Study of plasma induced nanostructure formation and surface morphology changes on tungsten and stainless steel at atmospheric pressure

Atmospheric pressure glows have been the subject of much interest in recent years. Such discharges are formed when high voltage is applied to a pin to plane electrode geometry usually in the presence of a helium fill gas. The discharge is typically low current (∼<500 mA at voltages less than 5 kV).¹ Although the discharge is generated at atmospheric pressure, its overall structure takes the form of a low-pressure glow discharge. The resulting atmospheric pressure glow does not transition into an arc largely because of the high ionization potential and high thermal conductivity of helium in addition to intentional resistive ballasting. This type of discharge is of interest because of its potential for material processing applications at atmospheric pressure, thus eliminating the need for pumps. Additionally, it has been observed that under certain conditions, the plasma attachment in these glows can self-organize into distinct patterns at the anode surface. These patterns consist of discrete packets of plasma, which have also been observed to move together as a collective, typically exhibiting rotation about an axis of symmetry.

As part of an ongoing effort to have a better understanding on the physical origin of the self-organization and accompanying pattern motion, the sensitivity of the discharge attachment response to the surface composition of the anode electrode was investigated. Here, the study aims to determine the effect of metal anode thermal conductivity and work function on discharge character. Significant surface nanotexturing was observed after plasma exposure; that is, in the general vicinity of the plasma attachment, organized, nanoscale-sized structures were observed. While a number of different metals were investigated, stainless steel and tungsten electrodes are the two metals that are highlighted here largely because they both exhibited the most pronounced nanotexturing. Tungsten has an order of magnitude higher thermal conductivity than steel, but they both share similar work functions. The melting point of tungsten is roughly twice that of stainless steel. Presented here are observations of nanostructure on an anode electrode in an atmospheric direct current (DC) glow discharge. While such structures are of interest for a variety of applications ranging from anti-icing, anti-fouling, optical detectors to catalysis, the formation of nanostructures on plasma-facing surfaces in tokamaks is actually a problem. It has been shown that various nanoscale structures can grow on plasma-facing surfaces when the helium ion flux exceeds 10²² s⁻¹ m⁻² at surface temperatures of order 1000 K and at helium ion energies of at least 60 eV. The problem with the appearance of such fibrous structures, dubbed tungsten fuzz, is that it could potentially flake and by doing so serve as a high Z dust thereby polluting the fusion plasma. Bremsstrahlung radiation amplification owing to the introduction of tungsten in the plasma would make it difficult to achieve ignition. Additionally, it has been shown that the nanostructure could greatly affect hydrogen retention and the thermo-optical properties of the surface in question.² 

The actual formation mechanism of the tungsten fuzz is still not well understood. Computational models suggest that implanted helium produces bubbles in the bulk material. These bubbles coalesce over time forming larger ones and then eventually penetrate the surface forming pinholes and ultimately over time the fibrous nanoscale structure.³ A key to all models, however, is the importance of substrate temperature and minimum ion energy. Research findings suggest helium penetration at energies as low as 10 eV.⁴ It is interesting to note that the nanotextured tungsten observed in this present work is morphologically quite similar to that produced in the fusion simulator experiments over similar exposure times (∼1–10 h).^5,6 This observation is somewhat remarkable, which suggests that despite the significant differences in exposure conditions, the physical processes at play may be similar eventually leading to the same type of nanostructuring.

In Sec. II, the experimental setup is described along with an overview of basic discharge operating conditions. Following this description, initial observations of the formation of nanostructures on the tungsten and stainless steel anode surface are presented. These observations motivated the execution of experiments aimed at time resolving the formation of nanostructures under controlled plasma conditions. Basic calculations aimed at estimating ion fluence, discharge current density, and surface temperature are also given in an attempt to understand the observed similarities between these structures and those observed to form under fusion plasma-wall simulated conditions.

