Erosion resistance of CMAS infiltrated sacrificial suspension sprayed alumina top layer on EB-PVD 7YSZ coatings

Wear 438–439 (2019) 203064

Contents lists available at ScienceDirect

Wear journal homepage: http://www.elsevier.com/locate/wear

Erosion resistance of CMAS infiltrated sacrificial suspension sprayed alumina top layer on EB-PVD 7YSZ coatings Lars Steinberg a, Christoph Mikulla b, Ravisankar Naraparaju b, *, Filofteia-Laura Toma c, Holger Großmann d, Uwe Schulz b, Christoph Leyens a, c a

TU Dresden, Institute of Materials Science (IfWW), D-01062, Dresden, Germany German Aerospace Center (DLR), Institute of Materials Research, D-51170, Cologne, Germany Fraunhofer Institute for Material and Beam Technology (IWS), D-01277, Dresden, Germany d Anton Paar Germany GmbH, D-73760, Ostfildern, Germany b c

A R T I C L E I N F O

A B S T R A C T

Keywords: TBC Solid particle erosion CMAS Alumina Suspension plasma spraying Erosion testing

The development of CMAS (CaO–MgO–Al2O3–SiO2) -resistant thermal barrier coatings is an urgent issue, as the operating temperatures of aero-engines exceed the melting point of the CMAS. Application of alumina as top layer on 7YSZ coatings has shown promising CMAS resistance behavior. It reacts with the CMAS and crystallizes into new stable phases, which seals the CMAS penetration paths in the top layer to save the underlying 7YSZ layer from infiltration. This paper deals with the study of the erosion behavior of porous thermally sprayed Al2O3 sacrificial layer on top of EB-PVD 7YSZ coatings before and after the CMAS infiltration and investigates the combined erosion/ corrosion regime. The TBC system has been infiltrated with three different CMAS compositions having a different Ca/Si ratio which resulted in different reaction scenarios and products. The change in the erosion behavior of the coating system as a function of CMAS infiltration depth was investigated and an erosion model has been pro­ posed. The key factors that dictate the erosion resistance of such alumina layers were found to be the thickness and morphology of the reaction layer, the porosity of the non-reacted alumina layer underneath the reaction layer as well as the coating adhesion mechanism between alumina and 7YSZ.

1. Introduction Higher operating temperatures in aero-engines are assured by the use of thermal barrier coatings (TBC) on the combustion chamber walls, as well as on the turbine blades and vanes [1]. The lifetime of TBCs is determined by several different extrinsic or intrinsic damage mecha­ nisms [2]. The state-of-the-art material used for the TBC is ZrO2 stabi­ lized with 7 wt.-% Y2O3 (7YSZ) having a tough t’-phase which is responsible for the mechanical stability as well as superior erosion resistance. Especially in sand-laden environments, the lifetime of TBCs is strongly dependent on the impact of small particles such as sand (CMAS; CaO–MgO–SiO2–Al2O3) which leads to erosion damage. However, cur­ rent engine operation temperatures, exceeding the melting point of CMAS (<1200 � C) or volcanic ash (VA) (<1050 � C), result in infiltrated TBCs causing an additional life threat by imposing chemical and me­ chanical degradation of standard 7YSZ [3,4]. Wellman et al. [5–9] as well as others [10–21] have thoroughly

studied the erosion behavior of TBCs at room temperature as well as at high temperature. The influence of CMAS infiltration on mechanical and chemical properties of TBCs has been a research focus of many groups [22–30]. Nevertheless, the combined impact of CMAS infiltration and erosion is less well understood. Wellman and Nicholls [31,32] have highlighted the change in erosion mechanism from individual column cracks towards a continuum behavior of the EB-PVD coating after CMAS infiltration, where cracks were found to propagate through several columns caused by the linking CMAS. Additionally, previous research by the authors [33] was focused on the influence of microstructure, CMAS infiltration time, and CMAS composition on the erosion behavior of EB-PVD 7YSZ coatings. Two different EB-PVD 7YSZ morphologies con­ sisting of different porous morphologies were tested. It was found that the more ‘Feathery’ 7YSZ layer (longer feather arms and smaller col­ umns) had a better erosion resistance compared to a ‘Normal’ (shorter feather arms and bigger columns) structure. However, under the influ­ ence of CMAS infiltration the effect was found to be reversed. In general,

* Corresponding author. E-mail address: [email protected] (R. Naraparaju). https://doi.org/10.1016/j.wear.2019.203064 Received 26 July 2019; Received in revised form 20 September 2019; Accepted 20 September 2019 Available online 21 September 2019 0043-1648/© 2019 Elsevier B.V. All rights reserved.

