Application of a pulsed atmospheric arc plasma jet for low‐density polyethylene coating

Authors: Dr. Dariusz Korzec|Dr. Stefan Nettesheim
First published: 05 November 2019 https://doi.org/10.1002/ppap.201900098
© 2019 relyon plasma GmbH. Published by WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim.

Abstract

The development of a method for deposition of low‐density polyethylene (LDPE) on rotating three‐dimensional aluminum alloy substrates is presented. LDPE powder with an average diameter of 50 µm is dispersed with nitrogen as a carrier gas. This suspension is injected into a plasma plume generated by a pulsed atmospheric arc plasma jet (PAA‐PJ). The optimized PAA‐PJ operating parameters are a pulse frequency of 63 kHz, a power of 1,500 W, a plasma gas (mixture of nitrogen with 3.3% hydrogen) flow of 54 SLM (standard litre per minute), a carrier gas flow of 9 SLM, and a powder rate of up to 5 g/min. The plasma head is moved at 200 mm/s, and the substrate is rotated at 10 rev/min. The net deposition rate reaches 2.5 g/min. A homogeneous, high‐quality film with 1‐mm thickness is deposited on an approximately 100 cm² substrate surface.

1. Introduction

The deposition of high‐quality low‐density polyethylene (LDPE) on metal surfaces with superior adhesion between the two materials is of broad interest for many industrial applications. This deposition allows the mechanical strength of the metal to be combined with the extreme chemical resistance, flexibility and low mass density of LDPE. Established methods based on hot‐melt distribution, such as extrusion coating, 1 hot‐melt spraying, or printing from filaments, do not ensure superior film adhesion. The low‐pressure plasma‐based methods, such as plasma polymerization or plasma‐enhanced chemical vapor deposition, exhibit low deposition rates. The deposition of polymer films using different types of atmospheric pressure plasma jets (APPJs) is reviewed in Fanelli et al. 2 However, for many practical industrial applications, none of these approaches offers sufficient deposition rates.

Different thermal spraying techniques are commonly used for the deposition of PE on pipeline components 3 with high deposition rates. The frequent applications of thermally sprayed PE coatings include protection against corrosion and corrosive wear. 4 The disadvantage of conventional thermal spraying lies in its use of plasma torches with high power levels, which are optimized for high gas flow rates and high mass flows of the deposited material. 5, 6 This disadvantage makes precise dosing and film quality control difficult. Due to the very high temperatures, the plasmas used in such instruments are also not suitable for surface activation.

A method should be found that combines the advantages of conventional thermal plasma‐spraying and low‐temperature APPJs. 7 On the one hand, the processing speed should allow coating of a 1‐mm LDPE film on an approximately 100 cm² surface in <5 min. On the other hand, good activation of the substrate surface through the use of the applied APPJ should be possible.

In this paper, the development of a cold plasma‐spraying technique is presented in which LDPE powder becomes molten and is distributed on aluminum substrates through the use of a pulsed atmospheric arc plasma jet (PAA‐PJ). In contrast to the established plasma‐spraying methods, the PAA‐PJ method allows for a much more precise thermal design of the coating process and control of the thermal flux to the powder, which is a prerequisite of such a process. This approach is realized by injecting powder with a well‐defined size distribution into the respective zone of the PAA‐PJ to realize effective melting and avoid overheating and consequently thermolysis of LDPE. A further challenge is the precise control of the temperature of the already deposited layer. The plasma plume should be as close as possible to the deposited surface to achieve effective surface activation for good adhesion of the next layer, while thermolysis should be avoided because it causes an inferior film quality. The power coupling should be adapted to the thickness of the deposited layer. Initially, the aluminum surface is activated to achieve good adhesion. The growing film acts as a thermal barrier, and the coupled power is gradually reduced by changing the plasma settings (e.g., increasing the distance of the plasma plume or the linear motion speed). The influence of the power coupled into the layer and substrate will be determined to optimize the growth rate of the LDPE layer and the temperature increase of the aluminum substrate. The settings of the distance of the plasma jet and the relative velocity (rotation speed and linear motion speed) fulfilling the thermal requirements and minimizing the overspray will also be discussed. The other important parameters of the process are the transfer rate of the LDPE powder and the amount of gas needed for the conveying process. The objective is the determination of the optimum delivery rate of the powder.

2. Coating system

The main task of the applied coating system is to melt the LDPE powder and distribute the molten droplets over the surface of the substrate. The applied coating method can be categorized as cold plasma‐spraying.8–10 The application of a pulsed high‐voltage (HV) arc for plasma and powder heating enables precise thermal management and high film quality. The main components of the coating system developed in the framework of this study are displayed in Figure 1. In the following section, we explain how this coating system works.

The LDPE powder from the powder tank (see Figure 1) is dispersed with a carrier gas. The different operating principles and variations of the dispersion systems evaluated in this study are described in Section 4.1. For all tests and processes, the LDPE spherical powder ADMER NR106, NS101 of Mitsui Chemicals, Inc. with D50 = 50 µm is used. 11 To avoid oxidation of LDPE, a nonoxidizing gas, specifically nitrogen, is used.

In the early stage of powder‐feeding experiments, clogging of the pumps or injectors frequently occurred. This problem was investigated in detail, and 1‐ to 3‐mm long macroscopic foreign substance particles, 3–5 particles/liter in the powder, were determined to be the reason. Thus, the entire powder used for conveying in the following experiments was sieved through a 600 μm vibrating analytic sieve. This procedure eliminated clogging events.

The LDPE powder and carrier gas dispersion is transferred over a pipeline system to the injector (see Figure 1). The tubing of the pipeline should be electrically conducting to avoid build‐up of static charge and parasitic discharges in the tubes. At the same time, the pipeline must be flexible to follow the linear movement of the plasma jet. An antistatic plastic tube exhibiting these two properties is applied.

In the plasma jet, the electrical energy of HV pulses is converted into heat in the pulsed arc (see Section 3.1 for details). In this arc plasma zone, the plasma gas is heated.

The powder injector forms the powder suspension into a stream and provides this stream with the speed needed to penetrate the arc plasma zone. When the LDPE particles cross the arc plasma zone, they are heated and melt. To evaluate the energy efficiency of the process, knowing the heat from the arc plasma coupled into the LDPE powder sufficient to melt it would be interesting. The net power needed for melting all of the LDPE particles flowing through the plasma at a powder rate can be expressed using the following formula 12: (1) where is the temperature of the LDPE powder before entering the plasma and is the temperature of completely molten LDPE, assumed for this calculation as 20°C and 140°C, respectively. is the specific heat of LDPE, C_LDPE = 2,300 J·kg⁻¹·K⁻¹. is the heat of fusion for 45% crystalline LDPE measured with a thermogravimetric analyzer, H_m = 135 kJ/kg. 13 Assuming the gross powder rate of R_P = 4.1 g/min (see Section 4.6), the net power needed for melting calculated using formula 1 is  W. This value is 64% of the net power coupled into the plasma, which is approximately 372 W (see the effective power estimation in Section 3.2 for such a powder rate). The remaining 36% of the power (a) is coupled into the LDPE powder but does not contribute to the film growth (overspray), (b) remains as heat not transferred to LDPE in the plasma and in the carrier gas blown out of the plasma nozzle, and (c) is radiated as heat from the arc mainly to the nozzle inner surface and subsequently transferred by heat conduction through the nozzle material to the outside environment. The amount of energy transferred to the particle surface during a particle’s flight through the plasma depends primarily on the process parameters, especially those that can be explicitly set:

• Geometrical properties such as the injector diameter or distance between the nozzle and the powder injector
• Carrier gas flow
• Temperature and speed of the plasma gas blown out from the nozzle
• Pulse frequency and power
• Type of plasma gas

As the powder injection is a critical process step, determination of the power coupled into the powder and its strong influence on the film quality are discussed in more detail in Section 4.2.