A schematic of the atmospheric pressure DC glow is shown in Fig. 1(a). The apparatus consists of a 6 mm diameter brass tube with a 0.5 mm diameter channel for gas injection. The anode was affixed to a water-cooled heat exchanger whose surface temperature was set at approximately 20 °C. A 0–3 kV DC high voltage power supply (Gamma High Voltage Research Inc., RR3-333R) with a current range of 0–400 mA is used to sustain the discharge. The discharge current was monitored using a precision multi-meter (Fluke 115) to measure the voltage across a 100 Ω sense resistor. The discharge voltage was read directly from the power supply front panel. The cathode electrode was attached to a Vexta (Model PK245M-01AA) stepping motor driven motion control system (Velmex, Inc., VXM-3), which was used to adjust the interelectrode gap. The cathode was supplied with helium via a mass flow controller (Alicat Scientific, MC-10SLPM-D/5M). For this work, the flow rates were fixed at 200 SCCM. In all the cases, the experiments took place in ambient air at atmospheric pressure. In one test case, the discharge was operated without helium gas flow for elucidating some of the mechanisms of the observed surface nanostructuring. The interelectrode gap used during the nanostructuring experiments was tested up to 12 mm base on each specific experimental purpose. The Reynolds numbers, calculated for a fixed flow of 200 SCCM of helium for the discharge gap lengths from 1 to 12 mm, respectively, ranged between 143 and 1715. This result indicates that the flow type is laminar since the Reynolds numbers are less than 2000.^7,8 The temperature of the anode surface was assessed using an optical pyrometer (Mikron, M90-V), which yielded a color or brightness temperature. In the work reported here, the brightness temperature was converted to the true temperature by correcting for the emissivity, as the electrodes are not true black-body emitters. In this effort, both polished and unpolished stainless steel (304) and tungsten (99.95%, metals basis) target plates were used as anodes. These electrodes measured in a dimension of 2.54 cm by 2.54 cm with 0.5 mm thickness. The exposure times for tungsten electrodes ranged between 2 and 30 min. Posttest the samples were analyzed using SEM, energy dispersive x-ray spectroscopy (EDS), and TEM to assess surface morphology, texturing composition, and crystallinity of the surface structures formed.

For all practical purposes, the structure of the atmospheric pressure DC glow discharge has all the characteristics of a glow discharge such as discharge current-voltage characteristics and discharge spatial structure.¹ These similarities were found in this work as well. For instance, visually the discharge has a distinct cathode and anode glow region as well as a defined positive column, as can be seen in Fig. 1(b). The current is limited by both power supply and a resistor that limit the maximum current to 0.3 A. This limitation prevents the discharge from constricting into an arc and therefore allows it to behave as a nonthermal glow. Even so, the gas temperature in the discharges of this type is typically high of order ∼1800 K but not as high as the electron temperature which has been estimated from laser Thompson scattering experiments to be a few electron volts, thereby making the T_e ∼10 times T_gas, thus nonthermal. According to Nikiforov et al.,⁹ at low ionization, high pressure plasmas, the electron density of an atmospheric glow discharge or plasma jet is in the range of 10¹³–10¹⁵ cm⁻³. In this work, the plasma density as inferred from Stark Broadening measurements through $Hβ$ line detection suggests densities of order 10¹⁴ cm⁻³. The relative spectrum can be found in Fig. 18.

Typical discharge behavior is resistive (roughly 16 Ω) with the discharge current increasing linearly with increasing discharge voltage for a given interelectrode gap. At a fixed voltage, the discharge current tends to decrease monotonically with increasing gap. This behavior is expected as energy and power losses associated with both scattering and attachment increase with discharge volume. Furthermore, the mean electric field and thus ionization rate in the positive column also decrease with increasing gap length. Such behavior is shown in Fig. 2 for helium plasma discharge in air with both tungsten and stainless steel anodes. In addition, for each case in this work, the plasma breakdown took place at a 1 mm interelectrode gap distance. For both stainless steel and tungsten anodes with helium flow, the breakdown voltages were found to be 600 and 800 V, respectively. A 1.3 kV breakdown voltage was recorded for an airglow without any helium flow.

After operating the discharge over a range of discharge currents (∼104 mA at voltage ∼2.3 kV), noticeable texturing was visually observable in the region of the plasma attachment. Figure 3 depicts a typical pattern that forms on the surface of tungsten if the discharge current is sufficiently high. Initially, it was suggested that this might be due to the localized melting at the anode surface. If the degree of material evaporation could be determined from images, in principle, it should be possible to estimate the energy deposited into the anode locally. This energy could then be used to estimate the magnitude of the local anode fall voltage, a parameter that is difficult to assess at atmospheric pressure using electrostatic probes owing to the small sheath thickness and the high collision frequency with background neutrals. The surfaces were analyzed using SEM to understand the nature of the possible melting. Surprisingly, the surface was not so much as melted but rather textured on a nanoscale. Surface nanostructuring was observed on both tungsten and stainless steel substrates with helium flow after exposure to the DC glow discharge. The tungsten substrate was treated for approximately 23 min with an interelectrode gap of 6 mm at a discharge current of 73.5 mA and an average discharge voltage of 1.75 kV. After exposure, the samples were analyzed via electron microscopy.