L. Steinberg et al.

Wear 438–439 (2019) 203064

Table 1 Summary of CMAS and Iceland VA compositions in mol.-%, main phases, viscosity and melting temperature based on data of [3,25,61,62] and (*) unpublished work. Mol.-%

CaO

MgO

Al2O3

SiO2

FeO

TiO2

Na2O

K2O

Main phases of the CMAS/VA deposits

Viscosity at 1250 � C in Pa. s

Melting range in � C

CMAS 1 ¼ C1 CMAS 2 ¼ C2 Iceland VA ¼ IVA

24.7 32.4 12.5

12.3 11.2 6.1

11.1 9.9 7.4

41.6 37.3 49.7

8.7 7.8 17.6

1.6 1.4 4.3

– – 2

– – 0.4

Pyroxene þ Anorthite Pyroxene þ Melilite Amorphous

6.9 4 292.9 (*)

1230–1250 1215–1245 1050–1200

completely CMAS-infiltrated EB-PVD 7YSZ TBCs exhibit a higher erosion resistance than the non-infiltrated ones due to continuum behavior of the TBC, as well as the reduction of the notching effect at the feather arms during the heat treatment. Application of CMAS resistant top coats on 7YSZ has been the current state-of-the art approach. CMAS resistant layers are classified in two major groups such as inert layers (Pd, Pt et al.) and sacrificial oxide layers e.g. Al2O3, TiO2–Al2O3, MgO, Sc2O3, Gd2Zr2O7 (GZO), 65 wt.-% Y2O3 balanced ZrO2 et al. [22,25,34–40]. While inert layers simply work as a non-reactive barrier, sacrificial layers chemically react with the CMAS. Components of the sacrificial layers that are dissolved into the CMAS melt generally modify the composition of the CMAS and hence the melting point and/or the viscosity of the CMAS. As a consequence, the modified CMAS undergoes changes in its crystallization behavior and precipitates into new phases with higher melting points. These new high temperature stable reaction phases such as apatite, garnet and spinel are formed, acting like a barrier against CMAS infiltration. Therefore, further infiltration of CMAS into the underlying TBC is inhibited [25,35,41–45]. However, the consumption of the sacrificial layer with respect to time defines the protective nature of such coating. Ideally, a sacrificial layer should react quickly with the CMAS and form the reaction products, which should not grow very fast. Dealing with reaction layers of different thicknesses, the mechanical properties of the reaction products are most important when erosion protection is considered. Such studies are rarely reported and one recent systematic study on the erosion behavior of a sacrificial double layer SPS GZO/8YSZ systems prior to CMAS infiltration [46] has shown that the single-layer SPS 8YSZ had a higher erosion resistance compared to SPS GZO/8YSZ systems. Furthermore, only a slight improvement in the erosion resistance in case of a reduced coating porosity was measured. However, its erosion behavior after the CMAS infiltration is not known. The porosity of the sacrificial layer and the additional interface between the TBC and the sacrificial layer had a great impact on the erosion behavior of the double layers. The reason for using alumina as a top layer is its comparable fracture toughness as well as the hardness of α-Al2O3 (bulk material: K1C ¼ 3.5–6 MPa m1/2, H300 mN ¼ 32.5 GPa) [47,48] with YSZ (bulk material: K1C ¼ 3 MPa m1/2, H200 mN ¼ 15.1 GPa) [49–51] even though, as coat­ ings these values may differ under the influence of microstructure and formed phases during the coating manufacturing procedure. Another reason is its better performance against CMAS degradation compared to that of 7YSZ [36,52,53]. Alumina has already been deposited as a CMAS resistant material on top of 7YSZ with different deposition techniques [35,52,54]. The most recent study was published by Naraparaju et al. [36], who deposited the alumina topcoat using EB-PVD technology. However, EB-PVD Al2O3-topcoats suffered locally from cracks that arise from crystallization and sintering shrinkage. As a consequence, the resistance against CMAS infiltration was insufficient due to the charac­ teristic morphology. It was found that the microstructure, the coating density and the distribution of the porosity were critical factors for the efficiency of sacrificial layers against CMAS infiltration and degradation. Furthermore, the reaction layer of EB-PVD alumina layer was found to be too thin [36] and their erosion behavior can not be evaluated as they can be removed very quickly during the erosion tests. As the Al2O3 microstructure is strongly influenced by the fabrication process of the material, new innovative coating methods are used to

