Nuclear fusion is a prospective energy source for the future, which brings the promise of a safe, relatively clean and large-scale energy source. Its successful realization is critically dependent on the availability of materials able to function in the extremely harsh conditions of a fusion device. In the recently published “Fusion Roadmap” [1
], three out of seven major challenges to its realization as an energy source are related to materials. Among the milestones on this road are the International Thermonuclear Experimental Reactor (ITER), currently under construction, whose purpose is to demonstrate the technical feasibility of a fusion reactor and to integrate and test the necessary components and technologies, and the Demonstration Reactor (DEMO), being the successor of ITER, whose aim is to demonstrate the feasibility of a complete, economically viable fusion power plant.
Tungsten is one of the candidate materials for the plasma facing components of ITER, and the prime candidate for the entire plasma facing surface of DEMO [2
]. These components must withstand the high heat and particle fluxes from the plasma. Among the favorable properties of tungsten are its high melting point, low vapor pressure, good thermal conductivity, high temperature strength and stability, high threshold for sputtering; also, it does not form hydrides or co-deposits with tritium. On the other hand, its high atomic number makes it a highly undesirable impurity in plasma, while further disadvantages include its brittle nature and difficult machining [3
]. Tungsten plasma facing components (also termed “armor”) can be fabricated by different methods, including bulk material processing (chiefly powder metallurgy + joining) and different coating techniques (physical and chemical vapor deposition (PVD, CVD) and plasma spraying (PS)) [5
]. This paper concerns plasma spraying, whose advantages include the following [7
Ability to coat large-area components, including non-planar shapes, with significant thickness
A single-step manufacturing technology, without the need for further joining
Possibility of in-situ repair of damaged parts
Easy formation of graded composites
Moderate heat input to the coated parts
High strain tolerance
The main disadvantage of plasma sprayed coatings is their low thermal conductivity, which makes them applicable only in regions of low to moderate heat fluxes [8
As with all coatings, adhesion is a critical issue; without sufficient adhesion, all other coating functions will be lost. Different coating adhesion mechanisms are usually divided into three groups: mechanical, physical, chemical/metallurgical [9
]. There are numerous factors that influence adhesion —e.g., the roughness, temperature and composition of the substrate, the presence/absence of adsorbates, bondcoats/interlayers, temperature and velocity of the impacting particles, powder feed rate, spraying pattern, torch traverse velocity, splat-substrate wetting, splat spreading, solidification, splashing, oxidation of splats and substrate, residual stress, coating thickness, etc.
, many of which are interconnected. Some of these factors will be briefly reviewed here, together with selected case studies.
Mechanical interlocking is generally accepted as the main adhesion mechanism in most thermal spray methods. Therefore, substrate roughening, e.g., by grit blasting is a common practice in thermal spraying. In general, higher substrate roughness leads to higher adhesion [11
]; this has been observed both for the deposition of fully molten (plasma sprayed) and semi-molten (HVOF-sprayed) particles [14
]. However, for higher roughness range (>50 μm), adhesion was also found to decrease with increasing roughness [15
]. The interlocking is most effective under the center of the splat, where the pressure upon impact is highest, and decreases towards the splat periphery [10
]. Pressure impulse at the point of impact increases with the kinetic energy of the incident droplet. Thermally induced bonding can occur when the conditions for incipient melting of the substrate are met [10
], which generally leads to improved bond strength. The necessary temperatures of the droplet and the substrate were calculated in [16
] for a variety of coating/substrate combinations. Furthermore, there has to be enough time for the splat to melt the substrate, and both to solidify before the impact pressure is dissipated [10
]. Localized substrate melting promotes the formation of metallurgical bonds, e.g., the formation of intermetallic compounds at the interface [16
]. However, under typical conditions of thermal spraying, the interdiffusion distance is of the order of units of nm [10
]. Even in the case of mutually insoluble materials (e.g., W and Cu) where no reaction interlayer forms, local melting promotes closer contact of the two materials and thus improves the bonding [16
]. For the above reasons, higher deposition temperatures generally lead to improved adhesion, but they may also induce or increase oxidation of metallic materials, which generally hinder adhesion [10
]. Therefore, a compromise has to be made for a specific material combination.