Figure 3 shows the effective “footprint” of the plasma attachment on the tungsten anode after plasma exposure along with surface morphology before plasma exposure. The texturing is noticeable at increasing magnification. Note that the center of the footprint is actually hollow which suggest that the discharge itself is hollow resulting in a ring shaped attachment at the anode. Such hollow features have been observed in DC glows as well as in helium plasma jets although the formation mechanism is not completely understood.^10,11 The current density is therefore significantly higher than the apparent circular footprint outline. As mentioned, the anode plasma attachment is localized. Figure 4 compares observed texturing outside the region without plasma exposure to a region inside the plasma attachment ring. Discrete, granularlike structures are clearly apparent on the surface of the plasma treated area.

Furthermore, most of the features shown have submicron characteristic dimensions. As depicted in Fig. 5, the region of plasma attachment featured in general three types of nanostructuring. Inside the ring region (region 1), the texturing is fibrous. Nano “fuzz” formation was observed just outside the ring attachment area (region 2). Somewhat further away from the midregion of the ring (region 3), nanoholes were observed. In detail, the fibrous zone of region 1 displayed rich texturing of two distinct forms: needlelike fibers and fibrous protrusions, which had somewhat of a flakelike appearance. These features have length scales ranging from 20 to 200 nm, which are displayed in Fig. 6. Just outside the attachment ring region, tungsten fuzz was observable. This nanostructuring is clearly seen in Fig. 7. As shown here, the structuring is uniform and extensive. Under the highest magnification, the structures are quite regular and seemingly interwoven. Nanoholes were observed outside the “footprint” of the plasma discharge as shown in Fig. 8. Presumably, this area is cooler and subjected to reduced ion flux. The observed holes are reminiscent of ruptured helium bubbles at the surface of substrates exposed to energetic helium flux.² In those studies, it is observed that the nanoholes can form when the ion energy is low (<20 eV) and if the surface is sufficiently hot. Nishijima et al. suggest that nanohole formation can occur at energies as low as 6 eV, the threshold for surface barrier potential energy for helium injection into the tungsten between atoms in the lattice.¹² This hole region, which located outside the region of the main attachment, is clearly surface textured indicating nonzero plasma bombardment albeit at presumably a reduced flux.

In order to assess the composition of the surface structures, EDS was used (Fig. 19). It was found that these surface features were predominantly tungsten and to a lesser extent, some oxygen was observed. It is interesting to note that in regions just outside the plasma footprint, EDS indicated the presence of copper as well. Presumably, this copper originated from the brass cathode and migrated via diffusion to the surface. The fact that it appears in regions outside the plasma footprint suggests that net deposition takes place here. Its absence in the ring footprint area suggests that those regions experience ion bombard at a rate that either matches or exceeds the copper deposition rate. This suggests that incident helium energies may exceed the sputter threshold for copper, which is approximately 25 eV.¹³ 

Similar to the tungsten, nanostructuring was also observed on stainless steel anode plates. The substrate was exposed to helium plasma for 68 min in total during the study of self-organization patterns. Voltages ranged between 0 and 2.74 kV with currents up to 142 mA. In the region of the plasma footprint, nanofuzz structures were observed. Figure 9 illustrates the fuzz structure with nanometer-sized features. EDS revealed these structures composed predominantly of iron and nickel with oxygen (see Fig. 20). However, there was no clear circular plasma footprint on the stainless steel surface after the helium plasma exposure. Since the total exposure time was long, this absence is likely due to plasma modification of the surface along with time (plasma driven resurfacing), which naturally sets an upper limit for exposure time. For comparison purposes, the discharge was also operated without helium gas flow. As stated earlier, the breakdown voltage for this operating condition was high, 1.3 kV (roughly twice of that with helium flow). The substrate was exposed to the air plasma for roughly 34 min at an average current of 62 mA. The discharge burned in regular room air. Under these conditions, nanostructures were not observed. For completeness, SEM images of this surface are shown in Fig. 10. As can be seen in the figure, the footprint of the discharge is apparent. However, there were no observable nanoscale structures. EDS analysis indicated the presence of iron oxide in the regions of the plasma footprint and the surrounding halo region (see Fig. 21). These data suggest that, at least over this range of operating conditions, nanostructuring does not occur, suggesting the importance of helium.