create the sacrificial layer with the desired morphology. Over the last years, extensive development efforts have uncovered the potential of thermal spraying with suspensions. Coating thicknesses, morphologies and properties can be varied over an extremely wide range, as presented i.e. [55,56]. Suspension plasma sprayed (SPS) alumina coatings have a thickness range from a few micrometers up to 2.75 mm, as shown in Ref. [56]. However, the typical thickness of an alumina layer will depend on the application. Direct processing of nano- and sub-micron-sized powders is possible with suspensions, but more important is the advantage of directly using the finely dispersed ceramic oxide powders of widely varying grain size, purity, etc. currently used in the preparation of sintered technical ceramics. In this study, porous suspension plasma sprayed alumina top layers were used on top of EB-PVD 7YSZ layers. In the literature the correlation between an increasing erosion damage and increasing coating porosity is shown by erosion tests on APS coatings [14,57]. Nevertheless, a slightly higher porous SPS alumina was chosen in order to form a thick and dense reaction layer which can resist several erosion intervals to determine a steady state erosion rate of the reaction layer. The current investigation focused on identifying and understanding the erosion behavior and corresponding mechanisms of a sacrificial alumina top layer with and without CMAS infiltration on an EB-PVD 7YSZ TBC layer. Furthermore, the chemistry of CMAS has shown a big influence on the erosion resistance in our previous study [33] and therefore, two CMAS and natural Iceland volcanic ash were used for infiltration tests in SPS Al2O3. From the erosion behavior according to the type of CMAS-infiltration, an erosion model will be proposed. 2. Experimental procedure 2.1. Coating system A 200 μm thick 7YSZ TBC layer was deposited by EB-PVD on sintered alumina substrates with a size of (34 � 20 � 1) mm3 at DLR Cologne. The deposition parameters used to obtain the ‘Feathery’ EB-PVD microstructure were published elsewhere [58,59]. On top of the 7YSZ layer, a 80–90 μm thick sacrificial alumina layer was deposited by means of SPS using a water-based suspension at Fraunhofer IWS Dresden. The suspension containing 25 wt.-% of a commercially available Al2O3 raw powder (Martinswerk, Germany; d50 ¼ 2.2 μm) was fed by an three pressurized-vessels suspension feeder developed by Fraunhofer IWS. The suspension was externally injected in an APS F6 plasma gun (GTV Verschleiβschutz GmbH Luckenbach, Ger­ many) with 6 mm nozzle and Ar/H2 plasma gas mixture. 2.2. CMAS infiltration The CMAS infiltration behavior of a similar coating system was studied in a more detailed manner in Ref. [60]. On the basis of this study, two synthetic CMAS compositions namely CMAS 1, CMAS 2 as well as natural Iceland VA (IVA) were chosen for the infiltration tests at the DLR. The two CMAS compositions and their main characteristic properties such as crystallinity, viscosity and melting range are pre­ sented in Table 1. The variation of the chemical composition influences the viscosity as well as the melting range and therefore the infiltration behavior of the CMAS. The CMAS powders were mixed with water and 2

L. Steinberg et al.

Wear 438–439 (2019) 203064

The feeding rate was adjusted to 0.25 g/min (Powder feeder GTV PF2/1, GTV Verschleiβschutz GmbH). The distance between the nozzle (5 mm diameter) and the sample was 30 mm. The erosion parameters are stated in Table 3. Before and after every erosion interval, the samples listed in Table 2 were weighed (Mettler Toledo Analytical Balances XA105DU) to mea­ sure the mass loss. Long term erosion intervals of 8 � 60 s followed by 1 � 180 s and 3 � 300 s were used. Erosions tests were carried on two samples for each condition to get reliable data. If the coating system failed prematurely, the erosion test was stopped. Especially the alumina layer in case of ‘as-coated’, ‘aged’ and ‘C1’ failed in the first intervals and therefore, shorter intervals than the original 60 s were required during the initial stage of erosion. Hence, further short term erosion tests were carried out on only one sample for each mentioned condition (different to the previous test) using 8 � 15 s intervals in order to detect the initial erosion behavior. All the investigated TBC samples have different porosity and density levels originating from the different manufacturing processes, coating materials, and presented different CMAS infiltration behavior. Conse­ quently, it is more appropriate to determine the erosion volume loss of the coating than to define the erosion rate in terms of coating mass loss vs. impinging erodent mass. Therefore, the coating volume loss was measured during the erosion test by confocal microscopy (μscan; Nanofocus) and used for a comparison of the TBC samples among each other. Coating mass loss was only used for the comparison with litera­ ture data if necessary.

Table 2 Nomenclature and reaction products [60] of the tested SPS Al2O3/EB-PVD 7YSZ samples under different conditions. Sample designation

Type of CMAS

Heat treatment

Reaction products of the CMAS/VA and SPS alumina layer after heat treatment

‘as-coated’ ‘aged’

– –

– –

‘C1’

CMAS 1

‘C2’

CMAS 2

– 300 min/ 1250 � C 300 min/ 1250 � C 300 min/ 1250 � C

‘IVA’

Iceland VA

spinel anorthite spinel anorthite gehlenite spinel anorthite pseudobrookite

300 min/ 1250 � C

Table 3 Erosion conditions of the used test rig. Erosion test rig operating conditions

Current conditions

Temperature Erodent Particle Size in μm

RT irregular-shaped corundum d10: 54.5 d50: 92.5 d90: 142.7 125 90� 0.25 30 5

Nominal velocity in m/s Impact angle Feed rate in g/min Distance nozzle-sample in mm Nozzle diameter in mm

