessity pulsed due to the difficulty of sustaining continuous plasmas at atmospheric pressure, however there are likely regions of the pulsed parameter spaced not yet investigated for their optimization. In this talk, results from modeling studies of low and high pressure plasma materials processing will be discussed with the goal of providing insights on how plasma properties and fluxes can be controlled using pulsed power. Examples will be used from control of electron, ion and VUV photon distributions in low pressure inductively coupled, capacitively coupled and magnetron plasmas; and from atmospheric pressure plasmas intended to treat biological materials.

The High Power Impulse Magnetron Sputtering (HiPIMS) has known a fast development driven mostly by the technological applications. However, the further optimizations of the HiPIMS deposition process, especially for specific coatings, demand necessarily the fundamental understanding of the plasma phase and the transport of particles in the reactor volume, at different moments of the discharge. At nanosecond scale, only electron distribution changes, sometimes inducing instabilities (‘spokes’, flares, etc.). At microsecond scale, the discharge current changes, so the ions move and the sputtering occurs. Tens to hundreds of microseconds represent the time scales for neutral species: gas rarefraction and the lifetime of the sputtered. Finally, milliseconds later, in the afterglow phase, the wall processes evolve towards an equilibrium state, especially in reactive case. The very different timescales of all these processes can be addressed via several models.

Pseudo-3D Particle-in-Cell (PIC) modelling brings information on the current-driven instabilities, which are driven by the relative ion-electron drift speed in these currents, both those along B and across it.

2D PIC modelling of the axisymmetric magnetron is achieved for tens of microseconds via massive parallel computation. It gives a microscopic spatial and temporal description of the phenomena into the plasma volume, e.g. ionization, excitations, energy deposition, electric field reconfiguration, cathode and anode currents, sputtering, etc.

3D a posteriori Monte Carlo modelling based on the self-consistent PIC results is very suitable for the numerical simulation of the flight of sputtered material from the target. It gives not only the fluxes of atoms, energetic and thermal, on the substrate and walls, but also their effect on the buffer gas used to run the discharge, namely its rarefaction.

Finally, the global 0D Ionization Region Model (IRM) gives a precise time occurrence of the key mechanisms during the HiPIMS pulse continuing far in the afterglow. IRM modeling reveals the dependency between rarefaction and ionization losses. It explains also the self-sputtering regime, the behavior of excited species, the plasma reactivity, etc.

Adding the important insight of each model, we can build a detailed composition of the HiPIMS plasma scenarios corresponding to real situations, attractive for many applications.

Fast imaging research of HiPIMS in the last few years has revealed the presence of dense, asymmetric, triangle shaped ionization zones that are more-or-less regularly spaced in the azimuthal direction. These zones have been observed to propagate at several 10³ m/s, up to 10⁴ m/s, yet much slower than the E x B drift velocity of electrons, which is on the order of 10⁵ m/s [1]. In this work, we combine streak and frame imaging to study the plasma in DC magnetron sputtering. Systematic experimental studies were performed at very low currents (less than 1.2 A) for a range of pressures (from 0.1 to 2.7 Pa) in end-on and side-on views. We have discovered reverse direction, sub-structures within ionization zones, arched plasma flares, and zone structural changes greater than previously anticipated.

In contrast to HiPIMS, the ionization zones in DC magnetron sputtering are much more elongated, with a wider “head” at the - E x B end of the zone. Increasing current and/or pressure splits the plasma from a single zone to multiple zones, and eventually, the zones merged and smoothed along the entire racetrack. The zones move with low velocities, on the order of 10³ m/s, in the - E x B direction, which is the opposite of zone motion direction in HiPIMS. The pattern of zones along the racetrack is more-or-less azimuthally symmetric but depends on current and pressure of the process gas.

In streak images we have also detected sub-zones moving in the same direction of electron E x B drift at high velocities on the order of 10⁴ m/s. In close-up frame images, we have observed sub-structures of light intensity having patterns that match the local magnetic field lines, suggesting that they may have been caused by electrons drifting in the E x B direction while gyrating along the B field lines. The shape of the observed sub-structures were highly unstable. From side-on frame images, we have detected arched plasma flares jetting along the racetrack. The discovery of reversal direction, sub-structures, and plasma flares in DCMS greatly promotes the particle transport theory governed by presheath energization, with a modified potential hump structure model and zone-related instabilities and turbulence.

[1] A. Anders, M. Panjan, Robert Franz, Joakim Andersson, and Pavel Ni, Appl. Phys. Lett. 103, 144103 (2013).

Highly ionised pulsed plasma processes have helped to improve coating properties by enabling the control of flux of sputtered ion species.

Deposition of magnetic materials, such as Nickel, is problematic with magnetron sputtering as the magnetic field necessary for the sputter process is reduced due to quenching of the magnetic field by the target material.

With Inductively Coupled Impulse Sputtering (ICIS) we can remove the magnetron. To generate the plasma, pulsed RF power is applied to an internal coil. Ar ions are attracted to the target surface by utilising high power DC pulses on the cathode and initiate sputtering. The sputtered material is then ionised as it passes through the coil volume, creating a highly ionised metal flux to the substrate.

For ICIS the creation rate for gas, metal and metal ion species are unknown.

