1. Introduction
In parallel with wear- or corrosion protection-related applications, thermally sprayed coatings or coating composites are used more and more in the electronics sector, e.g., in conductive tracks or integrated circuits [
1,
2], as they will allow to adaptively adjust the properties of the final product [
1,
3,
4,
5,
6]. Another application area that is currently being studied in great detail is the manufacturing of electrical components in automotive engineering. Particularly complete heating systems for vehicle passenger interiors are nowadays fabricated by a variety of coating technologies due to their high efficiency. An essential component for the proper functioning of such components is the ceramic insulating coating, which is usually based on Al
2O
3 [
5]. Currently, alumina coatings are mainly produced in the industry by atmospheric plasma spraying (APS) [
5,
7,
8]. On top of this isolating coating, a thin resistive heating coating is applied by atmospheric plasma spraying (APS). These resistive heating coatings are typically fabricated of NiCr 80/20 due to high consistency as well as low noise resulting in stable long-term electrical properties [
9]. Additive-subtractive methods are increasingly used in the development for e.g., temperature sensing or embedded microheaters [
5,
10]. To achieve homogeneous heating, the thermal spray NiCr coating (additive part) is usually refined by processes like laser texturing (subtractive). Atmospheric plasma spraying as a partly hot powder process, however, has some disadvantages in the production of thin resistive coatings. On one hand, there are the side process times until the plasma is stable and the deposition can be started. On the other hand, NiCr particles often oxidize strongly on an industrial scale and thus apart from a possible weak bond or cracking, partially interfere with subsequent processes, see
Figure 1.
Thus, simpler spraying technologies such as arc spraying (AS) are presently also being investigated for their technological suitability as a cost-effective alternative to plasma spraying in the manufacturing process for these components. Other technologies, such as HVOF spraying or wire flame spraying are potential options for achieving the desired properties too [
11,
12,
13,
14]. However, this is contrasted by higher costs due to either more expensive equipment in terms of HVOF or limited deposition in highly automized production lines when wire flame spray is compared with arc spray. Arc spray is an established technology and a good compromise in terms of high deposition rates, low investment costs and robust technology [
15,
16]. For this reason, this technology was chosen in agreement with the industry partner/heater manufacturer. Yet, in case of the proposed application, it must be clarified whether such thin coatings can be structured like the plasma-sprayed coatings and to what extent this affects the underlying ceramic coating. Moreover, the typical lamellar structure with partially big splats might be problematic. Yet, own comparative tests regarding the adhesive tensile strength showed promising adhesion of NiCr 80/20 coatings deposited by arc spray compared with the typical industrial system sprayed by APS, see
Table 1. Nevertheless, the different substrate materials can play a role in this regard. It should also be noted that further coatings are applied at the component level and that these values are therefore only of limited significance as a part of the entire system. The focus of the investigations will therefore lie more on the cohesive, above all electrical, properties of the coatings and heating units, i.e., after texturing.
Our earlier work and the related literature have shown that coatings sprayed with a mixture of nitrogen and hydrogen provide better coating properties in terms of oxidation, electrical conductivity as well as a finer overall microstructure [
16,
17]. Hence, solely alternative pressurizing gases and mixtures are utilized in this study. The approach in this paper can thus improve both the efficiency of the process and potentially the coating properties. The focus of the first part of the paper will therefore be on different variations of these and the effects on microstructure and functional properties. In the case of applications such as heating conductors, the coating thicknesses and masses are crucial in determining the functionality of a coating composite. For this reason, it is imperative to comply with a range of these coating properties defined in advance. As with all other coating properties presented below, the mass and thickness of the arc-sprayed coatings were therefore set in relation to the properties of the currently used plasma-sprayed coatings as a reference. Furthermore, certain critical influencing variables of the finished structured NiCr coatings must be kept in mind to allow a certain degree of flexibility in the design of the component. In addition to the resistivity of the heating element itself, this also includes the temperature coefficient of resistance (TCR) [
9], which should ideally be as close as possible to the bulk material.