2.4. Nano-indentation measurements The nano-indentation tests were performed by a NHT3 nanoindentation tester (Anton Paar Germany GmbH, Ostfildern, Germany) at the Technische Universit€ at Dresden. The tests were conducted on polished cross-section prepared samples using a large array of in­ dentations (10 � 8; stepsize 30 μm). This was done to study the nano hardness and Young’s modulus as a function of distance to the substrate/ EB-PVD 7YSZ interface with respect to the infiltration depth. A Berko­ vich diamond tip was used for these measurements. The Poisson’s ratio value for Al2O3, which is necessary for the calculation of the nano-indentation values, is found to be between 0.3 and 0.23 [47,63,64]. Asmani et al. [65] have found that the Poisson’s ratio decreases with increasing porosity. Since the Poisson’s ratio of CMAS is 0.26 [66] and CMAS infiltration reduces the porosity in the alumina layer, a Poisson’s ratio for the reaction layer as well as the alumina layer of 0.25 was considered. The Poisson’s ratio for 7YSZ was found to be in between 0.3 [67,68] and 0.1 [69] for EB-PVD TBC at room temperature. Due to the high anisotropy of the Poisson’s ratio for EB-PVD coatings depending on the crystallographic orientation, Fujikane et al. [51] has recommended an average Poisson’s ratio of 0.25. This value is in accordance with the Poisson’s ratio given by Jackson et al. [70] for CMAS infiltrated EB-PVD 7YSZ TBC and was used for calculations in this study. Since the Young’s modulus of a porous material is only valid in the very limited range of linear-elastic behavior, an increasing load starting with 25 mN was applied at each position and the Quasi Continuous Stiffness Measurement method was used [71]. In order to minimize the influence of underlying porosity and column boundaries, as well as any lateral bending of the columns during indentation, described by Well­ man [49], only the results of the minimum load of 25 mN are presented in this paper.

Table 4 EDS spot measurements to verify the partly CMAS 2 infiltration in ‘C2’ in Fig. 13. Spot Number 1 2 3 4 5 6

Chemical composition in at.-% Ca

Mg

Al

Si

Fe

Y

Zr

O

0.8 – 0.2 2.0 1.7 1.5

– – 0.3 – 0.3 –

0.3 – 0.8 3.7 2.4 2.3

0.3 – – 2.9 1.9 1.8

– – – – 0.4 –

2.2 2.8 1.4 1.3 1.6 1.7

32.5 33.4 32.0 25.2 28.0 28.1

63.9 63.8 65.3 64.9 63.7 64.6

the paste was applied in a concentration of 10 mg/cm2 on the Al2O3 coating surface and then dried at room temperature. The subsequent infiltration heat treatment at 1250 � C in ambient air lasted 300 min. At this temperature (and during this time) the CMAS melts interacts with the coating, i.e. it both infiltrates into the coating and reacts with it. After the treatment, the furnace was cooled down at a rate of 10 K/min. These longer annealing times were considered to have a thick reaction layer such that they survive few erosion intervals to assess the properties of the reaction products. Additionally, a set of samples (without CMAS on top) were aged, using the same heat treatment parameters (300min, 1250 � C, ambient air). Table 2 presents the samples nomenclature, treatments and reaction products. 2.3. Erosion tests The erosion tests were carried out using the erosion test rig of the €t Dresden which was described in detail in a Technische Universita previous publication [33]. The tests were carried out at room temper­ ature using irregular-shaped corundum particles with a size distribution of 54.5 μm (d10), 92.5 μm (d50) and 142.7 μm (d90) (Cilas, Cilas 1064). The particles impinged the coating at an angle of 90� and with a mean velocity of about 125 m/s (measured by a PIV system, PyroOptic Aps).

2.5. Microstructural characterization The samples were analyzed under a scanning electron microscope (SEM: DSM 982 Gemini, Zeiss, Germany). The CMAS infiltration zone was detected by analyzing the SEM-images in combination with energy dispersive X-ray-analysis (EDX: Noran System 7, Thermo Scientific). The 3

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 1. Cross-sectional SEM-images of the microstructure of samples of the SPS Al2O3/EB-PVD 7YSZ coating system and its infiltration depth/reaction layer for the case of CMAS and IVA infiltration. The dash lines represent the interface zones between infiltrated/non-infiltrated regions: a) ‘as-coated’ coating system; b) noninfiltrated coating system after 300 min/1250 � C heat-treatment (‘aged’); c) after CMAS 1 infiltration (‘C1); d) after CMAS 2 infiltration (‘C2‘); e) after Iceland VA infiltration (‘IVA’).

the ‘Feathery’ 7YSZ TBC (Fig. 1 a) is characterized by a high porosity [59,62]. The different compositions of CMAS used in this study have led to different reaction layers/infiltration depths, which are presented in Fig. 1 (c and d) [60]. CMAS 1 infiltrated almost 25% of the alumina layer and formed a thin, dense reaction layer, which stopped further CMAS infiltration (Fig. 1 c). Residual CMAS solidified on top of the Al2O3 layer. EDS spot analysis, which is not shown here, confirmed that the CMAS has not penetrated beyond the white dash line. The thickness of the CMAS 2 reaction/infiltrated layer varied between areas of fully infil­ trated alumina and 7YSZ coating system, and areas with partly infil­ trated/reacted alumina layer with non-infiltrated alumina and 7YSZ underneath (Fig. 1 d). In contrast, the Iceland volcanic ash (IVA) has infiltrated the alumina and 7YSZ coating system completely (Fig. 1 e). The surface morphology of as coated sacrificial Al2O3 coating is presented in Fig. 2. The coating mainly consists of small splats (1–6 μm) which are shown in point 1as well as non-molten particles (point 2). The coating contained an in-plane porosity of 32 � 4%. 3.2. Erosion results

Fig. 2. Surface microstructure of as-sprayed SPS Al2O3 coating: 1) deformed splats; 2) non-molten particles; 3) overspray clusters.

