Materials degradation in non-thermal plasma generators by corona discharge

Power–time measurements

Tests are performed to determine the behaviour of different corona electrodes materials exposed, during relevant amounts of time, to positive corona discharge at high voltages typical from real operation. Under the corona discharge effect, the thermophysical properties of electrode materials play an essential role in erosion. Tungsten (W), nichrome (NiCr) and stainless steel (SS) are evaluated as potentially interesting materials to develop the experiments, because they are electrically conductive, have good corrosion resistance and are commercially available. In the experiments presented, the wires used as corona electrodes have the 40 cm in length and the diameters vary between 25, 50 or 100 µm. Current–voltage characteristics curves are determined for different times of exposure at a fix voltage. This voltage typically depends on the geometric parameters of the device: distance between electrodes, wire diameter and configuration. As time increases, the current decreases; as a result, higher values for operating voltages are required, which reduces the performance and life span of the devices.

Based on the different values obtained by the CVCs curves the power consumption for the different materials is determined. As the exposure time to the corona discharge effect is increased, the corona power decreases. The experiments stop when the break point of the wire is reached or when the number of sparks forms a continuous stream that creates a short circuit. In the case of 25 μm wires, the applied voltage is 4.7 kV and a current of 0.2–0.25 mA. Meanwhile, the 50- and 100-micron wires are subjected to 5 and 5.4 kV of voltage, respectively, and 0.2–0.25 mA of corona current. All filaments are tested from 1 to 1.5 W of initial power.

To evaluate the different materials used as corona electrodes, a first round of experiments is performed using a 50-micron wire diameter for all of them. All curves represented in Fig. 2 are an average of the tests performed for each wire, in order to analyse power loss with increasing exposure of the devices to the corona discharge. For the 50 µm diameter wires both experiments of stainless steel and nichrome are stopped, as they break at 50 and 95 h, respectively. Tungsten wire does not break but the streamers formed make impossible the study when the 200 h are reached. We can observe how Ni–Cr and W wires have similar behaviour for the first 70 h which they only arrive at a power loss of 20%. After that point we can observe a drop of power for nichrome wire in its last 30 h of lifetime.

Stainless steel is considered to have a bad response to corona power consumption, as it breaks soon with a high-power drop. The misconduct of the nichrome and stainless-steel wires during the exposure of the discharge can be related to the boiling point of the alloys that constitute the material, which have low boiling temperatures with respect to the temperature at which the plasma is generated (< 10,000 K). Comparing all different materials for the same diameter, tungsten wire exhibits a better performance, as the wire does not break during any of the performed experiments and because it leads to the lowest values for the corona power reduction. Tungsten has a ceramic character that allows working at high temperatures, good resistance to aggressive environments and good mechanical performance. Therefore, tungsten is a proficient material to be used as an electrode. Subsequently, a second round of experiments is performed with tungsten wires and with varying diameters of 25 and 100 microns, in order to determine the useful life of these electrodes for its industrialization. As shown in Fig. 2, the tungsten with different diameter sizes presents a similar power drop around 60 h of use for all studies. Regarding the 25-micron wire diameter, when its power drops by up to 40%, it stabilizes for up to 450 h, when continuous arcs or streamers begin to form. On the other hand, the 50- and 100-micron corona electrodes stabilize after an exposure of about 200 h. In the case of the 100-micron corona electrode, the tests are stopped from 600 h of exposure of the discharge due to the generation of continuous electric arcs. Meanwhile, the 50-micron wire can overtake up to service times exceeding 1000 h. In the last two cases, the tests are stopped with different exposure times of the discharge to visualize these filaments by scanning electron microscopy.

Degradation of W, stainless steel 302 and nichrome corona wires electrodes: erosion rate (E)

Anode degradation is investigated in positive corona discharge in atmospheric air. The micro-craters produced by the ecton generation are related to the mechanism of erosion and oxidation in corona filaments. Investigation of anode degradation rate is developed for several different configurations shown in Table 2. Lifespan testing initiates with new W, Ni–Cr, and stainless steel (S30200) filaments of 50 µm. Nichrome and stainless-steel wires show an increase in electrical resistance in a few hours of discharge exposure, causing breakage or streamer formation. The generation of a uniform oxide layer is macroscopically visible in both previous materials with an ash gray color in the initial 40 h of testing. Regarding tungsten wire, in the first 80 h there is no appreciable change for its operation. However, after this period, the resistance of the filament begins to increase slowly until stabilizing to 200 h of procedure, when the gray-whitish generated oxidation layer is already visible. This behavior of the tungsten filament is verified for other sizes of 25 and 100 µm, respectively. Following the similar trend, an alteration in electrical resistance occurs within the first few hours but stabilizes over time until the electrical discharge is not uniform. For the TRAID configuration, it can be determined that the service life of tungsten filaments exceeds 1000 h but with significant damage due to the corrosion mechanisms that these wires develop.