However, for an accurate assessment of the microstructure, the best possible preparation is necessary. Yet, the preparation of thermally sprayed coatings already differs significantly from that of bulk materials during the sectioning process. Depending on the porosity, the materials used, and the grinding and polishing steps, coating defects can be artificially enlarged or even masked or covered up [
15,
18,
19,
20]. The individual coating behavior further depends on the coating material and the degree of automation of grinding and polishing [
15,
18,
19,
20]. Consequently, a strong dependence of the assessment of the coating quality on the preparation method can be assumed. This is also evident for the mounting method. For example, cold mounting is usually recommended for thermal-sprayed coatings instead of hot mounting due to the absence of excessive pressure and heat. In the following, after all, these coatings are mechanically processed, which can possibly distort the results as described before [
15,
18,
19,
20]. To minimize the influence of sample preparation, alternative methods have been established in recent years, such as broad ion beam (BIB) preparation, which is analogous to focused ion beam (FIB) preparation used for the TEM [
21], but covers a larger area of the sample. The advantages of this method are that almost no preliminary preparation is necessary and that temperature-sensitive materials can be processed. This means that even small and fragile structures can be prepared. Since such devices run fully automated, human influence during preparation is also minimized as far as possible [
21].
In summary, the objectives of this study consist of the dual approach of, on the one hand, optimizing the coating properties of heating elements by process changes during arc spraying and, on the other hand, evaluating the influence of specimen preparation on the results of coating analyses.
2. Materials and Methods
The arc spray experiments were carried out by using a Sulzer Metco Smart Arc, a PPG gun and two wires of NiCr 80/20 (Metco 8450, Ø 1.6 mm; system and wires both from Oerlikon Metco Europe GmbH, Kelsterbach, Germany). In general, the substrates were prepared and coated like the reference coatings. This means, that sheets of an aluminum alloy (t = 3 mm) were roughened and then coated with a NiCr 80/20 bond coat and a ceramic Al
2O
3 top coat by plasma spraying with a GTV Delta torch (GTV Verschleißschutz GmbH, Luckenbach, Germany; powder process; modern three anode system). Afterwards, the arc-sprayed coatings as well as the plasma-sprayed reference coatings were then applied directly to the ceramic top coat without further substrate preparation. The experimental setups for the arc spray and plasma-sprayed reference coatings are depicted in
Figure 2.
Yet, the plasma-sprayed reference coatings were fabricated on prototype components (same material) and not sheet metal. For these coatings, the same GTV Delta torch was used as described above. The spray parameters are listed in
Table 2.
It is visible that Variation 1 (S1–S6) was about the use of various air caps, gas types, as well as stand-off distances (SOD), while the aim of Variation 2 (S7–S9) was to further optimize gas flow by use of a secondary gas. For Variation 1, the composition of the gas was varied because different amounts of hydrogen have additional reducing effects as well as add energy to the process for improved particle adhesion. In addition, the air caps as well as the SOD were adjusted to specifically influence the particle impact behavior and thus the cohesion of the coatings. For Variation 2, the wire feed rate of specimens S8 and S9 was increased slightly in order to achieve a proper coating mass. While specimens S1 to S9 were part of the parameter determination study (1 substrate each), series S11 and S12 (10 substrates each) are the preferred parameter sets of the two variations with respect to directly measurable coating properties, particularly mass, surface quality, and resistivity. Process kinematics were maintained the same as for the plasma-sprayed reference coatings, i.e., the sample sheets were rotated and coated with the same number of passes and by an identical robot traverse speed.
Following the spray process, the coating masses were determined using one sample each, with the exception of series S11 and S12 (10 samples each), using an SBS-LW-2000 A precision balance (Steinberg Systems, Zielona Góra, Poland). In addition, the coatings were visually checked with a CANON EOS 700D and ImageAccess software (Imagic Bildverarbeitung AG, Glattbrugg, Switzerland) with respect to the sample surface.
The specific electrical resistivity of the coatings was examined by four-terminal method using a Loresta GX MCP-T700 system (Mitsubishi Chemical Analytech Co. LTD, Kanagawa, Japan; constant current 10 mA), with 7 values recorded at 6 measurement points for each sample, see
Figure 3.
Since coating thickness has a major influence on the calculation of the resistivity, the corresponding values were used for samples S1 to S9. For S11 and S12 the available data are more reliable, since for these the mean values of 3 specimens were applied. Attention was paid to the coated substrates with the largest, the smallest, and the medium coating mass. In this manner, the entire spectrum of the test series is well represented. However, all coating thicknesses were only determined at the same spot by means of a microsection, as described subsequently, and not at each single measuring point following
Figure 3b.
After spraying, the specimens were textured with a laser (Rofin D100, ROFIN-SINAR Laser GmbH, Hamburg, Germany), just like the plasma-sprayed coatings in industrial application, so that a geometrically defined pattern in two distinguishable loops could be obtained on the coatings. As the controlled heating would now take place by applying a voltage, the coatings can henceforth be termed heating conductors or heating elements.