The stream of the plasma gas drags the LDPE droplets in the direction of the substrate surface. The still molten droplets splatter and adhere initially on the substrate and then on the deposited film surface. The heat of the molten LDPE droplets is transferred to the substrate. The droplets themselves solidify and build‐up the coating.

To distribute the liquidized LDPE over the entire surface of the substrate, linear movement of the plasma jet along the injector and rotational movement of the substrate are implemented. For the motion of the plasma jet, an ISEL GFV 44/33 xyz‐robot is used. The coating system is enclosed in a box with transparent walls and gas extraction. The position of the plasma head can be set in Cartesian coordinates with an accuracy of 1 mm. Optional adjustment of the tilt angle is possible. The maximum speed of the plasma head is 250 mm/s. The rotation of the substrate is realized using a motor with an electronically controlled rotation speed. A photograph of the coating plasma in the vicinity of the substrate is shown in Figure 2.

3. Plasma Generation

The most popular form of atmospheric plasma for industrial applications is the APPJ in its numerous variations. 7, 16 Among them, the low‐temperature arc jet 17 has the largest potential for LDPE deposition because it allows the generation of plasma with high local plasma densities. One of these low‐temperature arc jets is the PAA‐PJ. Its discriminating features are the generation of the arc via HV pulses and the stabilization of the arc in space by utilizing a gas vortex. To realize these features, a special construction of the plasma jet and an adequate HV pulse generator are needed. The technical solutions applied in this study are described in the following sections.

3.1 Operating principle of the plasma jet

For all processes developed and investigated in this study, a PG31 plasma jet 14 is applied. The cross‐section of this device is shown in Figure 3. A subsystem critical for the operation of the plasma jet and determination of the process parameters is plasma gas management. The gas is introduced into the plasma jet by use of a plastic tube (refer to Figure 3). The flow divider shapes four equal gas flows following the helical channels crafted into the poly(ether ether ketone) PEEK helix. The purpose of such gas guiding is to set the gas into the rotation. An ancillary effect of the gas flow is cooling of the plasma jet body and especially that of the anode.

The arc needed for plasma generation is ignited between the positively biased inner electrode (anode) and the grounded nozzle (cathode). For this purpose, short positive voltage pulses are supplied to the anode. Considering an air gap of 8 mm, striking of the arc occurs at approximately 15 kV. After establishing the arc, the current between the cathode and anode increases. In typical plasma torches, 18, 19 the arc sustained between the cathode and anode has a small voltage, below 100 V, and very high currents of many amperes. To limit the current rise, vortex stabilization of the arc, such as that in “Linde‐type” gas heaters 20, 21 is applied. In PG31, the ignited arc is dragged by the gas flow along the inner surface of the nozzle in the direction of the nozzle outlet, reaching a length of over 5 cm. This elongation causes an increase in the electrical resistance and consequently a higher operating voltage and a lower current.

The frequency of the pulses in the range of tenths of kHz is high enough to avoid complete extinguishment of the arc between two pulses. The discharge channel remaining along the arc current trace is much easier to reignite than the non‐preionized air gap.

3.2 HV pulse generator

For the generation of HV pulses, a PS2000 OEM commercial power supply 22 is connected over a 10‐m coaxial HV cable to the plasma jet. Owing to the microcontroller system, the operation of the PS2000 OEM is optimized for operation with the PG31 plasma jet. The microcontroller system can not only recognize erroneous conditions, such as a missing plasma jet, lack of plasma gas flow or missing safety switches but it also collects the crucial operation data (mean voltage, current, and power) of the plasma jet and communicates them to the host computer over the CAN‐open interface.

In conventional HV supplies, the ac voltage is transformed to the HV level. Unfortunately, the HV transformers at frequencies in the Hz range are very bulky, which is not the case if signals in the kHz range are transformed from low to high voltage. The PS2000 OEM utilizes this principle in its electronic architecture. The electronic circuit shapes the short, low voltage signals in the kHz range (40–65 kHz), and a compact, low‐loss transformer produces an HV signal.

The power supply controller increases the voltage until the arc is ignited. 23 If breakdown occurs for a voltage lower than the arc ignition value, then the voltage is reduced until the set value of the output power stabilizes. As the sustainment of the arc requires a signal with a much smaller voltage than the plasma ignition value, depending on the plasma gas, values in the range of 1,500 to 3,000 V stabilize. Typical time‐dependent voltage and current curves during the supply of a single pulse to the plasma long after the first ignition, measured for the HV cable with an HV probe and a current probe and collected by an oscilloscope, are shown in Figure 4. The voltage gradually grows over the load capacity, consisting of the cable capacity and correction capacitor capacity until the breakdown voltage of approximately 1,500 V is reached. During the breakdown, a high current peak of over 6 A flows, discharging a capacitance of approximately 1 F through the plasma. The power coupled into the plasma during a single pulse period can be calculated using the general integral formula: (2)

The result of the curves shown is approximately 3 mJ. Multiplying this value by the 50‐kHz frequency used for this measurement, the effective power of 150 W is calculated, in comparison with the 700 W of electrical power consumed by the PS2000 OEM.

As the total energy transferred from the power supply to the arc is proportional to the number of single HV pulses, increasing the frequency of these pulses causes an increase in the coupled power. The limit of 65 kHz for the maximum frequency is related to the speed of the semiconductor power switches used in the power supply. However, the frequency can be almost doubled by connecting two HV pulse units in parallel. Such a configuration (see Figure 1) was used for most of the coating experiments in this study. The effective power for both pulse generators at 62 kHz results in an estimated effective power coupled into the plasma of 2 × 62 kHz × 150 W/50 kHz = 372 W.

3.3 Arc movement

After arc ignition, due to the azimuthal component of the gas motion in the nozzle, the cathodic arc foot sweeps along a spiral trajectory on the conical inner surface of the nozzle to out of the nozzle and continues the rotation motion around the nozzle lip.