The steady state erosion rate is usually used as a base to estimate the erosion behavior of any TBC coating [5,11,12]. For a complete assess­ ment, however, the time of occurrence of coating failure and its corre­ sponding total volume loss has to be considered. The coating volume loss depends on the mass flow rate of the erodent as well as erosion time (erosion intervals: 8 � 60 s followed by 1 � 180 s and 3 � 300 s). The results for as-coated and infiltrated TBC system are presented in Fig. 3. With regards to the total coating volume loss at the time of occurrence of coating failure, the Iceland volcanic ash infiltrated sample, subsequently called ‘IVA’ (8.07 mm3 after 1560 s), showed the best erosion resistance. The high total coating volume loss of the CMAS1 infiltrated sample, subsequently, ‘C1’ even exceeds those of the ‘as-coated’ and ‘aged’ samples, which are similar. The ‘aged’ and ‘C2’ coating systems

porosity was measured on five SEM images at magnifications of 3000� using ImageJ software. The erosion damage of the coatings was observed in SEM images taken after the individual erosion test. 3. Results and discussion 3.1. Microstructural characterization of the SPS alumina/EB-PVD 7YSZ coating systems before and after CMAS infiltration The SEM cross-section images in Fig. 1 show the microstructure of the coating with and without CMAS infiltration. The microstructure of 4

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 3. a) Erosion data for long erosion intervals with coating volume loss plotted against erosion time as well as mass erodent exposure (slope of linear fitted erosion data as steady state erosion rate Es of stage (i) or (iii)) and marking of the beginning of the alumina layer spallation by dashed lines in combination with b) the corresponding (to the dashed lines) top view images of the coatings.

Fig. 4. Erosion data for short erosion intervals with coating volume loss plotted against erosion time as well as mass erodent exposure (slope of linear fitted erosion data as steady state erosion rate ES). 5

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 5. a) Nano Hardness (H), b) Young’s modulus (E) of the nano-indentation test of the infiltrated samples plotted against distance to the substrate/EB-PVD 7YSZ interface; sub-divided into three different regions (i) alumina (reaction-) layer, (ii) 7YSZ/alumina interface and (iii) 7YSZ layer.

delaminate from the substrate after 660 s and are, therefore, the first samples to fail, followed by ‘as-coated’ and ‘C1’ (Fig. 3). As the tested samples have an alumina layer on top of the 7YSZ layer, the erosion process can be sub-classified into three different stages: (i) surface erosion and initial spallation of alumina layer, (ii) mixed erosion mode with alumina spallation and initial 7YSZ erosion, and (iii) domi­ nated by 7YSZ erosion followed by delamination and failure of the TBC. The residual CMAS on top of the Al2O3 layer is highly brittle and is usually removed during the initial erosion time intervals. The vertical dashed lines in Fig. 3 a) indicate the beginning of alumina delamination and the transition of stage (i) towards stage (ii). Fig. 3 b) shows the top view images of the coatings directly after the erosion interval in which the first complete alumina delamination appeared. While the large size of the erosion spot of ‘as-coated’/‘aged’ (black/green) as well as ‘C1’ (red) indicates a spallation of the alumina coating, the small erosion spot of ‘C2’ (purple) and ‘IVA’ (blue) refers to a progressive removal. Stage (iii) follows after a constant coating volume loss per erosion interval is achieved. The steady state erosion rate was determined for the stages (i) (ES(i)) and (iii) (ES(iii)) and is shown in Fig. 3. The stages are described more precisely in section 3.4. Since the spallation of ‘as-coated’, ‘aged’ and ‘C1’ alumina layers starts in the first two erosion intervals (dashed lines in Fig. 3), further erosion tests with shorter intervals of 15 s were done in order to detect the initial erosion behavior (Fig. 4). The ‘as-coated’ and ‘aged’ samples started to spall during the first 15 s and therefore, in erosion stage (i) no steady-state erosion rate could be determined.