The erosion rate is a difficult parameter to estimate because numerous phenomena overlap during corona discharge. Several authors such as Islamov and Krishtafovich²⁷ estimated the erosion rate of various materials in their configuration. They related to this physicochemical phenomenon with the volume of filament subjected to discharge, the intensity, and the time of exposure. The average erosion rate formula e_a for a wire is expressed in μg/C and is shown below:

where E is the global rate of erosion as a function of Q(t) which is the charge transferred per unit length of the cable in C/cm. The charge transferred per unit length is defined as (mathrm{Q}left(mathrm{t}right)={int }_{0}^{mathrm{t}}mathrm{I dt}), where I is the current of the corona filament per unit length in µA/cm. On the other hand, A_d is the association with the cross section and the density of the electrode material, A_d [µg/cm] = 10⁶ · (uppi cdot ) (r [cm])² · (uprho ) [g/cm³]; R₀ and R_a are, respectively, the initial and final values of the resistance per unit length of the electrode in the active exposure zone of the corona discharge. Estimating the erosion caused during the corona discharge process is a complicated method of determining when it depends on time. By improving the accuracy of this degradation rate, successive measurements can be used. It is considered that the overall erosion rate is proportional to the derivative of the corona electrode resistence R₀/R in relation to the minority transferred charge Q.

As previously discussed, the erosion phenomenon of the corona anode is one of the main reasons why filament rupture occurs. In Fig. 3, the worst degradation behavior of stainless steel is determined, presenting a rupture at Q = 3 C/cm and an average erosion rate of 29 μg/C. A similar behavior is shown by the nichrome electrode which has a limited useful life, breaking at a transferred charge of 16 C/cm. However, the tungsten electrode has an average erosion rate 1.9 μg/C that remains unaffected with the exposure time, being the lowest of all electrodes.

The erosion rate is approximated exponentially for all the cables studied. An exponential regression and interpolation model of all the data obtained by the data logger is created. The erosion rate for 50 µm tungsten, nichrome and stainless-steel filaments is determined considering the model fit errors and a coefficient of determination R² that is in a range from 0.61 to 0.88.

The results of the average erosion rate for the different W wires studied dependent on the charge transfer are shown in Fig. 4. It is observed that an average erosion rate for the W filament of 25 µm remains approximately 3.7 µg/C. On the other hand, the 50- and 100-µm anodes, unpredictably, do not have a similar average erosion rate in most of the tests performed. The 50-micron wire is remarkable because it has a low erosion rate compared to the other two wire sizes, being from 1.9 to 0.9 µg/C. Instead, the erosion rate performance for the three different diameters of tungsten shows a stable and constant trend when a charge transfer of about ~ 25 C/cm is exceeded. However, the erosion behavior of tungsten shows an appreciable difference with our electrode arrangement being higher than that reflected in the literature. These results may be associated with the prototype geometry and the different applied electrical parameters. For the TRAID configuration it is estimated that it has an erosion rate exposed below:

In general, it is determined that nichrome and stainless steel are not competitive electrodes for use in non-thermal plasma in atmospheric air (for corona discharge applications) due to their accelerated corrosion and their limited-service life. On the other hand, tungsten is still the best material due to its refractory behavior and good resistance to degradation. However, due to the aggressive environment where the discharge has effect, the erosion and oxidation mechanisms are the main dominant phenomena that produce an electrode lifespan of 1630 h without cleaning and maintenance.

Accelerated degradation under corona discharge phenomena

While studying the degradation of electrodes under corona discharge, it has been identified time as a critical factor. The amount of time needed to obtain consistent values for service life in electrodes could require years of tests. Due to this, the importance of accelerating the amount of damage generated on electrodes while being tested and being able to quantify that damage has been one of our goals.

Fourteen 50-micron filaments have been subjected to accelerated degradation tests using a voltage applied of 5.3 kV and a current between 0.35 and 0.4 mA. The acceleration of time damage on electrodes has been done by calculating the cumulative energy density that the electrode handles in MJ/cm². By using that parameter, we can on one side, obtain typical values of cumulative energy density in which the electrode breaks and on the other side set a strategy to increase the energy density the electrode handles in each time step to reduce testing length.

The selection of energy density has been made since this energy has been proved to be the driving force of all the chemical reactions and therefore it directly influences service life³⁰. Energy density has been chosen instead of energy because it’s easier to scale it to different kind of electrodes by taking its surface as reference. Also, the exposed surface is the place where most of the reactions take place in between chemical elements, molecules, electrodes, etc.