When photographs of the plasma jet are taken with a very short exposure time (in the ms range), the shape of the arc appears similar to that sketched in Figure 5a. The cathodic arc foot can be seen on the nozzle lip. The anodic arc foot remains on the tip of the anode. Due to the swirling motion of the air, a drag on the cathodic arc foot exists, which causes its movement around the nozzle lip. This rotation is very fast—a couple of thousand rotations per second. During one rotation cycle, 10–100 current pulses sustain the arc. When taking photographs with a longer exposure time, instead of a single arc, an overlap of the arc at many subsequent positions can be seen, as indicated in Figure 5b. The fast, rotational motion of the arc results in a characteristic pencil‐shaped bright primary plasma, observed at the nozzle orifice during operation of the plasma jet.

3.4 Diffuse plasma

The secondary plasma zone shown schematically in Figure 5b—the so‐called diffuse plasma—consists of the products of interactions between the arc and the gas flowing through it. When the charged particles are confined by electrostatic forces in the core of the arc, neutral particles can freely enter and leave the arc zone, driven by thermodynamic forces. This phenomenon is the reason why the diffuse plasma contains mainly electrically neutral species. However, during the dwell time in the arc, the neutral particles can be energized by electron and photon impacts, resulting in electronically, vibrationally, and rotationally excited species and atomic and molecular radicals, which can transfer the accumulated energy to the substrate surface.

The temperature in the diffuse plasma measured for nitrogen and different compressed dried air (CDA) flows is displayed in Figure 6 as a function of the distance to the nozzle tip. The temperature was statically measured using a K‐type temperature sensor embedded in tubing to avoid electromagnetic compatibility (EMC) interference in the sensor electronics by arcs transferred onto its surface. This temperature does not represent the actual temperature of the surfaces exposed to the plasma plume moving with high speed. The dynamic temperatures are much lower.

In this study, a gas mixture of nitrogen with a small admixture of hydrogen (<5%) is used for chemical and physical reasons. Application of nitrogen, instead of the CDA commonly used for PAA‐PJ operation, should reduce the oxidation of LDPE. The small admixture of hydrogen reduces oxides growing on the hot nozzle, and thus, ensures smooth operation and less nozzle erosion. At the same time, the nitrogen–hydrogen gas mixture causes an increase in the operating voltage from the 1.2 kV typical for CDA to 2.5 kV, which in turn results in better power coupling in the arc.

3.5 Focused plasma

In the diffuse plasma mode, the arc starts at the inner electrode, is expelled several mm out of the nozzle and ends on the nozzle lip. If the nozzle with ignited plasma approaches an electrically conducting and grounded surface or a surface with a large electrical capacity, then the arc jumps over to this surface, forming a focused plasma. A diffuse plasma can no longer be seen, and the arcs are concentrated on a small spot at the substrate. The establishment of a transferred arc is no abrupt event. When decreasing the distance between the substrate surface and the nozzle, the number of arcs ending on the substrate instead of on the nozzle lip increases. To describe such a transition, the definitions of two distances are required: The distance at which single arcs are already transferring to the substrate (limit of the purely diffuse mode) and the distance at which no arcs are ending at the nozzle lip (complete arc transfer). Both transition distances between the nozzle tip and the grounded surface have been determined for nitrogen–hydrogen gas mixtures 24 and are displayed in Figure 7. Both distances, for the transition to the purely diffuse operation mode (upper line in the diagram) and for the transition to the purely transferred arc mode (lower line), decrease with increasing hydrogen percentage. The difference between the start of transfer and complete transfer distances is approximately 3 mm.

An important advantage of the focused plasma is that no erosion of the nozzle occurs, and consequently, the lifetime of the nozzle can be prolonged by an order of magnitude. 25 A further advantage is an increase in the surface roughness, resulting in improved adhesion. Despite these advantages, the focused plasma mode was not applied for pretreatment of the rotating substrates. As the substrates are fixed between holders made of PEEK, they are not grounded. The floating potential induced in the substrate by a transferring arc can cause EMC problems, and consequently, disturbance in the robotic motion and rotation control. To prevent the transfer of arcs onto the metal surface, a safe distance should be kept for all pretreatment processes, over 20 mm for 2% hydrogen in a hydrogen‐nitrogen gas mixture. 24 The distance from the grounded powder injector to the nozzle tip is typically much smaller than this value. To avoid parasitic arc transfer to the tip of the powder injector, the safe radial distance from the nozzle axis is empirically determined to be r_min = 3.7 mm.

3.6 Lifetime of the nozzle

The foot of the arc on both the cathode and anode undergoes erosion due to local melting. 26 The sweeping movement of the cathodic arc foot is crucial for the operation of the plasma jet because it avoids the arc dwelling on one position long enough to cause melting and evaporation of significant amounts of the nozzle material. Owing to this motion, in combination with the special alloy selected for the nozzle tip, the erosion of the nozzle is in the range of micrograms per hour. Thus, a considerably long lifetime of the nozzle approaching 1,000 hr can be achieved.

The rotational motion of the cathodic arc foot can be obstructed by different factors. The simplest case is when some mechanical damage is inflicted on the nozzle lip, which can easily occur because the material of the standard nozzle is quite soft. The arc foot tends to linger much longer at such a position, causing local overheating followed by increased erosion and end of the service of the nozzle within a considerably short period of time. This issue is why the nozzles should be handled with care, which is required when removing residual LDPE deposits close to the nozzle tip. The application of tools made of a material softer than copper is recommended.

In general, the erosion of the nozzle is very slow, but after 500–700 hr of operation in the diffuse mode, some material is removed from the nozzle lip, causing the arc foot to move around the orifice with varying speed. The places that are swept slower by the arc are more exposed to erosion. After a further hundreds of hours of operation, this process causes a macroscopic asymmetry of the erosion ring. The plasma jet starts to tilt, causing degradation of the process accuracy. Such irregular erosion with an arc operated in CDA is shown in Figures 8 and 9a). The erosion profile is much more regular if nitrogen with an admixture of hydrogen is used as the plasma gas (see Figure 9b).

The nozzle erosion is more than doubled during operation with two parallel‐connected HV generators. The inner diameter of the copper nozzle grows from 4.2 to 5.8 mm within 50 hr of operation. As a consequence, the arc is blown out of the nozzle slightly, causing an increase in the thermal load of the deposited film. Compensation of this effect is needed and can be achieved by a slight increase in the nozzle distance or an increase in the linear motion speed.

The asymmetrical erosion process is sped up when some deposits have built upon the nozzle lip. On the scale of hours, the nozzle heats up, and overspray powder sticks on its surface. Due to the local powder injection, the LDPE film grows faster on one side of the nozzle lip than on the other. The regions coated by the thinner layer of LDPE are more frequently impacted by the arc, which results in asymmetric erosion of the nozzle lip (see the nozzle profile in Figure 8). To avoid this effect, the residual melt is removed from the tip of the nozzle after each coating process, and the azimuthal position of the nozzle is changed every 8 hr of operation.