where a maximum lateral displacement of the columns is possible. In the case of the ‘IVA’, the complete coating system (7YSZ and alumina layer) was infiltrated uniformly. As a result uniform high nano hardness and Young’s modulus values were measured over the coating thickness (average values: H(iii) IVA ¼ 12.7 GPa; E(iii) IVA ¼ 173 GPa). The nano-indentation values of the ‘C2’ in region (iii) vary signifi­ cantly (H(iii) C2 ¼ 6.4–12.8 GPa; E(iii) C2 ¼ 98.4–166 GPa). As the 7YSZ layer was not fully infiltrated, the non-infiltrated 7YSZ layer exhibits similar values as the non-infiltrated 7YSZ layer of the ‘C1’ sample whereas values of the infiltrated 7YSZ layer are comparable to the ‘IVA’ one. (ii) EB-PVD 7YSZ/SPS alumina interface ‘IVA’ showed the highest nano-indentation values at the 7YSZ/ alumina interface (H(ii) IVA ¼ 25.5 GPa; E(ii) IVA ¼ 329 GPa). In contrast to that, the values for ‘C2’ (H(ii) C2 ¼ 12.0 GPa; E(ii) C2 ¼ 166 GPa) and ‘C1’ (H(ii) C1 ¼ 10.0 GPa; E(ii) C1 ¼ 110 GPa), were significantly lower. (i) SPS alumina (reaction) layer In all the infiltrated cases, the alumina (reaction) layer shows an increase in nano hardness and Young’s modulus compared to that of the 7YSZ layer. These increased values reflect the lower porosity due to infiltration of the alumina layer and the superior hardness of Al2O3 itself compared to 7YSZ. The highest values were reached directly at the CMAS-alumina reaction interface. Only sample ‘C1’ could be measured at 310 μm, which is above the 300 μm thickness of the coating system. This was due to the rapid formation of the reaction layer, which sealed the alumina surface and led to a solidified CMAS layer on top of the alumina layer. Higher values were measured on the ‘IVA’ samples (H(i) IVA ¼ 25.3 GPa, E(i) IVA ¼ 299 GPa) compared to ‘C2’ (H(i) C2 ¼ 19.0 GPa, E(i) C2 ¼ 257 GPa) and ‘C1’ (H(i) C1 ¼ 15.2 GPa, E(i) C1 ¼ 194 GPa). The accuracy of the values measured in the ‘C1’ samples at 250 μm were limited by the extremely high porosity of the non-infiltrated part of the alumina layer below the reaction layer, which strongly influences the measurement.

3.3. Nano-indentation The nano-indentation results of the infiltrated layers are sub-divided into three different regions, which correspond to the different erosion stages presented in section 3.2 and named in the same manner: (i) alumina (reaction-) layer, (ii) 7YSZ/alumina interface and (iii) 7YSZ coating (Fig. 5). (iii) EB-PVD 7YSZ layer Due to the formation of a protective reaction layer after CMAS 1 infiltration, the 7YSZ coating underneath the SPS alumina was not infiltrated by CMAS. On this non-infiltrated 7YSZ coating the lowest range of nano-indentation values (H(iii) C1 ¼ 4.5–8.0 GPa; E(iii) C1 ¼ 77–125 GPa) were measured. The lowest value was measured in the middle of the 7YSZ layer at a distance of 130 μm from the substrate,

3.4. CMAS infiltration of the coating system The CMAS infiltration behavior in alumina layers has been investi­ gated intensively [60]. Nevertheless, in order to understand the erosion behavior of sacrificial alumina layers it is important to discuss the 6

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 6. SPS alumina/EB-PVD 7YSZ interface and change of its adhesion mechanism based on CMAS and IVA infiltration a) ‘as-coated’, b) ‘C1’, c) ‘C2’, and d) ‘IVA’.

progressing spallation. The small particles used in the SPS process enabled the splats to adapt to the roughness of the EB-PVD tips as shown in Fig. 2 b). These particles were also found in between the EB-PVD columns (Fig. 6 a). From this observation it can be concluded that the mechanism behind coating adhesion strength is the mechanical inter­ locking [72,73]. During the spraying process, columns tips were not damaged (Fig. 6 a). Since no CMAS was found underneath the reaction layer, the alumina/7YSZ interface of the ‘C1’ sample (Fig. 6 b) was similar to the ‘aged’ sample. In both cases, the feather arms of the 7YSZ showed sin­ tering effects after the annealing step, but no significant changes in the non-infiltrated alumina layer (Fig. 6 b). Therefore, bond strength is ex­ pected to be comparable to the ‘as-coated’ sample, in which the strength was derived from the mechanical interlocking. If CMAS infiltrates through the alumina layer into the 7YSZ TBC, an additional term related to chemical bonding is added to the mechanical interlocking component. This improves the coating’s adhesion strength [72,73]. In the case of ‘C2’, areas which were fully infiltrated by CMAS 2

changes in the alumina microstructure after CMAS infiltration. The principal function of the sacrificial layers is to form a protective reaction layer, which seals the TBC layer from further CMAS infiltration. The general mechanism behind any reactive TBC layer is that the dissolution and re-precipitation of the reactive layer within the CMAS melt induces the formation of new crystalline phases which prevent the CMAS from infiltrating into the unreacted coating by sealing gaps and cracks [35,36, 60]. The reaction products in case of alumina with different CMAS compositions are listed in Table 2, whereby only spinel has been shown to have sealing properties [36,60]. Fig. 1 shows that only in case of ‘C1’ a uniform spinel-containing reaction layer was rapidly formed, sealing the gaps and protecting the underlying TBC from further CMAS infiltration. The reaction products of ‘C2’ and ‘IVA’ showed less sufficient sealing properties. Further information about the CMAS infiltration behavior of alumina coating can be found in the literature [36,60]. Fig. 6 shows the alumina/7YSZ interface of all the tested cases in high magnification. Good adhesion strength of the SPS alumina layer at the 7YSZ interface is essential in order to prevent the alumina layer from