As discussed previously, various nanoscale structures were observed on both tungsten and stainless steel anodes when exposed to helium plasma under atmospheric pressure condition. These structures were observed over a range of discharge operating conditions and exposure times. Of particular interest is the development and evolution of these features with time under carefully controlled discharge conditions. In this manner, one can correlate their occurrence under specific operating conditions and yield insight into the formation mechanism. In order to achieve this, a controlled set of experiments was performed to understand the structure development. First, in order to minimize the possibility of nanostructure seeding via surface roughness,¹⁴ tungsten samples were mirror polished. The polished surfaces were then exposed to the helium plasma for set time intervals and then examined. After each treatment interval, the electrode was displaced to allow the discharge to attach at an untreated location on the 2.54 cm by 2.54 cm substrate accumulating approximately seven treatment sites on a given substrate. Note that the attachments were approximately 2.5 mm in diameter. After treatment, the various locations were examined using SEM. This method removed any bias regarding the polishing quality since each attachment was generated on a surface with the same surface treatment. This sample was also water cooled as described previously. For these time resolved measurements, the applied voltage was fixed at 1.8 kV with a constant discharge current of 53 mA and a fixed interelectrode spacing length of 12 mm. The plasma exposure time for a given treatment site ranged from 2 to 30 min. In this manner, the time evolution of the nanostructures was determined.

Figure 11 illustrates the observed macroscopic surface changes on the tungsten substrate as a function of time. On average after 2 min of plasma exposure, the hollow-shaped plasma footprint with a textured (light in color) central region is apparent. On subsequent exposure of duration up to 20 min, the circular footprint pattern is less apparent. After 30 min of plasma exposure, the circular footprint gives a way to lighter colored spots on the surface. This transition is related to the time evolution of the discharge itself, which over time self-organizes into a discrete ensemble of plasma attachments, essentially glowing dots. SEM analysis of the spotted regions (to be shown in Fig. 13) revealed a uniform distribution of fuzz similar to that shown in Fig. 7.

In-depth analysis of the spotted regions is presented in Fig. 12. While a range of structures and surface changes were observed via SEM analysis, as one can see in its subfigures, fuzz growth is apparent in many of the images. Figure 12 focuses on structures located at the middle of the attachment. As can be seen after 2 min of exposure, submicron-sized fuzz is apparent. Cavities are also present after 2 min of exposure. After 10 min of exposure, surface near center contains what are perceived to be particles and cavities, some of which with dimensions less than a micrometer. After 16.5 min, the surface is heavily textured with submicron-sized holes. Particles appear to be present on the surface as well. However, no fuzz structures were observed in the center region at this time point. After a full 30 min of exposure, fibrous fuzz is apparent in the center region along with cavities. Fuzz and holes are nanoscale in size. What is interesting about these features is the insight that the yield regarding the evolution of the discharge. Over time, the discharge apparently evolves from a double ringlike attachment to multiple dot attachments. The key question is whether the texturing observed on the surface is a consequence of discharge evolution or whether the texturing generated via plasma exposure affects the attachment pattern. In other words, this latter question concerns whether or not plasmas self-organization into the dot patterns is a consequence of surface changes. Furthermore, if the pattern changes are a type of feedback in response to the surface morphology, it may be possible to seed self-organization by a priori texturing the surface to test this possibility. This question is left to future work. It should be pointed out that EDS confirms that the observed fuzz and cavity are tungsten in composition. In the damage studies on tungsten due to helium plasma irradiation, microcracks or cavities appear similarly on the tungsten surface under fusion condition due to the bombardment of helium ions.^15–17 