Even when regularly cleaning the deposits from the nozzle, after approximately 80 hr of operation, the erosion, as shown in Figure 9b), is so pronounced that significant changes in plasma properties can be observed. To reduce the thermal drift of the process and extend the maintenance period, a nozzle with a core made of tungsten, as displayed in Figure 9c, is applied. After 15 hr of operation of such a nozzle, no measurable increase in the orifice’s inner diameter can be observed. Moderate erosion of the nozzle lip after 80 hr of operation is documented in Figure 9d. The width of the eroded ring is 0.63 mm, compared with 1.08 mm for the copper nozzle after the same operation time. Assuming rounding of the nozzle lip with a radius of 0.7 mm, the tungsten nozzle erosion volume is approximately one quarter of the copper nozzle erosion volume. Consequently, the lifetime of the tungsten core nozzle can be expected to be four times longer. The amount of the eroded material in the grown LDPE film is in the range of few ppb.

4. Deposition Process Development

The generic LDPE coating process on an aluminum surface consists of the following steps:

• 1. Plasma pretreatment of the aluminum substrate surface to activate it and increase its temperature close to the LDPE melting point for better adhesion of the first LDPE layer on the metal surface.
• 2. Deposition of the first layer of the LDPE with a small distance from the nozzle to the metal surface to ensure high heat transfer. The smallest distance to avoid arc transfer is applied, that is, 20 mm.
• 3. Deposition of further layers of LDPE with gradually increasing distance from the nozzle to the coating surface to avoid overheating of the already deposited layer. Additionally, the growing thickness of the film is compensated for by increasing the distance between the nozzle and substrate.

Manual operations, such as unloading of the substrate, and removal of the LDPE melt from the masking components and plasma head, contribute to the gross process time. The latter should be done before the LDPE melt solidifies.

4.1 Evaluation of conveyors for LDPE powder

Five different powder feeder configurations for conveying the LDPE power were evaluated. The results of these evaluations are presented below.

4.1.1 Palas RBG‐1000

In the framework of earlier projects, the aerosol generator of Palas GmbH, model RBG‐1000, 28 was successfully applied for dispersing different organic powders, such as polyester powder 29 or adipic acid. 30 Thus, it was selected for the evaluation of conveying of the LDPE powder. The powder to be dispersed is filled into a cylindrical powder reservoir and compressed by a tamper into a type of elongated pill. The powder, which is uniformly compressed along the filling height, is pressed with a piston onto a rotating tightly woven precision stainless steel brush at an exactly controlled feed rate. An adjustable gas volume flow streams over the brush and pulls the particles out of it. As the available LDPE powder cannot be compressed into a type of pill, from which the rotating metal brush can strip the powder, no positive result has been achieved with this operating principle.

4.1.2 Plasmadust conveyor

The first successful film deposition utilized a Plasmadust™ dispersion unit. 31 The schematic structure of this device is displayed in Figure 10a. Before operation, the powder is filled into the powder container. Four nozzles introduce carrier gas pulses into the powder volume, causing dispersion. The gas pulses are controlled by electrically switched valves, opened and closed according to an optimized sequence. The dispersion process can be optionally enhanced by vibrations generated with a pneumatic rotary vibrator. The main disadvantage of the Plasmadust™ dispersion unit is the small volume of the powder container, which is insufficient for more than one substrate.

4.1.3 Powder processing unit with a diaphragm pump

More than one liter of powder can be continuously conveyed by using the powder processing unit (PPU) developed by Reinhausen Plasma GmbH. The operating principle of this system, shown in Figure 10b, is based on a vibrating suction needle. 32 The suction needle is positioned very close to the surface of the powder stored in a cylindrical container. The vibration of the needle causes local dispersion of the powder. The suspension generated in this manner is sucked through the needle using diaphragm pumps. By x–y movement, the needle strips the powder layer by layer from the powder container. The motion speed of the needle is controlled on the basis of the weight changes determined with a precise electronic scale on which the powder container is placed. This method delivers a fair powder rate with satisfactory time stability. However, after longer operation times, some parts of the diaphragm pump become quite hot. This heating was not a problem before when the pumps were used for conveying metal powders, which was the main application of the PPU up to this time. However, the temperature is high enough to cause melting of the LDPE powder, sticking of the LDPE onto the pump components and subsequent clogging of the pump channels.

4.1.4 PPU with a Venturi pump

To avoid problems with diaphragm pumps in the operation of the PPU, different pump types were tested. The pump based on the Venturi principle was found to be an efficient solution. The usage of a Venturi ejector instead of a diaphragm pump for powder suction in the PPU system allows the powder overheating problem to be overcome. Several commercially available Venturi ejectors of J. Schmalz GmbH were tested. Finally, the SEG 05 HS‐S 10.02.01.00272 for 5 bar was selected as the best fit. Unfortunately, the Venturi ejector was found to be sensitive to clogging due to insufficient underpressure in the suction needle. This pressure can be increased by increasing the carrier gas flow, but a consequence is deterioration of the powder utilization efficiency. Another problem was that even though the applied carrier gas was nitrogen, environmental air was sucked together with the powder through the suction needle, causing increased oxidation of the powder and nozzle.

4.1.5 Flowmotion

Finally, a commercial powder feeder called Flowmotion™ from Medicoat AG was evaluated. This system is based on volumetric powder flow control. The powder is continuously fed out of a hopper by an oscillating channel. The amplitude of this oscillation defines the volumetric quantity of powder (cm³/min). This feeder fulfills the main requirements of the process: nitrogen as a carrier gas, a high and a stable powder rate (see Figure 11) and a sufficiently large (2.5 or 5 L) powder reservoir. This dispersion system is the final recommendation concluded from the study.

4.2 Powder injection

The aim of the injection of the powder suspension into the plasma is to allow heat transfer from the plasma to particles. 33 The transfer of energy to the powder, which finally causes its melting, can proceed over different channels. The main mechanism is the heat transfer from the hot gas across the boundary layer by heat diffusion. A large portion of the energy can be provided by the IR (infrared) radiation from the arc. Furthermore, chemical channels must be considered. 34 The chemical radicals produced in the arc approach the particle surface and undergo some chemical reactions, which are typically exothermic and provide thermal energy to the surface. Another type of chemical energy deposited on the powder surface is the recombination heat that is released when two atoms of the same kind combine into a two‐atom molecule using the surface of the particle as a catalyst. Furthermore, the quenching of different types of excitation, especially, vibrational excitation, is a source of energy flow to the powder particle. A small part of the heat is provided by the bombardment of electrically charged particles: electrons and positive and negative ions. A systematic presentation of the atomic and molecular phenomena at the boundary between the plasma and the substrate can be found in the work of Lieberman and Lichtenberg. 35

Different architectures of powder injection into the plasma are used for spraying. 36 An extensive investigation of direct metal powder injection within the output channel of the PAA‐PJ nozzle has been conducted. 37 The design with internal injection proved to be very efficient when working with metal powders. However, due to the elevated temperature of the nozzle operated with a double power supply in combination with the low melting point of LDPE, the powder sticks at the outlet of such an internal injector, causing blockage. The arc in the mouth of the nozzle is so hot that overheating of the powder easily occurs. The geometry of such injection is fixed and consequently does not allow the geometrical settings to be adjusted when gas flow, powder rate or power is varied. For these three reasons, the external injection shown in Figure 12 was chosen instead.