Fig. 7. Stereomicroscope top view images: enlargement of the erosion spot with increasing erosion time of ‘C1’ sample. 7

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 8. Top view SEM-images of the change of SPS alumina microstructure caused by a compaction process on ‘as-coated’ during erosion a) microstructure before erosion, and b) microstructure after erosion.

(Fig. 6 c) as well as the fully infiltrated ‘IVA’ (Fig. 6 d), it is expected that the coating adhesion strength is improved. The chemical bonding be­ tween all phases present at the former interface between alumina and 7YSZ contributes to changes in Young’s modulus. Since the mechanical interlocking effect is supposed to be equal for all samples, a low Young’s modulus may indicate a trend towards a weak point at the interface of the coating system during erosion. However, it must be noted that the Young’s modulus depends on the amount and shape of the porosity

which was more pronounced for the ‘C1’ samples [65,74,75]. As shown in Fig. 6 b), the bond strength of the ‘C1’ interface is based only on mechanical interlocking (H(ii) C1 ¼ 10.0 GPa; E(ii) C1 ¼ 110 GPa), while ‘IVA’ exceeds this by adding a chemical bonding component due to the process of interface infiltration with IVA (H(ii) IVA ¼ 25.5 GPa; E(ii) IVA ¼ 329 GPa). The non-uniformly infiltrated ‘C2’ interface results in measured hardness values between those obtained for the samples ‘IVA’ and ‘C1’ (H(ii) C2 ¼ 12.0 GPa; E(ii) C2 ¼ 166 GPa).

Fig. 9. a–c) SEM images of the ‘as-coated’ cross section in the individual stages during erosion of the alumina coating; d) confocal microscopy on the surface of the erosion spot of the ‘as-coated’ condition after 15 s of erosion. 8

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 10. SEM cross-sectional images of a) vertical crack propagation through residual CMAS and the reaction layer (white line) of ‘C1’ after successive removal by surface spallation (<40 μm), b) spallation of the reaction layer due to the interaction of different erosion mechanisms.

Fig. 11. Removal of fractured regions of the reaction layer of ‘C1’ after less than 60 s of erosion.

9

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 12. Erosion behavior of ‘IVA’ a) surface spallation, b) interfacial cracks through alumina layer and EB-PVD 7YSZ tips, c) successive removal of the infiltrated alumina layer.

3.5. General erosion behavior of different individual erosion stages of SPS alumina/EB-PVD 7YSZ coating system

example. The 3-D Gaussian distribution of the eroding particles results in a more concentrated erosion in the center portion of the impact. In case of eroding multilayer systems, it leads to a composite erosion sit­ uation which makes it difficult to separate the effects of the different layers. So once the first layer is penetrated in the center of the erosion spot the second layer starts eroding while the periphery of the erosion spot is still eroding the first layer. Since the periphery of the erosion zone has seen significantly lower particle flux than the center, the local erosion rate is lower. This results in concentric circles (layers) whose diameters grow with time. The particular erosion behavior of different infiltration scenarios are described in detail in section 3.6.

A visualization of the different stages of the erosion process of infiltrated alumina/7YSZ layers is shown in Fig. 7 on the ‘C1’ as

(i) Surface erosion and initial spallation of SPS alumina layer During stage (i), the residual CMAS and the reaction layer, which forms during the initial reaction/infiltration of CMAS with the sacrificial alumina layer, was progressively removed. The thickness of residual CMAS and the reaction layer is decisive for the erosion resistance in this stage. Based on the comparison between the different erosion processes of ‘IVA’ and ‘C1’ samples it can be assumed that residual CMAS and the reaction layer are gradually removed to the point where the layer no longer withstands the load of the particles. At this point vertical cracks are believed to be initiated in the remaining residual CMAS/reaction layer and propagate through them. This process increases the load on the underlying porous non-reacted/non-infiltrated alumina layer. At this point, the mode of erosion changes to compaction and significant spallation of the reaction layer occurs, leading to stage (ii) erosion. The infiltrated alumina layers have comparable steady state erosion rates throughout all CMAS compositions (ES(i) ¼ 1.25–2.96 mm3/g). These rates were found to be much lower than the rates of non-infiltrated alumina which clearly indicates an enhanced erosion resistance due to infiltration and reaction between alumina and deposits. However, in the case of the non-infiltrated ‘as-coated’ and ‘aged’ samples, a compaction