The surfaces of the specimens were inspected visually (same equipment as described above). Since the temperature coefficient of resistance (TCR) and the resistivity of the heating conductor are key properties of the entire heating unit, they were measured after texturing. For this purpose, the resistivity was measured using a digital multimeter at room temperature and at 140 °C, after contacting the textured coatings via an additional copper coating onto which contact wires were soldered. Whereas the resistivity of the heating unit can be considered as the product of measured ohm resistance and coating mass, the TCR is computed as described in [
9].
For microstructural analyses of the previously described variations, the samples were first hot-mounted (HM) using an ATM Opal 410 Hot Mounting Press (ATM GmbH, Mammelzen, Germany) and then ground and polished stepwise (6 µm, 3 µm suspensions, finally oxide polish). The coating thickness was analyzed using a Leica DM6000M optical microscope (OM) (Leica Microsystems GmbH, Wetzlar, Germany) and the ImageAccess software tool, taking 5 readings, 3 times, for each sample. Moreover, a JEOL JSM-IT100 scanning electron microscope (SEM) (JEOL Germany GmbH, Freising, Germany; accelerating voltage 20 kV, backscatter detector) was used for coating thickness measurements of the reference samples with 3 times 7 values for each sample.
The coating morphology was investigated representatively by SEM analyses (see above for type and conditions) at various magnifications at 3 spots in each section. In addition, energy dispersive X-ray spectrometry (EDS) was performed at the same 3 locations within the cross-sections of the arc-sprayed and reference coatings using a JEOL Dry SD25 (JEOL Germany GmbH, Freising, Germany; accelerating voltage 15 kV). This allowed the local chemical composition of the coatings to be identified. Specifically, the oxygen content was examined exclusively in the NiCr coating, which was defined as region of interest (ROI) for the analysis, see
Figure 4. This enabled a quantitative side-by-side comparison of the coatings.
For the sake of comparing different preparation methods in the second part of this work, some parts of the same samples (S11, S12) were (I) additionally cold-mounted (two-phase system: liquid hardener and powder resin) and gradually ground and polished (6 µm, 3 µm suspensions, finally oxide polish). The plasma-sprayed reference coatings were also cold-mounted like the arc-sprayed samples S11 and S12.
For further comparison purposes, segments of the same specimens S11 and S12 were (II) ion beam polished (IP) using a JEOL cross-section polisher (JEOL Germany GmbH, Freising, Germany). This was carried out by mechanically cutting sample parts first. Subsequently, the specimens were polished using a voltage of 5 kV for 8 h in a cycle of 45 s each using the electron beam and a 15 s pause. Afterwards, characterization of these specimens was carried out by OM, SEM, and EDS using the same equipment and parameters as described before, while only magnifications were different for microscopy.
Finally, SEM images (type and conditions see above) and the software ImageJ (National Institutes of Health, USA; using Despeckle filter in the ROI, normalization and finally the Trainable Weka Segmentation tool) were used for examining the content of porosity/oxidation/cracks inside the coatings for every preparation method and for both materials, the NiCr coatings (magnification 2000×) and the underlying ceramic Al2O3 coating (magnification 500×). Three images were taken for the hot-mounted and cold-mounted specimens, while it was one image for the ion beam polished specimens. However, due to the much larger area for the latter one, the examined specimen area was comparable in size. This procedure was also carried out for HM samples from the first part of the study, i.e., for the arc-sprayed coatings.
4. Discussion
Similar to the results section, the discussion will also be divided. While in the first section of the discussion the coatings and their functional properties will be evaluated, the heating elements in their entirety will be discussed in the second section. The various preparation methods will be evaluated in the final section of the discussion.
4.1. Evaluation of the Arc-Sprayed Coatings
In case of using NiCr as coating material, it was not clear whether the changes in the spraying process would be sufficient for the coatings to meet the previously defined requirements, such as the specific coating mass or the resistivity of the finally textured heating elements, so that the arc-sprayed coatings could eventually be equivalent to the plasma-sprayed reference coatings.