The following considerations were done to evaluate the geometrical constraints of the injection process.

The angle α between the injector axis and the axis of the nozzle has a strong influence on the power transfer from the arc zone plasma to the particles. With decreasing α, the length of the crossing path through the plasma becomes inversely proportional to cos(α), resulting in a longer thermal interaction time and, consequently, stronger heating of the powder. At the same time, the mean temperature of the crossed zone decreases, resulting in decreasing thermal exchange. The best compromise of these tendencies is empirically found at α = 45°. For a visible plasma plume diameter of d_P = 5 mm, the powder covers the distance of approximately s_cross = d_P/cos α ≈ 7 mm across the arc plasma zone. The time of this thermal interaction t_cross can be estimated as: t_cross = s_cross / v_particle,  (3) where v_particle is the mean speed of the particle. According to Equation 3, too high a speed of the powder results in too short an interaction time between the hot plasma zone and the powder, and consequently, in incomplete melting of the particles. Too low a particle speed results in a long thermal interaction with the plasma and consequently, in overheating of the powder, possibly leading to LDPE decomposition. The optimal speed is a compromise between these two tendencies.33 Assuming, that particles entering the plasma have the same speed as the carrier gas coming out of the injector, this speed can be estimated as a function of the carrier gas (nitrogen) flow F_N2. The simplified formula expressing this speed is then: N₂ v_particle = 4/π F_N2/ d²_inj (4) where d_inj is the diameter of the injector circular outlet.

The inner diameter of the injector strongly affects the speed of the injected suspension, which increases with decreasing diameter. For a typical carrier gas flow of 7 SLM (standard litre per minute), the mean gas speed at the outlet of the injector with a 1.5‐mm diameter calculated using Equation 4 is 66 m/s, compared with 9 m/s in the pipeline with an inner diameter of 4 mm. For the injector with a diameter of 1 mm, this speed reaches 150 m/s. Such a high speed of the powder entering the arc zone results in a very short time of thermal interaction between the powder and the hot arc zone, and consequently, a high percentage of nonmolten powder. However, an increase in the injector diameter results in a decrease not only in the suspension speed (37 m/s for a 2‐mm diameter) but also in the poorly focused flow of the suspension. As a consequence, only a small percentage of the powder crosses the central arc zone. Despite the prolonged thermal interaction time, powder utilization is degraded. The injector diameter of 1.5 mm was determined to be the optimum value and was applied for most of the substrate‐coating experiments (see Table 1). Table 1. Process conditions for five process recipes

For the estimation of the suitable carrier gas flow, the interaction time between the particle and the plasma that is sufficient to reach the melting point in the entire volume of the particle must be determined. This time can be estimated as a function of the diameter of a spherical LDPE particle using a simplified one‐dimensional heat diffusion model. In Figure 13, the results of such a simulation are shown. Using the trend line of this diagram, the time required for the melting of a particle with a diameter of 50 µm can be determined as 104 µs. Using Equations 3 and 4 with injector diameter d_inj =1.5 mm, a nitrogen flow of approximately 7 SLM can be calculated.

Commercially available powders consist of particles of different sizes. The statistical distribution of the particle size, mass or volume around the respective mean value can be specified. The characteristics of the statistical distribution of the powder have a direct influence on the quality of the deposited film. Consider an LDPE powder containing not only particles with a diameter of 50 µm but also some particles with much smaller and much larger diameters. Looking at the diagram in Figure 13, particles with diameters larger than 50 µm would clearly need a time longer than 104 µs to become completely molten. Thus, for a 104‐µs interaction time, such large particles will not become completely molten, and particle structures will be observed in the deposited film. For particles smaller than 50 µm, complete melting is reached after a time shorter than 104 µs. Thus, for the interaction time of exactly 104 µs, these particles will reach temperatures much higher than the melting point in parts of their volume. The smaller the particle is, the higher the risk of overheating and thermal decomposition. A significant number of small particles causes inferior quality of the film, manifested by a brown color and poor mechanical properties. To ensure a high film quality, a powder with as narrow a statistical size distribution as possible should be used.

A decrease in the temperature of the plasma gas and consequently, of the droplets restricts the distance between the nozzle and the substrate. Such solidified particles recoil from the substrate and are sucked out of the process chamber by the extraction system or remain on the floor of the chamber. Consequently, the powder utilization ratio degrades. Too small a distance results in local overheating of the already deposited coating by the plasma and deformation of the soft film by the focused gas flow.

The temperature of the gas flowing out of the nozzle depends on the distance to the nozzle tip (see Figure 6), which is why the powder suspension is injected into the plasma plume as close to the nozzle tip as possible. The limiting factor is the increasing risk of arc transfer to the tip of the injector when it is too close to the outlet of the nozzle. Transferring of arcs onto the injector causes overheating, which can cause melting of the powder inside the injector and blockage. In the worst case, damage of the injector by melting of its tip is possible.

To obtain an optimal heat flow to the powder, the actual powder trajectory (see the graphical explanation in Figure 14) should cross the axis of the central arc zone. However, if the injector axis exactly crosses the middle point of the arc zone, then most of the powder is deflected and does not reach the arc (see the uncompensated trajectory in Figure 14). As the speed of the injected suspension is similar to the azimuthal speed component of the plasma gas, the plasma gas flow drags the injected powder stream off‐axis. Consequently, the heat transfer to the powder deteriorates. Compensation of the azimuthal drag can be realized by a small correction of the injector position. The injector has to be approximately 1 mm off the axis of the nozzle to obtain the optimal powder heating.

The optical emission from the powder plasma is helpful for finding the precise optimum position of the injector. The best position results in the brightest light emission from the powder‐arc zone. This type of adjustment is crucial and very specific for nozzles with vortex stabilization of the arc.

4.3 Substrates

Substrates strongly varying in shape were investigated. Structures that are difficult to cover with a continuous LDPE film include thin and deep groves and sharp steps in the topology of a coated surface. The problems that occur due to such formations include recoiling of the droplets at very flat impact angles and shadowing effects. The result of such a shadowing effect is demonstrated in Figure 15. A triangular void, which can be observed between the left wall of the aluminum grove and the LDPE material filling the grove, is created as a result of shadowing by the left side LDPE layer due to the slightly tilted spray direction, as depicted in Figure 15. The cutting plane shown in the photograph was masked by a PEEK surface during LDPE spraying. The smooth variation in the radial position (5–10 mm) of the substrate surface affects the thickness of the coating. Depending on the specified film requirements, this variation can be tolerated or not. Additionally, the systematic variation in the film thickness is present on conical surfaces because approximately the same dwell time of the plasma head and consequently, the same amount of LDPE is distributed over a larger area with increasing radius.