Structures similar to those observed in simulated plasma-facing tungsten samples were also observed outside the center of the discharge attachment. Figure 13 illustrates these structures with how the overall surface evolves as a function of time outside the central attachment region. After 2 min of exposure, cracklike cavities and holes were observed. These features were observed both between the double ring attachment “footprint” and on the ring impression as well. Fuzz growths were also observed. At 10 min exposure, in addition to the cavities and fuzz, nanoholes and rod structures were observed. The formation of the rod structures on the surface is somewhat puzzling. Measured width of rods observed ranged between 20 and 80 nm. A rodlike structure has been observed in tungsten sputter deposition experiments. Here, the so-called Beta-phase W (100) nanorods have been observed. These simple cubic structures tend to have a faceted, pyramidal shape at the ends.¹⁸ While some of these morphological features are apparent on the observed rods, it is not clear if the same mechanism is at play as strictly speaking tungsten vapor is not likely under these conditions.¹⁹ Tungstate nanorods have been fabricated and studied for semiconductor applications. One synthesis method involves exposing the hot tungsten to an oxygen environment. Cracks in the natural tungsten oxide layer serve as reservoirs of tungsten. It has been shown that cracks expose (110) orientation planes most readily absorb oxygen. In this case, those cracks exposing (110) crystalline prominences were the location sites for nanorod growth, essentially along the crystalline plane. In that work, a tungsten wire was heated to around 800 °C in the presence of oxygen.¹⁹ These conditions are similar to that prevailing in this experiment. The morphology of the nanorods observed in this work is very similar to that observed by Tokunga. In this regard, this suggests that the role of the plasma in this case is simply thermal—a source of localized heat deposition to give rise to crack formation in the natural oxide layer. Nanohole diameters are measured to have a diameter between a few nanometers and 20 nm. Interestingly, the cavities at 10 min exposure also appear to be nanotextured. This can be seen in panel (b) of Fig. 13. The nanostructuring of microcavities has also been observed on the surface to tungsten plates exposed to fusionlike plasma. In this case, it was reported that nanoglobule structures were observed inside pinholes and microcracks.¹⁴ 

Heavily textured, intricately nanostructure surface features were observed beyond the central region after 16 min of exposure, as can be seen in panel (c). Significant nanohole coverage is observed in regions outside the main plasma attachment “footprint,” in presumably the cooler, lower plasma density region of the attachment. There the morphology at closer inspection appears almost weblike, nanoflake and fiber shaped structures were found at the edge of the plasma ring with rodlike structures appearing in between. EDS indicates that the fuzz is in the composition of tungsten (see Fig. 22). In particular, according to Kajita et al.,² such an extended rodlike structure can only be seen when the fluence is rather low, it is difficult to distinguish it from flakes when the fluence is high. In the outside of plasma print region, nanoholes or bubbles are observed as shown in panel (c) as well. After 30 min of plasma treatment, the surface both on plasma dot footprints and beyond appears to be densely covered with nanofuzz patches as shown in panel (d). EDS measurements confirm that this fuzz is indeed tungsten (see Fig. 23). Cavities are present as well. Surface particles also appear to be covered with cavities, nanoparticles, and nanoholes. Surface particles range in size from ten nanometers to a few micrometers in diameter. Apparently, these particles also exist in the nanocavities. In particular, nanoholes (∼80 nm diameter) and cracks with widths between 20 and 90 nm are observed on the surface of the large growth depicted in the figure. The nanostructuring action apparently takes place on any material present on the surface.

In these time-solved tests on mirror-polished tungsten (which minimizes the likelihood of prior features seeding structure), the same nanostructured features (fuzz, flakes, fibers, holes, and particulates) observed in fusion simulators have been observed in this atmospheric pressure DC glow. These data suggest that similar processes are taking place and that nanostructuring is not germane to fusion type plasma.