The first layer of deposited splats represents the contact between the coating and the substrate. Therefore, their formation is crucial for the development of adhesion. An important factor in this aspect is the deposition temperature. In many materials, a transition temperature was found, above which contiguous, disk-shaped splats typically form, and below which they tend to splash and fragment [17
]. This has been attributed to adsorbates and condensates on the substrate surface [18
], which evaporate under the hot splat and form a gaseous barrier between the two surfaces or are trapped under the splat in the form of bubbles [19
], in both cases increasing the thermal resistance. On heated surfaces, these adsorbates/condensates are evaporated and the splat–substrate contact is improved [21
]. Disk-shaped splats are generally preferred, as they are associated with high adhesion and cohesion, low porosity and good mechanical and thermal properties of the coating. Coatings formed from splashed splats typically have poor adhesion and cohesion, and high porosity [20
]. The detrimental effects of splashing include the formation of voids that are difficult to fill by the subsequent droplets, reduced time for substrate melting and reduced pressure in the splash drops [10
]. The presence of oxides or hydroxides at the interface might inhibit the contact between the two materials; if these are thin enough, however, localized substrate melting, jetting and interdiffusion are still possible [22
]. Preheating appears to improve the wettability of the substrate by the particles, thereby enhancing the adhesion, even in cases of splashing [15
]. Rough substrates often lead to splats with higher thickness and lower diameter [10
], while their splashing is suppressed [23
]. Substrate roughness also presents higher thermal resistance, with only limited contact points, and this leads to longer thermal interaction [10
]. Mutual bonding between the splats is also important, as coating delamination can occur at the interface (adhesive failure) or within the coating (cohesive failure), or in a combination of both. Among the factors positively affecting interparticle bonding (cohesion) are the particle temperature and deposition temperature. Particle velocity, although generally accepted to produce higher density coatings, was found to be rather detrimental to interlamellar bonding [24
In cases of largely dissimilar materials, bonding interlayers (or “bondcoats”) are often introduced. The most common example is thermal barrier coatings in jet engines, where sprayed Ni- or Co-based bondcoat is introduced between the Ni-based turbine blade and plasma sprayed zirconia-based topcoat. In [25
], tungsten as plasma facing material was deposited by vacuum plasma spraying (VPS) on graphite with SiC and Ti interlayer; only coatings with Ti exhibited sufficient adhesion strength in thermal exposure tests. In [26
], the case of VPS-W on Cu was considered by finite element modeling (FEM), with interlayers of W/Cu, Ti and NiCrAl. From these simulations, the W/Cu appeared the best in reducing the stress concentration at the interface without significant increase of the surface temperature. Such mixed layers can be advantageously formed by plasma spraying [27
]. VPS and PVD were successfully used to produce functionally graded Eurofer97/tungsten coatings that proved suitable as interlayers for joining Eurofer97 and tungsten bulk material by diffusion bonding [30
]. The structures bonded this way survived several thermal cycles up to 650 °C, without new phase formation or change in chemical composition, and showed a marked improvement over direct diffusion bonding.
This paper focuses on the adhesion of plasma sprayed tungsten on steel substrates. Two factors are considered—substrate roughness and the presence of W and Ti interlayers. The shear adhesion test is complemented by detailed characterization of single splats, interfaces and the fracture surfaces.
