**1. Introduction**

Thermally sprayed oxide ceramic coatings have an outstanding importance in many technological areas. Coatings sprayed from the single oxides and some commercially available binary compositions of the Al2O3-Cr2O3-TiO2 system have multi-functional properties and are widely used as wear and as (sealed) corrosion resistant coatings. Depending on the composition, they are electrically insulating or conductive [1,2].

Plasma spraying and, in particular, conventional atmospheric plasma spraying (APS), is the most widely used thermal spray process for oxide coating manufacturing. Water stabilized plasma spraying (WSP) is a special high-energy process for coating manufacturing and is applied for special purposes, such as manufacturing of ceramic tubes. Coatings with excellent properties can be deposited by high velocity oxy-fuel spraying (HVOF), but often lower powder feed rates and deposition e fficiencies are considered as drawbacks. More detailed descriptions of the spray processes used for oxide coating manufacturing are given elsewhere [2–4].

Most commonly feedstock powders with a typical particle size in the range 15–45 μm are used. The main feedstock powder manufacturing method for the single oxides is fusing and crushing. For avoiding the formation of metallic chromium for Cr2O3 sintering and crushing is also common. For the commercially available binary compositions, the variety of the manufacturing methods are significantly broader. This includes mechanical blends of separately fused and crushed single oxide powders, jointly fused and crushed powders, but also agglomerated and sintered powders. Depending on the oxide particle or grain size, there are large di fferences in the homogeneity of mixing of the metallic elements in these methods. The homogeneity is lowest in case of the powder blends from single oxide powders and highest in the case of agglomerated and sintered powders, where finely dispersed oxide powders are mixed in the manufacturing process. Suspensions as feedstock become increasingly important, but are currently limited to single oxides [5]. Suspensions with two components are under investigation [6].

Each of the single oxides shows a specific material behavior during spraying, which is detrimental to the processing and/or coating properties [1,2].

For alumina, the detrimental phase transformation from α-Al2O3 (corundum) existing in the feedstocks to metastable phases, predominantly γ-Al2O3, in the coatings is well known [7–10]. The reason is the high cooling rate and nucleation of undercooled melt. Some content of the remaining α-Al2O3 in the coating is usually explained by the occurrence of non-molten particles [7]. An increase of the α-Al2O3 content in the coatings is described as an important measure to improve the coating properties, such as wear, electrical, and corrosion resistance. Except the addition of other oxides (e.g., Cr2O3 [8–10]), there are several technological measures for this, as the selection of special spray process conditions [11], use of suspensions [1,5] or heat post-treatments [12] as well as plasma-electrolytic oxidation of arc-sprayed and flame-sprayed aluminum coatings [13]. However, each of these technological measures has certain limitations and the stabilization of the α-Al2O3 by a tailored powder would be favored.

Chromia has a low deposition e fficiency due to oxidation from Cr2O3 to volatile CrO3, which immediately reconverts to Cr2O3 when cooling down [1,2,14]. Formation of Cr(VI)-oxyhydrates in wet atmospheres is another detrimental reaction [14,15]. Although the appearance of hexavalent chromium is below the maximum allowable concentrations under normal process conditions, increased safety regulations are permanently under discussion. By adding Al2O3 [16] or TiO2 [1,2] to Cr2O3 and forming respective solid solutions, both the formation of hexavalent chromium is suppressed and the deposition e fficiency is increased.

Since titania TiO2 readily loses oxygen in a reducing environment such as during the fusion step of feedstock manufacturing, fused and crushed feedstock powders are non-stoichiometric and preferably designated as TiO*<sup>x</sup>*. The oxygen content can also change during the spray process. The oxygen defects are often disordered in coatings under the strong nonequilibrium conditions of the thermal spray process, but also often forms ordered-structures for certain O/Ti ratios (Magnéli-phases) in feedstock powders, manufactured with lower cooling rates. Due to a eutectic with an oxygen content corresponding to *x* of about 1.78 in the Ti-O phase diagram (in the two-phase region of the Magnéli-phases Ti4O7 and Ti5O9). The temperature of appearance of a melt is decreased from 1857 ◦C for TiO2 down to 1679 ◦C at the eutectic point [2,17].

It was found that the addition of a second oxide can improve the coating properties [1,2]. There are some indications in the literature that the coating properties can be further improved by ternary compositions and by the addition of the third oxide of the system. Examples of such potential improvements can be the stabilization of target phases, an acceleration of the formation of a compound and improved sintering properties. This can relate to the feedstock powder manufacturing step and/or the spray process.