The nanostructuring observed on the tungsten and steel surfaces in atmospheric pressure is very similar to the tungsten fuzz structures observed in fusion plasma surface interaction studies. In the fusion work, the formation of such structuring is believed to related with helium bubble formation.^2,3 In many of these studies, the sample is biased negatively so as to attract ion flux to the heated target at the desired beam energy. Recent studies suggest that at least for helium ions on tungsten, the nanostructuring occurs in the 1000–2000 K range. Nanostructures appear at ion energies above 20 eV though nanoholes can apparently appear at energies below this. The typical helium flux required for nanostructuring on tungsten in this temperature range is ∼10²⁵ s⁻¹ m⁻².^2,20 While the target plasma exposure approach featuring the DC glow discharge differs from the plasma surface interaction studies in supporting fusion research, it is worthwhile to examine conditions prevailing in the glow to see if similar mechanisms are at play. Perhaps the biggest difference is the operating pressure, which is 1 atm for the DC glow—the presence of ambient air and high collisionality. The maximum temperature on tungsten at the location of the self-organized patterns without cooling was estimated to range between 1000 and 1300 K by using Wien's approximation. During the estimation, tungsten was considered as a gray body, which has a spectral emissivity²¹ less than that of a black body. As a result, these surface temperatures are in reasonable agreement with operational temperatures where nanostructures and bubbles form.^3,22 The ion fluence is more difficult to compare. If we assume that the 1 atm DC glow discharge has a negative anode fall voltage (plasma potential higher than anode), then we can expect ion flow to the surface at some energy acquired over the last mean free path.²³ This assumption is not unreasonable, as one would expect an accelerating potential present to give rise to the surface texturing and damage. In this case, the ion flux at the cathode matches the ion flux at the anode. The metered current is therefore a combination of ion and electron current. The electron current to the anode is just twice the ion current to first order if one neglects the secondary emission current. This neglect of the secondary electron current is justified, since at the discharge voltages observed in this experiment the secondary electron emission coefficient is much less than one. The actual electron current to the electrode is actually twice the metered current. Moreover, since the electron current “emitted” at the cathode is equal to the ion current, ion current is just half the total electron current incident on the anode. The metered current is simply the sum of the electron and ion current. The ion current can be easily calculated from density and plasma temperature data. Density estimate in this type of discharge is ∼10¹⁴ cm⁻³. If we assume an electron temperature of a few electron volts, then we can estimate the ion flux: ∼10²⁴ s⁻¹ m⁻².²⁴ This value, though on the lower end, is consistent with fluxes associated with nanostructuring reported in the literature.¹³ Densities at the anode surface could be considerably higher than this during self-organization as the estimate depends on both local density and temperature. The incident ion energy may be estimated as well. Because the halo region of the plasma footprint was due to the absence of deposited copper derived from the cathode rod, the incident ion energies must be at least greater than the sputter threshold for this element. This means that the ions must be hitting the surface with energies at least higher than approximately 25 eV. Furthermore, this suggests that the anode is indeed below the local plasma potential by this much. This value is consistent with the reported literature as well in regard to the lower limit for observed nanostructure formation. In summary, the prevailing conditions, based on these crude estimates, suggest that temperature, energy, and flux meet necessary lower limits for nanostructuring. The role of oxide formation is not addressed here. However, it most certainly plays some role in structure formation since the process in this case takes place in ambient air, which explains in part the presence of the tungsten rods observed.

The tungsten nanostructure formation was investigated to a limited extent using TEM to look for the presence of helium bubbles, to determine the crystallinity, and to assess the subsurface structure. Figure 14 shows the TEM electron diffraction pattern from the tungsten filaments of a helium plasma induced nanostructure. The discrete bright areas indicate crystalline structures on the surface. TEM images also revealed the presence of subsurface bubbles as can be seen in Fig. 15. This tungsten TEM sample was prepared by using focused ion beam in the bright field at a high beam voltage of 300 kV. Here, the bright regions constitute the voids. The characteristic dimension of the larger of these voids measured several hundred nanometers similar to that observed in the literature.^3,13,25 The fact that the voids were present in the material suggests that nanostructuring processes taking place in the fusion simulator samples are similar to those observed in the helium atmospheric pressure glow.