The parameter sets were investigated directly after the spraying process by immediate analyses of coating mass, thickness, surface quality, and specific resistivity. In particular, the relationship between a high specific conductivity or a low specific resistivity and minor quantities of coating defects such as oxides and pores confirmed this approach, which is displayed in
Figure 15. The primary gas mixtures of pure nitrogen as well as nitrogen and 4% of hydrogen can be regarded as the largest factor for the reduction in oxygen content in all the specimens. In terms of oxidation, all coatings are superior to the reference coatings. This fact can be attributed mostly to the shroud effect. The shroud effect describes that particles exhibit less oxidation by the use of inert gases and their mixtures due to protection from the atmosphere [
7]. As assumed, the specific resistivity decreases almost linearly in dependence of the measured oxide content, see
Figure 15.
Furthermore, the change in air cap and thus gas flow largely influences the results of Variation 1 samples, compare S1 and S2 in
Figure 15. The use of additional hydrogen (S3–S9) should have further reduced oxidation and promoted an increase in elastic energy of the particles. However, the samples S3 and S2 coincide partly regarding their oxygen content and specific resistivity. The subsequent reduction in stand-off distance from 120 mm (S3) to 60 mm (S6) changes this to a certain extent, i.e., oxygen content and specific resistivity both decrease. Yet, it must be noted, that the shortest stand-off distance does not lead to reproducible results anymore, which is proven by the high standard deviation of the resistivity and the irregular, partially stained surface. Considering these results, the optimum coating quality for Variation 1 is reached by specimens of S5 and S11, which almost equal the reference coatings considering the specific resistivity.
In the case of Variation 2, the gas flow optimization by change in the air cap and the use of nitrogen as secondary shielding gas further enhanced the gas shroud effect. As already described for Variation 1, reproducible results cannot be achieved with the lowest spray distance. Thus, the specimens of S8 and S12 of Variation 2 reach the most suitable coating properties regarding the whole test series, compare
Figure 8 and
Figure 15,
Table 2.
Despite the superior coating properties achieved by arc spraying, a certain amount of cracking was observed in the arc-sprayed coatings as well as in the ceramic coatings underneath. A reason for cracking can be residual stresses, resulting from the superposition of quenching stresses by subsequently evolving cooling stresses [
7]. Nevertheless, it is more likely that the preparation of the specimens has the greatest influence on the occurrence of these cracks since the influence of residual stresses is weakened by the use of alternative gas mixtures [
7]. Hot mounting takes place under increased temperature and pressure, while the subsequent grinding and polishing processes artificially expand potential cracks or even particle boundaries. This shall be further discussed in
Section 4.3. Findings from the advanced analytical methods.
4.2. Suitability of the Coatings as Heating Units
Aside from reaching very positive results regarding the coating quality, the coatings needed to be approved for use as heating units by measuring their electric characteristics.
Due to the very low oxygen contents and the uniform surfaces, see
Figure 5,
Figure 10 and
Figure 11a, the coatings could be textured successfully using the same parameters as for the plasma-sprayed reference coatings. Nevertheless, the coatings became visibly darker and more irregular due to the heat input of the texturing process, which indicates a partial surface oxidation of the coatings. This might explain the partial scattering regarding the quantitative evaluation of oxygen content, especially for Variation 1, which was affected more than Variation 2.
As already presented before, the specimens S8 and S12 reach the most suitable coating properties. Furthermore, these specimens also happen to exhibit the lowest TCR values, which are far lower than the reference value, see
Figure 16.
Overall, the relationship of coating oxygen content and TCR resembles the findings in the previous chapter. With regard to the use as a heating element, a low TCR is crucial, as it ensures a low temperature dependence of the resistivity [
9]. A low TCR therefore also increases the reliability of the heating element to a certain extent. In view of these correlations, the arc-sprayed coatings can be considered as suitable for this kind of application. This can mainly be attributed to the almost oxide-free structure of these coatings, which in turn, is a result of using wires as feedstock. Coatings manufactured using powder feedstock with particles of varying sizes tend to show a greater scattering in their characteristics due to the different specific surfaces of the particles and thus display a high level of local oxidation.
However, the disadvantage of these very oxide-free coatings is that the resistivity of the heating element is only slightly higher than for the currently used plasma-sprayed reference coatings, see
Figure 16b. However, this was to be expected from a physical point of view, since the resistivity of the heating conductor tends to correlate with the specific coating resistivity. For future applications, e.g., for differently designed heating conductor geometries, this could be a problem. Nevertheless, it should be noted that the specified minimum value of resistivity regarding the heating conductor was reached and even slightly exceeded.