The steps to calculate the energy density are shown in Eqs. (4), (5) and (6).

where W refers to power in [W], I refers to current in [mA], V refers to Voltage in kV, S_c refers to the electrode surface in cm², ρ_W refers to power density in W/cm² and ρ_E refers to energy density in MJ/cm².

Integrating Eq. (6), for the experiments done on the degradation benches, the results shown in Fig. 5 are obtained.

These test benches have been studied at a higher service power to obtain the limit of accumulated energy density that an electrode can handle. At approximately 1550 h, the maximum value obtained is 166 MJ/cm².

With that cumulative energy density result and the power loss on electrodes we can give a rough service life estimation at designed operating condition for electrodes³¹. This result has been calculated for several power densities that can be seen in Fig. 6 (in this case, we use linear power density in W/cm for design purposes).

The horizontal red line represents the current cumulative energy density limit which grows while the time the test electrodes are exposed to high power grows. Right now, the estimated service life for electrodes at average power is 4980 h. At minimum power it can be increased up to 10,000 h and at maximum power it decreases to 3300 h.

To improve the reliability of this model and for being able to obtain a higher cumulative energy density limit in a shorter period, it is planned to accelerate even more the test by increasing the energy that the electrode handles. Several ways to increase that energy have been studied³². The easiest one is increasing the voltage in between electrodes but it can make the discharge less stable. That is why in the future tests will be done in other atmosphere with higher temperature and lower pressure to increase the energy density while keeping the discharge stable.

The aim of increasing the energy density is to reach an acceleration factor of more than 10 (predicted hours/tested hour), being this number right now is around 3.2.

Surface morphology: SEM analysis before and after exposure corona discharge

Corona electrodes are analyzed using scanning electron microscope (SEM) images. The study of different diameters of tungsten wires are observed over the central length exposed to the corona discharge. First, the diameter tungsten wire of 25 µm for different hours is visualized obtaining images like the ones shown in Fig. 7. The advance of oxidation is observed, according to the formation of plasma over time in a new filament (see Fig. 7a), beginning to originate oxides on the surface not uniformly after 20 h (Fig. 7b).

Considering a longer exposure time, reaching up to 400 h of use, an irregular oxidative film based on spherical and fractal structures can be seen, mostly in the ionization zone, as shown in Fig. 7c and d. This corrosive phenomenon causes significant changes in the discharge as well as in the direction and intensity of the electric field due to the presence of this film, causing a smaller effective discharge area. Therefore, electric arcs and streamed are created and are tried to go through the corroded film. It is important to consider that although this continuous formation of oxides due to erosion and the oxidative medium where it occurs is quite unstable and most of the oxides volatilize.

The tungsten anode of 50 and 100 µm after corona discharge is shown in Fig. 8. An oxide layer with a rough surface form on the 50 µm filament after 250 h of discharge over the entire surface of the anode (Fig. 8a,b). Two significant regions are shown: on the one hand, areas with stable oxide layers around the anode; on the other hand, sections affected by the impact of an arc which has caused the rust shell to break and local fusion in that area.

On the exposed surface in front of the ground electrode, nanometric craters are visualized at 40 and 185 h (see Fig. 8c,d). Around these micro-explosions, nanometric oxides begin to form, overlapping several effects: local temperature changes due to the impact of electrons with temperatures above 10,000 K, atmospheric air as a corona discharge medium that induces accelerated oxidation and the application of a high electric field that directly influences the degradation mechanism. An exposure time of 470 h is evaluated in the SEM images (Fig. 8e,f) in which two significant regions are also observed where the oxide presents a uniform velvet pattern and fused areas that cause discontinuities in the discharge.

In addition to the oxidation film created, electron bombardment-induced erosion results in significant mass loss over time. It can be clearly seen how the wire in a certain area experiences a considerable decrease in the cross-sectional area (Fig. 9). This could be because of combination of the localized losses of oxide layers, due to microarcs produced, and plastic deformation promoted by the operating temperature during these phenomena. The Ductile–Brittle Transition Temperature (DBTT) of tungsten is in the interval of 200–450 °C^33,34, but it depends on metallurgical factors that can do it to change. Plastic deformation in tungsten only is possible if the temperature is over this DBTT. Therefore, the reached temperature in these areas has been higher than the transition temperature. This change in dimensions directly influences the operation of the discharge, this area being a critical point of electrode breakage. The presence of longitudinal cracks can be caused by the wire manufacturing process and the excessive tension applied to fix the filament on the prototype. These structural imperfections also influence corrosion, accelerating the entire process.