Systematic process development (see Section 4.6) was conducted using rotationally symmetric substrates machined of aluminum alloy. However, for a small number of tests, stainless steel substrates were used. Due to the different thermal properties of stainless steel in comparison with aluminum alloys, an adjustment of the process parameters was needed to realize coating films of the same quality. The main changes were implemented in the plasma pretreatment step, which should result in an increase in the substrate temperature of approximately 100 K. Let us consider substrates machined of AISI‐301 stainless steel weighing 800 g and substrates made of aluminum alloy AlSi9Cu3 with similar volume and coating area. The specific heats of these two materials are 500 and 880 J·kg⁻¹·K⁻¹, respectively. For the required temperature increase, if the different thermal losses are disregarded, 40 and 28 kJ, respectively, should be coupled into the substrates. This result means that the stainless steel substrate requires 43% more thermal energy to reach the same temperature as the aluminum substrate. This value fits the empirically determined increase in the thermal load quite well, which was reached by doubling the net exposure time to the plasma (e.g., two times more passes of the plasma head) with a slightly increased (on the order of 1 mm) distance of the nozzle at the same time.

4.4 Masking

In many cases, some areas of the coated substrate should remain uncoated. In the example considered in this study, the screw thread segment and the bottom side of the substrate should remain uncoated. For this purpose, two masking cylinders made of PEEK, one small (see Figure 2) and one large, were used. PEEK was chosen for this purpose because it withstands such high temperatures as 300°C, which is much higher than the melting point of LDPE. PEEK is also mechanically strong, which makes the removal of superfluous deposits from its surface easier. An advantage of the removal of parasitic coatings is also the poor adhesion of the LDPE melt to PEEK. A photograph of a substrate segment coated with a mask mounted on it is shown in Figure 16.

4.5 Aliasing

For combinations of the linear speed of the plasma head and rotation speed of the aluminum substrate, for which the plasma head repeatedly passes over the same positions, modulation of the film thickness can be observed (see Figure 17b). This effect is most prominent on the outer edge of the workpiece, where the distance between coating traces is larger than the width of the spray stream. Closer to the substrate axis, strong overlapping of the spray streams causes unification of the film. The condition for the occurrence of aliasing patterns is 38 that the time of n_rot full rotations of the substrate is the same as the time of n_move linear back‐and‐forth movements (n_rot and n_move—natural numbers). This condition can be expressed by the following formula: (5) where t_rot and t_move are the rotation time of the substrate and the time of the back‐and‐forth motion of the plasma head, respectively.

The examination of the aliasing pattern seen in Figure 17b shows that the number of aliasing maxima counted on the perimeter of the substrate is 23, corresponding to the angular distance between neighboring maxima of 15°39′ depicted in the figure. The time for the back‐and‐forth motion of the plasma head according to the data from Table 1, recipe 3, is twice the shift of 75 mm divided by the speed of 210 mm/s, giving t_move = 0.68 s. However, assuming that the aliasing pattern was created during only one rotation, the time between aliasing maxima for the rotation speed of 12 rev/min (t_rot = 5 s) would be t_rot/n_move = 5 s/23 = 0.22 s, which is much less than the calculated t_move. More plausible is the assumption that n_move = 23 back‐and‐forth motions occur during n_rot = 4 rotations. In such a case, according to Equation 5, each fourth aliasing maximum would be sprayed in a time interval of t_move = 0.68 s. The actual interval for back‐and‐forth motion of 0.88 s, 30% larger than the calculated 0.68 s, is realistic because the setting of the plasma head speed (in our case, 210 mm/s) is significantly higher than the effective speed. The reason for this discrepancy is the acceleration and deceleration of the plasma head at the turning points. Therefore, the actual duration of the back‐and‐forth movement of the plasma head is longer than the estimated one.

The aliasing effect can be minimized when the distance between the aliasing maxima is significantly smaller than the width of the spray stream. By the appropriate choice of t_rot and t_move, the coating result shown in Figure 17a can be achieved. The risk of aliasing can be completely avoided if fast, variable speed rotation of the substrate and slow stop‐and‐go motion of the plasma head are implemented. However, hardware allowing such a motion scheme was not available for the reported experiments.

4.6 Process parameter optimization

Preliminary coating processes were conducted on aluminum plates. For powder supply, the Plasmadust™ dispersion system described in Section 4.1 was used. The experiments with LDPE spray directed on a cold aluminum plate resulted in no adhesion of the melt, independent of all the other process conditions. As heating of the substrate improves the distribution of molten polymer droplets, 39 a heat plate with controlled temperature was used to preheat the aluminum alloy plates. Later, plasma preheating was applied. If the substrate surface is heated but the temperature is far from the melting point, then, the adhesion of LDPE is poor. The deposit growing on such a plate is shown in Figure 18. The small separate balls seen in Figure 18b, corresponding in size to the LDPE powder particle diameter, are not wetting the metal surface. A tendency toward the growth of clusters with the uncoated metal surface between them is also observed (black areas in Figure 18). Such clustered deposition can result in more than 1‐mm high columnar structures and macroscopic voids in the film. The reason for such growth dynamics is the much better adhesion of the molten droplets to the surface of already deposited LDPE. As LDPE has a much lower heat conductivity (≈0.33 W·m⁻¹·K⁻¹) than the aluminum alloy (≈150 W·m⁻¹·K⁻¹), the droplets touching LDPE remain liquid much longer and can spread over the already solidified LDPE surface much better than over the cold aluminum alloy. Preheating of the plates up to approximately 120°C improves the wetting of the aluminum surface with liquid LDPE and the growth of homogeneous films.

In these first successful experiments, the surfaces of the aluminum plates were pretreated with the forming gas 95/5 plasma in addition to the hot‐plate prewarming. The shear force test was conducted according to DIN EN 1465. 40 Pairs of plates with sizes of 3 × 25 × 100 mm made of AlCuMg2 aluminum alloy were used. The 10 × 25‐mm large surface at one end of each plate was sprayed with 1‐mm thick LDPE. The parameters for the spraying were a single HV pulse generator with a pulse frequency of 54 kHz, a distance from the nozzle to the substrate—of 12 mm, a substrate heater temperature of 160°C, a plasma jet speed of 8 mm/s, a gas flow of 56 SLM with 2.5% hydrogen, and powder conveyed by a Plasmadust disperser with a carrier gas (nitrogen) flow of 6 SLM. The plasma pretreatment was conducted twice with a speed of 50 mm/s. Two plates were clamped together with the coated surfaces facing each other and heated in a Nabertherm C250 muffle furnace up to 145°C to create a connection. The shear force was determined using a Zwick Z010 universal tensile‐testing machine. The mean value determined for the six‐plate couples was 2,892 N (bond strength of 11.6 N/mm²). The maximum force as a function of displacement was reached at 1.8 mm.

The process time and powder utilization ratio evolution over the course of process development for five process recipes are shown in Figure 19. The corresponding process parameters are summarized in Table 1. The utilization ratio of LDPE powder depicted on the right axis in the diagram is defined as the ratio of the mass of a film deposited on an aluminum substrate to the gross mass of powder used for the production of this film. The film mass was gravimetrically determined for each substrate by weighing the substrate before and after coating. The mass of the used powder was controlled by weighing the powder added into the tank of the powder feeder.