In summary, it can be stated that both variations resulted in coatings that can potentially be used for heating elements. While the best coating properties are achieved by secondary atomization in Variation 2, the use of the High Velocity air cap with optimization of the stand-off distance in Variation 1 is a workable compromise, see
Figure 15 and
Figure 16. Using a High Velocity air cap could provide further advantages, e.g., regarding the coating of modified heating element designs.
4.3. Findings from the Advanced Analytical Methods
First of all, it can be stated that the preparation methods CM and IP have great advantages over HM, especially with respect to the evaluation of ceramic coatings. For example, the previously observed horizontal cracks in the underlying Al2O3 coating, as already suspected, clearly originate from the preparation method of hot mounting, and could not be observed for the other methods. Only smaller, non-critical material separations near the texturing and at inter-particle boundaries in the NiCr coating were recognizable. The latter probably are a result of the preparation parameters, which have not yet been fully optimized. For example, no variations were made in the preparation voltage for IP. Furthermore, in the case of CM, it must be noted that the materials NiCr and Al2O3 have very different mechanical properties, e.g., hardness, which strongly affects the preparation. A uniform, optimum mechanical preparation is thus very unlikely. In contrast, IP already shows very good, uniform microstructures for both materials. For Al2O3 in particular, this manifests in a finer degree of preparation. For NiCr, on the other hand, some small oxidation effects due to IP were detected via EDX and must be taken into account. Furthermore, the origin of the material separations between the splats visible in the IP specimens cannot be clearly explained. On one hand, it is conceivable that already existing weakly bonded areas were artificially widened as a result of the preparation time and voltage. The slight change in the coating oxides chemical composition also suggests this. Accordingly, these facts would require an optimization of the preparation parameters. However, another possibility that must be considered is that arc wire spraying leads to relatively long and large particles compared with, for example, powder-based coating processes. These smaller cracks could therefore also indicate a suboptimal bonding of the rather thin coatings. Reasonably, a superposition of these two effects can therefore be assumed.
Nevertheless, from direct comparison it can be concluded that the preparation by means of IP provides the most reliable and best results, which is also reflected in the quantitative evaluation of the coating defects.
5. Conclusions and Outlook
The aims of this study were twofold. On one hand, several mixtures and combinations of alternative pressurizing gases and further process modifications of the arc spray process, such as change in air cap and stand-off distance, were tested to optimize the coating characteristics regarding their use as heating elements. On the other hand, advanced preparation methods were used in order to quantify the influence of sample preparation on the coating analysis results. The key findings of the study are listed below:
The process optimization results demonstrate significantly reduced oxygen contents and improved coating morphologies compared with the currently used plasma-sprayed reference coatings. This also affects the measured defect content overall positively, especially in terms of oxidation. It can be assumed that a further reduction in oxygen content by parameter adjustment alone is difficult to achieve, as a residual amount of oxygen is always present with arc spray processes. However, a change to a modern power source and an improved torch design might improve the coating properties.
The improved microstructure positively affects the surface quality and the specific resistivity of the coatings. Nevertheless, some micro cracking was observed inside the arc-sprayed coatings and the ceramic coatings underneath. Although the cracking most probably originates from the preparation of the specimens by hot mounting, as was evident from the comparison of different sample preparation methods, characteristics of the arc spray process must be considered too.
Due to the improved surface structure and reduced oxide content, the arc-sprayed specimens can be subsequently textured just as successfully as the plasma-sprayed reference coatings. Moreover, the temperature coefficients of resistance (TCR) and the resistivity of the heating elements were found to be superior to conventionally manufactured coatings.
It can be shown that the preparation method is essential regarding the analysis results, especially for ceramic coatings. Compared with hot mounting, cold mounting and ion beam polishing were clearly less influential on the coating characteristics and led to more representative analysis results. The latter allowed a very fine preparation of the coatings, which even provided insights into the dendritic microstructure depending on the position within the coating splats.
The key characteristics of the coatings (adhesive and cohesive), which can be found in the previous figures and tables throughout the manuscript, are summarized again in the following
Table 4.
The previously listed findings and
Table 4 suggest that the improved properties of the arc-sprayed coatings, including superior temperature coefficients of resistance and resistivity, hold promise for potential industrial applications in heating element production. However, the technology requires further validation and additional tests. Future research could include, for example, prototyping and testing the technology on actual heating element components over many load cycles to validate its industrial applicability.
In conclusion, this study successfully optimized the arc spraying process for heating element coatings, demonstrating improved properties and, moreover, highlighted the significance of sample preparation methods in coating analysis.