This addition can be made in di fferent ways, e.g., during the feedstock manufacturing process (joint processing in fused and crushed as well as agglomerated and sintered powder manufacturing) or by blending single oxide powders. The latter case is the simplest one, and blended powders have a high importance in the current industrial practice of thermal spraying. This is valid for commercially available blends in the Al2O3-TiO2 system, such as Al2O3-40%TiO2 and Al2O3-13%TiO2 [18,19], but also

for blending at the production site [20]. In general, an interaction between components of the blends will occur only during the spray process and is, in general, expected to be low due to the large particle size. However, an intensive interaction during the spray process was found in the case of an Al2O3-40%TiO2 blend [21]. Another example are blends of alumina-rich compositions of the Al2O3-Cr2O3 system, which can lead to a stabilization of the α-Al2O3 during WSP [10]. Blends of the binary Cr2O3-TiO2 system were investigated as well [17,20]. However, ternary compositions of the Al2O3-Cr2O3-TiO2 system were not studied neither as powder blends, pre-alloyed powders, nor suspensions.

In this work, the interaction of single oxide ternary powder blends of the Al2O3-Cr2O3-TiO2 system during APS and their sliding wear properties are studied. Coatings from the single oxides are investigated for comparison.

#### **2. Materials and Methods**

Commercial fused and crushed Al2O3, Cr2O3, and TiO*x* powder grades, compiled in Table 1, were used in this work. The particle size distribution was determined by laser diffraction analysis in a Cilas 930 device (Cilas, Orléans, France). To prepare the powder blends, dried powders were mixed using a tumble mixer, according to the compositions given in Table 2.




**Table 2.** Compositions of the powder blends.

Feedstock powders were investigated with a scanning electron microscope (SEM) LEO 1455VP (Zeiss, Jena, Germany) with an acceleration voltage of 25 kV. By using a backscattered electron detector (BSD), the material contrast is visualized by different grey levels. In addition, phase composition was studied by X-ray diffraction (XRD) using a Bragg-Brentano geometry operating with Cu *K*α radiation with a D8 Advance diffractometer (Bruker AXS, Billerica, MA, USA) in a range of *2*θ = 15◦–120◦ with a step size of 0.02◦ and 3 s/step.

Low carbon steel (S235) samples with a diameter of 40 mm were used as substrates, which were grit blasted with alumina (EK-F 24) (3 bar, 20 mm distance, 70◦ angle) and cleaned in an ultrasonic ethanol bath before applying the coating. The coatings were produced by atmospheric plasma spraying using an F6 torch (GTV, Luckenbach, Germany) and the spraying parameters, according to Table 3. In order to ensure good comparability, all coatings in this work were sprayed with this parameter set. The parameter set was chosen in such a way that a shift of the composition of the powder blends during processing was avoided. Interruptions of the coating process ensured that the substrate did not heat up to more than 200 ◦C.

The cross sections of the coatings were prepared by the standard metallographic procedure. The analysis of the microstructure was conducted using an optical microscope GX51 (Olympus, Shinjuku, Japan) equipped with a SC50 camera (Olympus, Shinjuku, Japan) as well as by SEM with the

same device used for powder analysis. The coating thickness was ascertained at 10 evenly distributed points. To determine the porosity, five images of the coating were evaluated by the image analysis method with the software ImageJ. Furthermore, the hardness of the coatings was measured on the cross sections using a Wilson Tukon 1102 device (Buehler, Uzwil, Switzerland). For this purpose, 10 Vickers indentations with a test load of 2.94 N were examined. The XRD patterns of the coatings were recorded with 3003 TT diffractometer from GE Inspection Technologies in a range of 2θ = 15◦–80◦ with a step size of 0.03◦ and 4 s/step. The high-resolution microstructure and the local chemical composition of individual splats in the coatings from blends were determined using a FESEM Ultra (Zeiss, Jena, Germany) equipped with an EDS detector X-Max80 and using a voltage of 13 kV. At least five measurements for the three typical individual splat compositions in each coating were performed. The calibration for the quantitative EDS analyses were performed using a cobalt standard.


**Table 3.** The parameters of the APS process with the F6-torch.

The processing of the coatings by grinding and polishing was investigated. All coated samples were simultaneously ground on grinding wheels up to a number of 1200 for 3 min at 300 rpm and 25 N. Polishing was then carried out with diamond suspensions down to 1 μm at 200 rpm and a slightly reduced force in order to obtain comparable surfaces for the wear tests. The roughness of as-sprayed and ground coatings was recorded with a tactile profilometer (Jenoptik, Jena, Germany) with five measuring tracks. The characterisation of the sliding wear behavior of the coatings was carried out using a ball on disc test (Tetra, Ilmenau, Germany) using the parameters summarized in Table 4. For each coating, three wear tracks were generated. The evaluation of the wear tracks with regard to the wear depth and the wear volume was carried out using a 3D profilometer MikroCAS (LMI, Teltow, Germany).

**Table 4.** Ball-on-disk test parameters.