The starting point of development was the conventional configuration of plasma generation commonly used for high‐speed surface activation. 41, 42 The so‐called hot nozzle was applied, which allows extension of the arc out of the nozzle. A single power supply unit was used. The main difference in comparison with activation processes was the utilization of a nitrogen–hydrogen gas mixture (5% hydrogen) instead of CDA as the plasma gas. This recipe 1 of the substrate‐coating process (see the parameters in Table 1) was based on the PPU with a vibrating suction needle connected to a Venturi pump, as described in Section 4.1.

The recipe 2 process improvement was mainly due to doubling of the power by adding a second HV pulse generator. To avoid local overheating of the film, both the substrate rotation speed and the plasma head linear motion speed were more than doubled. Additionally, an optimized injector with a diameter of 1.5 mm was introduced. These improvements resulted in a reduction in the process time by a factor of two and more than doubling of the powder utilization ratio (see Figure 19).

Although the PPU worked properly and allowed deposition of films on the substrates, its powder transfer ratio was strongly limited. The minimum nitrogen flow through the Venturi pump that avoided clogging of the suction line was 7 SLM. This gas flow mixes with the air and powder coming from the suction needle. The amount of this air was not precisely determined. Consequently, a large amount of cold gas was injected into the plasma plume, causing cooling, thus, resulting in a reduction in the powder utilization ratio. The other drawback was that the carrier gas included environmental air sucked from the surface of the powder tray, causing stronger oxidation of the LDPE powder and the nozzle. Evidence of this process includes the slight brownish coloration of the film deposited using this hardware configuration. The application of the Flowmotion™ powder feeder (process recipe 3) resulted in a reduction in the process time by a factor of almost 2 (see Figure 19). At the same time, the powder utilization ratio increased from 25% to 29%.

The improvements in the process conducted according to recipe 4 are based on the adjustment of the vibration frequency of the Flowmotion™ powder feeder. For earlier processes, the oscillation frequency of 63 Hz delivered satisfactory powder flow. However, the resonance frequency of the oscillating channel loaded with LDPE powder is over 120 kHz, which means that the oscillator worked with the second harmonics instead of the first harmonics. The final optimized oscillation frequency of 123.2 Hz allowed a much higher powder conveying rate to be reached. With this oscillation frequency, not only was the powder rate increased but also, the flow was more stable, allowing the flow of carrier gas (nitrogen) to be reduced from 9 down to 7 SLM. This recipe allowed a reduction of the total gas flow by 10% (from 60 to 54 SLM) without increasing the plasma gas dilution. With decreasing plasma gas flow, the temperature in the plasma plume increases, as shown in Figure 6. This increase compensates for the increase in the amount of conveyed powder, resulting in no significant decrease in the powder utilization ratio (see the powder utilization curve in Figure 19).

The last essential process improvement (process time shortened to 7 min and powder utilization ratio increased to 65%) was achieved by careful programming of the plasma jet motion, allowing a loss of coating material on the masking components and dead times when the powder is conveyed but no coating occurs to be avoided. The main change in process recipe 5 was the shortening of the linear motion from 75 to 47 mm. The number of coating steps was increased from two to three to better adopt the transferred heat to the growing film thickness. The application of the nozzle with a tungsten core instead of copper allowed for high time stability of the process due to the significantly reduced erosion of the nozzle.

An important outcome of this study is the establishment of the process configuration and process conditions that allow the fulfillment of the time requirements of the customer. The conditions are summarized as recipe 5 in Table 1. A process with very good reproducibility has been achieved. The mean square variation of <2.7% in the film weight on 140 substrates is caused mainly by manual removal of the masking components. The failure rate of the process conducted according to recipe 5 is below 0.7% (140 substrates were coated without failure). From the net LDPE film weight of 11 g, assuming the mass density ρ_LDPE = 930 kg/m³, the average film thickness of 1.02 mm can be calculated. For a net spraying time of 270 s, a net deposition rate of 2.5 g/min is obtained. Including the material deposited on the PEEK holder, the gross deposition rate of 4.1 g/min can be calculated.

5. Conculsion

As demonstrated, a PAA‐PJ can be used for low‐temperature spraying of LDPE on a preheated aluminum surface with a gross deposition rate in the range of 4.1 g/min, a powder utilization ratio of 65% and a failure rate below 0.7%. Three powder conveyors were evaluated for 50 µm LDPE powder, and the most suitable was selected for tests with a large number of substrates. Due to plasma pre-treatment of the aluminum surface, adhesion of the films has been achieved, showing only cohesive failure by shear tests. Without plasma treatment, adhesive failures frequently occurred.

In the course of process development, the optimized set of process parameters was found. The best results were achieved with a 1500‐W electric power (two PS2000 OEM HV pulse generators connected in parallel operated at 63 kHz) and with a gas mixture of 96.7% nitrogen and 3.3% hydrogen at a total gas flow of 54 SLM. The frequency of the Flowmotion™ oscillator for D50 = 50 µm LDPE powder was 123.2 Hz. The optimal carrier gas flow at this frequency for a 1.5‐mm injector was 7.2 SLM of nitrogen.

Although a coating process with attractive specifications has been developed, further improvements can be achieved when using different substrate motion schemes, for example, the substrate could be rotated much faster with varied speed to achieve a constant azimuthal treatment speed, and the plasma head could follow an arbitrary contour of the substrate.

Before transferring the generic process to a fully automated system, several issues must be considered in more detail. Examples include the automated loading and unloading of the substrates and masks, removal of the surplus material, management of the overspray or automated cleaning of the plasma nozzle. Solutions for these problems could be proposed but are outside the scope of this development study.

The presented method was applied not only for LDPE deposition on aluminum but also for PEEK, PVC, and PTFE coatings. Due to the short time‐frame available for these feasibility studies, the achieved results were not satisfactory.

Acknowledgments

The author thanks Rolf Kuhn from Medicoat AG for efficient support during evaluation, installation and operation of the Flowmotion™ powder feeder. Thanks also to Bernd Grundmann for all the kinds of mechanical machining in the course of the study. The coating processes with the PPU powder feeder were conducted by Raul Schmidt. Alexander Wiedemann contributed substantially to the development and characterization of the Venturi version of the PPU.