It is worth pointing out as discussed earlier that neither surface polishing nor controlled experimental parameters affected the appearance of nanostructures on samples. Also notable is that the anode pattern actually evolved over time. Figure 16 illustrates the variation in the discharge pattern over time. The evolution of the observed nanostructuring over time is believed directly related to the changing discharge and perhaps changes in surface morphology. As can be seen in the figure, the local current density tends to increase with exposure time. Here, the average current density is determined from the average size of attachment spots, the total number of spots, and the total discharge current. The fact that the discharge transitions from a ringlike attachment to a self-organized array of dots may be related to thermal effects. Indeed, the observation of spots on metal electrodes in arc discharges has been well documented and modeled. It has been asserted that cooling effects may drive an instability that leads to the self-organization. The time dependent nonequilibrium model Trelles used was able to show the formation of anode spots in a high current-free burning arc discharge.²⁶ The self-organization observed in this case is similar to the anode patterns observed in atmospheric pressure glows. This model suggests that anode cooling plays a key role in self-organization. The model suggests an imbalance between heavy particle cooling at the anode, which lowers heavy particle temperature and electron collection at the anode. To maintain equilibrium, electron temperature must increase to maintain plasma conductivity. No chemical or electrode effects were included in the model, yet it reproduced the spots, thus highlighting the cooling effect. Other studies suggest that the presence of a resistive layer located at the surface can lead to self-organization. The ohmic barrier would tend to regulate and thus stabilize the current flow and the resulting pattern. Raizer and Mokrov suggest that space charge can give rise to filamentation and thus the redistribution of the electric field at the surface, which can lead to self-organization.²⁷ Muller suggests that pattern formation is a consequence of a bistable layer in contact with a resistive zone at the anode surface.²⁸ What is clear is that over time, the surface in contact with the discharge heats up and in turn gives rise to increases in material resistivity. The increase in resistivity for tungsten is roughly quadratic with temperature.²⁹ Therefore, energetically, it would be more favorable for the ring attachment to spread out or break up into smaller current carrying attachments to minimize local resistivity and thus maintain current flow at a fixed voltage. Spreading out is not viable since at 1 atm diffusion path lengths are so small that the footprint of the attachment is inherently small. Breaking up into discrete attachments is, however, viable. Breaking into smaller attachments also allows for attachment sites to lose heat laterally and along the surface between spots. The fact that the spots tend to be in motion as well is consistent with this reasoning. The attachment spots are in motion to minimize local heating and thus local resistivity at the attachment point. As time wears on, the average plate temperature increases, necessitating the need for more spots. In this regard, the self-organization and the resulting texturing may be the basis of a minimization principle. Here, if the discharge is to be maintained at roughly constant power, self-organization into dots is therefore necessary. In this case, the self-organization behavior may be cast as a variational calculus problem. Phenomenologically, the configuration that minimizes the power expenditure is the “path” that is actually chosen. Assuming a fixed voltage, the plasma therefore essentially evolves such that the integral $J=∫x1x2⁡f{y(x),y′(x)}dx$ is an extremum, where in this case the minimum of this function is sought. Here, the x parameter represents the surface area associated with the actual attachment, and the function y is electrical conductivity at the surface associated with plasma heating at x, while the compound function f is the current density. If these functional forms and function y are determined, then the self-organization may be predictable. These same physical processes may prevail in DC glows with liquid anodes as well.

Surface nanostructuring generated by plasma exposure to a metal plate target serving as the anode of a DC helium-fed, 1 atm glow has been characterized. These structures were present on the plasma-exposed surfaces of both tungsten and stainless steel. Time resolved images revealed surface evolution in which nanoholes, various shapes of tungsten fuzz, and nanoparticles were observed. Figure 17 summarizes qualitatively the structures observed relative to the location of the plasma footprint. Interestingly, the morphology depended on the relative spatial location to the discharge attachment footprint. Presumably, large gradients in plasma conditions and temperature exist in the general vicinity of the attachment. The discharge attachment tended to evolve over time, transitioning from a ringlike attachment to dots. With this evolution over time, the average current density was observed to increase. It is possible that the discharge attachment morphology was a consequence of surface morphological changes. Furthermore, nanorods were also observed and appear to be a consequence of localized thermal heat treatment. Helium bubbles similar to those observed in fusion target studies were observed in this work too. Overall, it was found that observed nanostructuring was similar in morphology and size to that observed on tungsten samples exposed to near-wall, fusion simulated helium plasmas at low pressure as shown in Table I. Remarkably, in both cases, surface temperature and helium ion fluence were similar despite the pressure difference. These findings suggest that similar mechanisms may be active in the surface evolution of substrates exposed to the 1 atm glow and those exposed to fusionlike plasma in the presence of helium.

This work is supported by the U.S. Department of Energy with an Award No. DE-SC00-18058 and the Michigan Memorial Phoenix Project (MMPP) seed grant program.

This section provides additional evidences in order to support the experimental results in Sec. III. The optical emission spectroscopy example with estimated plasma density is shown in Fig. 18. Figures 19–23 show the typical EDS with composition identification that are associated with the nanostructures in Sec. III B.