In the conducted experiments both DC and RF power supplies were running at a duty cycle of 7.5% and were synchronised to ensure the pulse on time was the overlapping. A power-pressure matrix was created to examine the influence on the plasma. At a constant pressure of 13 Pa the applied RF power was varied from 1000-4500 W, at constant applied RF power of 3000 W the pressure was varied from 3-26 Pa. The pulsed DC voltage to the target was kept constant at 1900 V. OES, energy resolved MS and I-V curve measurements were taken for each examined element.

OES measurements for increasing power have shown a linear increase in intensity with increasing power. The slopes of gas and metal species in a log-log graph exhibit a factor 1 increase from the Ar neutral intensity for each excitation and ionisation step, suggesting electron collisions to be the main excitation mechanism.

The IEDFs measured by MS show two sharp peaks, one high intensity peak at 20 eV which corresponds with the plasma potential and is ideal for increased surface mobility without inducing lattice defects. The second lower intensity peak, of high energetic ions, is visible at 170 eV.

We will be discussing the voltage and current waveforms relationships of the slope factor β in an intensity-power graph, the IEDFs vs. power and pressure relations and the origin of higher energetic ions.

In addition to well-established plasma diagnostic methods (Langmuir probes, optical emission spectroscopy, mass spectrometry etc.) we perform examples of “non-conventional” diagnostics [1] which are applicable in technological plasma processes for particle formation, surface structuring and thin film deposition:

i) The total energy influx from plasma to substrates can be measured by special calorimetric probes (passive or active, respectively) as well as by fluorescent micro-particles. Examples for rf-discharge and magnetron sputtering will be provided [2-5]. By comparison with model assumptions on the involved plasma-surface mechanisms the different contributions to the total energy influx can be separated.

ii) For a variety of thin film applications it is essential to determine the sputtering yield as well as the angular distribution of sputtered atoms. For this purpose we developed a novel and rather simple method, the so-called sputtering-propelled instrument (SPIN) [6]. It is stack nearly without friction and exposed to a vertical ion beam, rotating due to momentum transfer by the released particles, i.e. sputtered target atoms and reflected ions. Comparison of the measurements with simulation yields valuable information on the sputtering mechanisms and support validation of related sputter codes. The angular distribution of sputtered particles has also been measured by a sensitive pendulum which is commonly used for thrust measurements in ion beam sources for space propulsion systems.

References

[1] T. Trottenberg et.al., Plasma Phys. Control Fusion 54(2012) 124005.

[2] H. Maurer, H. Kersten, J. Phys. D: Appl. Phys. 44(2011) 174029.

[3] I. Levchenko, M. Keidar, S. Xu, H. Kersten, K. Ostrikov, JVSTB 31(2013) 050801.

[4] K. Nishiyama et.al., J. Nucl. Mat. 438(2013), S788-S791.

[5] H. Kersten, H. Deutsch, H. Steffen, et.al., Vacuum 63(2001) 385-431.

Few layer crystalline and amorphous boron nitride (BN) films produced by laser ablation exhibit large area, stoichiometric, pinhole free growth on a variety of substrate materials at much reduced temperatures from traditional CVD growth methods. By adjusting the incoming laser flux as well as the background pressure within the deposition chamber, a strict control of deposition rates, stoichiometry, and energies of incoming plasma species can be achieved. In this study, chemistry, energies, time of flight data, and specific component’s spatial distributions within the ablated plume were investigated using an ICCD camera with a spectrometer and characterized to determine optimal film growth conditions. Laser ablation occurred from a 248 nm KrF excimer laser from an amorphous boron nitride target in vacuum and in various pressures of nitrogen gas. The identification of energetic plasma species formed from background gas collisions, secondary excitation at the plume/substrate interface, as well as the increased dwell time of the excited species at the condensate surface are discussed and provide insight into understanding crystal growth at reduced substrate temperatures. Dynamics of the atomic and molecular species in the ablated plume were spectroscopically analyzed and used to optimize growth of 1-2 nm thick films over large areas, confirmed by techniques including Raman, AFM, TEM, and XPS.

The transient analysis of the voltage and current waveforms recorded during plasma electrolytic oxidation of aluminium was performed. The decomposition of the experimental bipolar pulses uncovered three significant transients, two in the current, one in the voltage. The transients were approximated with a superposition of exponential functions which helped to propose an equivalent circuit diagram. Functions being the solutions for second order aperiodical transient process problem have been used for the analysis. The average value of R² for all the fits is 0.94 which is consistent with the accuracy of the voltage stabilization.

An inverse problem of an equivalent circuit synthesis was solved on the base of Kirchhoff’s differential equations, and its correctness was justified by a simulation using a circuit modelling software. It was shown that the components can be attributed to certain parts of the coating, and their values evolve with the coating growth.

Investigation of the transients over all the experimental design shows that the time constants and steady values vary with the PEO treatment time and with the positive and negative pulse voltages. Therefore, the transient analysis of the waveforms will help to uncover the PEO electrolyzer equivalent circuit, and the evolution of its component values will help to evaluate the coating thickness and other surface properties during the treatment.

Finally , the proposed method can be applied in a process control system for diagnostics of the unobservable surface properties during plasma electrolytic oxidation.