References

1. T. Bezigian, Handbook of Industrial Polyethylene and Technology, John Wiley & Sons, Hoboken, NJ 2017, p. 429. Wiley Online Library Google Scholar

2. F. Fanelli, R. d’Agostino, F. Fracassi, Advanced Plasma Technology, Wiley‐VCH Verlag GmbH & Ca. KGaA, Weinheim, Germany 2008, p. 353. Google Scholar

3. L. Pawlowski, The Science and Engineering of Thermal Spray Coatings, John Wiley & Sons Ltd., Chichester, England 2008, p. 560. Google Scholar

4. P. Fauchais, A. Vardelle, Advanced Plasma Spray Applications, Intech, Rijeka, Croatia 2012, pp. 1– 38. Google Scholar

5. P. Fauchais, J. Phys. D: Appl. Phys. 2004, 37, R86. Crossref CAS Web of Science®Google Scholar

6. Z. Dong, K. Khor, Y. Gu, Surf. Coat. Technol. 1999, 114, 181. Crossref CAS Web of Science®Google Scholar

7. J. Winter, R. Brandenburg, K.‐D. Weltmann, Plasma Sources Sci. Technol. 2015, 24, 064001. Crossref CAS Web of Science®Google Scholar

8. L. M. Wallenhorst, S. Dahle, M. Vovk, L. Wurlitzer, L. Loewenthal, N. Mainusch, C. Gerhard, W. Viöl, Condensed Matter Phys. 2015, 2015, 930. Google Scholar

9. L. Wallenhorst, L. Gurău, A. Gellerich, H. Militz, G. Ohms, W. Viöl, Appl. Surf. Sci. 2018, 434, 1183. Crossref CAS Web of Science®Google Scholar

10. R. Köhler, G. Ohms, H. Militz, W. Viöl, Coatings, MDPI 2018, 8, 8090312. Google Scholar

11. Powder Technology Handbook (Eds: H. Masuda, K. Higashitani, H. Yoshida), Taylor & Francis, New York, NY 2006, pp. 3– 93. Google Scholar

12. R. Molz, D. Hawley, in Proc. Int. Thermal Spray Conf. (Eds: B. R. Marple, M. M. Hyland, Y.‐C. Lau, C.‐J. Li, R. S. Lima, G. Montavon), Material Park, OH 2007, 6. Google Scholar

13. K. Gaska, X. Xu, S. Gubanski, R. Kádár, Polymers 2017, 9, 4. Google Scholar

14. relyon plasma GmbH, Operating instructions—Plasma generator PG31, https://www.relyon‐plasma.com/wp‐content/uploads/2016/05/plasma‐generator‐pg31‐manual‐EN_F0298601.pdf, 2014, (accessed: April 2019). Google Scholar

15. D. Burger, MSc Thesis, Regensburg University of Applied Sciences (Regensburg, Germany) 2011. Google Scholar

16. X. Lu, M. Laroussi, V. Puech, Plasma Sources Sci. Technol. 2012, 21, 034005. Crossref CAS Web of Science®Google Scholar

17. F. Fanelli, F. Fracassi, Surf. Coat. Technol. 2017, 322, 174. Crossref CAS Web of Science®Google Scholar

18. E. Pfender, Gaseous Electronics Vol. I. Electrical Discharges, Academic Press, New York, NY 1978, pp. 291– 398. Google Scholar

19. E. Pfender, Thermal Plasma Applications in Materials and Metallurgical Processing, Academic Press, 1992, pp. 13– 30. Google Scholar

20. R. C. Eschenbach, G. M. Skinner, Development of stable, high power, high pressure arc air heaters for hypersonic wind tunnel, Speedway Research Laboratory, Linde Company, Division of Union Carbide Corporation, Aeronautical Systems Division, Technical Report WADD, Technical Report 61–100, 1961. Google Scholar

21. W. S. Brzozowski, Z. Celinski, Bull. Acad. Pol. Sci., Ser. Sci. Tech. 1962, 10, 1. Google Scholar

22. relyon plasma GmbH, Operating instructions PS2000 power supply, https://www.relyon‐plasma.com/wp‐content/uploads/2017/07/ps2000‐power‐supply‐manual‐EN_F0307802.pdf (accessed: April 2019). Google Scholar

23. M. Goldman, A. Goldman, Gasous Electronics, Electrical Discharges, Vol. I, Academic Press, New York 1978, pp. 219– 290. Google Scholar

24. M. Hoffmann, MSc Thesis, Fernuniversität Hagen (Hagen, Germany), 2016. Google Scholar

25. D. Korzec, D. Burger, S. Nettesheim, adhäsion KLEBEN DICHTEN 2015, 59, 26. Crossref CAS Google Scholar

26. J. Rager, PhD Thesis, University of Saarland (Saarbrücken, Germany) 2006. Google Scholar

27. F. Hoppenthaler, Abschlussbericht Plasmaerzeuger PB3, Reinhausen Plasma GmbH Technical Report AP010, 2013. Google Scholar

28. Palas, RBG series powder disperser for extremely low and medium mass flows, 2014, https://www.palas.de/en/product/rbg1000, Product Data Sheet V0020811 (accessed: October 2019). Google Scholar

29. R. Köhler, P. Sauerbier, H. Militz, W. Viöl, Coatings, MDPI 2017, 7, 7100171. Google Scholar

30. D. Korzec, S. Nettesheim, A. Ammon, ak‐adp Jubiläums‐Band, Arbeitskreis Atmosphärenduckplasma—ak‐adp, Jena, Germany 2019, pp. 1– 9. Google Scholar

31. Reinhausen Plasma GmbH, Plasmadust Dispergierer. Betriebsanleitung, 2011, http://www.reinhausen‐plasma.com, Product Data Sheet V0020811 (accessed: June 2014). Google Scholar

32. S. Nettesheim, US Patent 9327919 B2, 2016. Google Scholar

33. R. B. Heimann, Plasma spray coating: Principles and applications, Wiley‐VCH Verlag, Weinheim, Germany 2008, p. 79. Google Scholar

34. A. Fridman, Plasma Chemistry, Cambridge University Press, New York, NY 2008, p. 425. Google Scholar

35. M. A. Lieberman, A. J. Lichtenberg, Principles of Plasma Discharges and Materials Processing, John Wiley & Sons, Inc., Hoboken, NJ 2005, p. 425. Google Scholar

36. M. Vardelle, A. Vardelle, P. Fauchais, K.‐I. Li, B. Dussoubs, N. J. Themelis, J. Therm. Spray Technol. 2001, 10, 267. Crossref CAS Web of Science®Google Scholar

37. M. Szulz, MSc Thesis, Silesian University of Technology (Gliwice, Poland) 2011. Google Scholar

38. A. V. Oppenheim, A. S. Willsky, Signals and Systems, Prentice‐Hall International Limited, London, UK 1997, p. 527. Google Scholar

39. M. Ivosevic, R. A. Cairncross, R. Knight, Int. J. Heat Mass Transfer. 2006, 49, 3285. Crossref Web of Science®Google Scholar

40. Adhesives—Determination of tensile lap‐shear strength of bonded assemblies, 2009, https://www.beuth.de/de/norm/din‐en‐1465/115724482 (accessed: October 2019). Google Scholar

41. M. Noeske, J. Degenhardt, S. Strudthoff, U. Lommatzsch, Int. J. Adhesion Adhesives 2004, 24, 171. Crossref CAS Web of Science®Google Scholar

42. M. Kim, D. Song, H. Shin, S.‐H. Baeg, G. Kim, J.‐H. Boo, J. Han, S. Yang, Surf. Coat. Technol. 2003, 171, 312. Crossref CAS Web of Science®Google Scholar