2. Experimental Section
For the experiments, six conditions of the substrates were used: bare steel substrates with two levels of roughness (grit-blasted, termed ‘fine’ and ‘coarse’) and steel substrates with two types of thin interlayers—Ti and W, again at two roughness levels each (as-machined and grit-blasted substrates). Ti was chosen as a compliant interlayer with the prospect of reducing the stress concentration at the W/Fe interface, caused mainly by the different thermal expansion. Ti was also successfully used in joining of W to steel by hot isostatic pressing (HIP) [31
]. The W interlayer was tested because of its expected good adhesion to the substrate and chemical affinity to the sprayed coating (same element). Low alloyed steel of S235JRC (1.0122) type was used for all the substrates; substrate dimensions were5 × 10 × 30 mm for the shear adhesion tests and 2.5 × 25 × 25 mm for the observation of individual splats and coating cross sections. Different roughness levels of the bare steel substrates were achieved by different sizes of the alumina grit and air pressure during the grit blasting. Roughness was measured either after grit blasting (bare steel substrates) or after the PVD coating (substrates coated with Ti and W), using a Surtronic 3P surface profilometer (Rank Taylor Hobson, Leicester, UK). The roughness values are presented in Table 1
Magnetron sputtering of the Ti and W interlayers was performed in a Hauzer Flexicoat 850 (Hauzer Techno Coating, Venlo, The Netherlands) equipment. First, the samples were ultrasonically cleaned in acetone, flushed with ethanol and dried, before insertion in the coating chamber. The chamber was evacuated to 2 × 10−5
mbar and preheated. The targets were cleaned for 15 min and the samples were cleaned for 20 min by Ar ion bombardment, with plasma current 60 A and bias 200 V. The deposition conditions are summarized in Table 2
The W coatings were sprayed by a WSP® water stabilized plasma torch (Institute of Plasma Physics, Prague, Czech Republic), using a 5:1 mixture of tungsten (Alldyne, Huntsville, AL, USA; 63–80 μm) and tungsten carbide (Osram, Bruntál, Czech Republic; 40–80 μm) powders. The WC is a sacrificial additive, undergoing decarburization during spraying; the carbon reacts with oxygen, making less of it available for oxidizing the tungsten [32
]. X-ray diffraction was used to check that the WC converted completely to W. The spraying parameters were as follows: torch current 500 A, powder feed rate 33 kg/h, carrier gas Ar + 7% H2
, feeding distance 35 mm, spraying distance 200 mm. All substrates were preheated to 160 °C by one passage of the torch prior to coating deposition. The substrate temperature was monitored by an infrared camera. Samples for the observation of individual splats were produced by one fast sweep of the torch in front of the substrates at 500 mm/s traverse velocity. Full coatings for the adhesion tests and metallographic observations were produced by a rectangular meandering pattern across all substrates at a traverse velocity of 300 mm/s. To prevent overheating of the samples, each deposition cycle was followed by several cooling cycles (again with Ar + 7% H2
as the cooling gas, helping to suppress oxidation) until the deposition temperature decreased from ~250 °C to ~160 °C. The resulting coating thickness was about 650 μm.
The adhesion strength was measured using a standardized shear test (EN 15340) [33
]. The main advantage of the shear test, compared to the common tensile adhesion test, is the absence of glue, which may otherwise affect the results, and the relatively uncomplicated sample and test configuration [34
]. The test samples were prepared in the shape of a prism with dimensions of 10 × 5 × 30 mm; coatings were deposited on the 10 × 5 mm face. The test was performed using a universal tensile test machine Instron 1362 (Instron, High Wycombe, UK). Loading was applied in the direction perpendicular to the sample longitudinal axis by means of a carbide cutting edge SPEW 1204 ADEN: 8230 (Pramet Tools, Šumperk, Czech Republic) moving at velocity of 3 mm/min.
Overview of surface conditions prior to plasma spraying.
Overview of surface conditions prior to plasma spraying.
|Notation||Surface treatment||Interlayer||Roughness Ra (μm)|
|R1||grit-blasted, coarse||-||7.8 ± 0.4|
|R2||grit-blasted, fine||-||5.4 ± 0.2|
|T1||as-machined||Ti||1.7 ± 0.2|
|T2||grit-blasted, fine||Ti||6.0 ± 0.6|
|W1||as-machined||W||1.6 ± 0.1|
|W2||grit-blasted, fine||W||5.6 ± 0.6|
Magnetron sputtering parameters for the Ti and W interlayers.
Magnetron sputtering parameters for the Ti and W interlayers.
|Chamber preheat (°C)||400||250|
|Process pressure (mbar)||2 × 10−3||2 × 10−3|
|Deposition time (h)||3||3|
|Ar flow rate (sccm)||95||90|
|Cathode power (kW)||2 × 4||1 × 4|
|UBM coils current (A)||3||4|
|Coating thickness (μm)||2||1.5|
Observations of individual sprayed splats, metallographic cross-sections of the full coatings and fracture surfaces of the tested samples were performed in an EVO MA15 scanning electron microscope (Carl Zeiss, Oberkochen, Germany) in backscattered and/or secondary electron modes. Elemental analysis was performed by energy-dispersive spectroscopy (EDS) in the SEM, using an XFlash 5010 detector (Bruker, Berlin, Germany). Post-mortem analysis of the shear test samples was focused on the fracture location—either through the coating (coating cohesive failure), along the interface of coating-interlayer (coating adhesive failure) or along the interface of interlayer-substrate (interlayer adhesive failure). Detailed fractographic analysis of the fracture surfaces was also performed.