Fig. 13. SEM-micrograph with EDS mapping (calcium element, red) to detect the CMAS 2 infiltration behavior (white line) into coating system (‘C2’) including EDS spot measurements (Table 4) before erosion testing. The white dotted lines separate infiltrated from non-infiltrated regions. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) 10

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 14. SPS alumina/EB-PVD 7YSZ interface failure of ‘C2’ during erosion and the formation of SPS alumina CMAS infiltrated islands.

and fast spallation of the layer was observed within the first 15 s (Fig. 4) due to the missing reaction layer. Obviously, for both the latter samples stage (i) could not be identified even when using short erosion intervals.

erosion simultaneously, which is demonstrated by the enlargement of the erosion spot (Fig. 7; 120 s–360 s). During this entire stage, the alumina layer and the 7YSZ coating were simultaneously eroded. It is characterized by a non-constant erosion rate, which is initially high (dominated by alumina spallation) and decreases over time (dominated by 7YSZ erosion). As a consequence, the erosion rate cannot be divided into a fixed proportion due to the mixed alumina and 7YSZ removal. For this reason, no steady state erosion rate could be determined for this stage. The adhesion at the alumina/7YSZ interface is decisive for the

(ii) Mixed erosion mode of SPS alumina spallation and EB-PVD 7YSZ erosion After the initial erosion stage a mixed erosion mode (stage ii) follows, consisting of fast and progressive removal of the alumina layer and 7YSZ

Fig. 15. Erosion mechanisms in stage (iii) of ‘C2’ a) non-infiltrated: individual column crack, b) infiltrated EB-PVD 7YSZ areas: cracks propagated through several columns; c) different erosion depth of non-/infiltrated EB-PVD 7YSZ areas. 11

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 16. Top-view images showing the higher erosion depth of infiltrated compared to non-infiltrated EB-PVD 7YSZ coating areas in ‘C2’ after erosion exposure for 660 s: a) stereomicroscope image, b) confocal microscope image.

course of this erosion stage, since a low adhesion allows the alumina layer to spall off easily and thus causes a larger erosion rate. Without complete infiltration of the alumina layer, this adhesion is mainly determined by mechanical interlocking of alumina particles with 7YSZ columns as shown in Fig. 6. However, this has been changed in case of C2 and IVA where CMAS infiltration/reaction has added chemical bonding, which can hugely change the erosion mechanism in this stage as already mentioned in section 3.4.

alumina layer. Therefore, the erosion mechanisms of both ‘as-coated’ and ‘aged’ samples are illustrated as scenario (1) lately in Fig. 17. The impact of the erosive particles on the highly porous alumina layer leads directly to a compaction damage (Figs. 8 and 9 a). Eaton et al. [76] described a similar mechanism for highly porous APS coatings as the formation of crack networks interconnecting the pores results in a high erosion rate. The bonding between the splats or the splats themselves are directly broken as shown by red arrows in Fig. 8. Since the Young’s modulus and the fracture toughness are directly dependent on the bonding interphases between the splats, the erosion resistance decreases due to their cracking [75]. In the area of the highest stress during the impact of the particle a delamination crack is initiated, which leads to the spallation of large areas of the alumina. A delamination crack and the resulting spalled area of the compacted layer are shown in Fig. 9 b. If the majority of the alumina layer is spalled off, the underlying 7YSZ columns are exposed directly to the erosion medium as shown in Fig. 9 c. The nano hardness and the Young’s modulus for the alumina/ 7YSZ interface of the ‘C1’ sample (Fig. 5) can be used to explain the erosion mechanism in this region, since this is also a non-infiltrated interface. The low mechanical strength, which represents the weak interlocking, can be assumed to be the cause for an early failure of the alumina layer in this area upon erosion. Fig. 9 d) shows a 105 μm deep erosion spot of the ‘as-coated’ sample after 15 s. Since the alumina layer has a thickness of 90 μm, the mech­ anism of compaction damage and alumina/7YSZ interface failure occurs in stage (i) and was completed within the first 15 s. In this stage, as well as in stage (ii), no significant differences between the coating volume loss of the ‘as-coated’ and ‘aged’ alumina can be seen (Fig. 4). Only in stage (iii), due to the slight formation of sinter bridges between the 7YSZ columns an increased erosion rate of the ‘aged’ sample occurs, as these bridges allow the cracks to propagate over several columns. For this reason, the steady state erosion rate for this stage increases from ES(iii) as3 3 coated ¼ 2.57 mm /g to ES(iii) aged ¼ 3.67 mm /g and the time to delamination decreases from 960 s to 660 s (Fig. 3).

(iii) Dominating EB-PVD 7YSZ erosion followed by delamination and failure of the TBC This stage begins after the alumina layer has already been removed in the area of the main erosion spot. In this stage the dominant erosion mechanism relies on the properties of the infiltrated or non-infiltrated 7YSZ coating and were already described in detail [33]. In the case of a non-infiltrated coating, separate columns were individually fractured. While the reported erosion rate of as-coated ‘Feathery’ EB-PVD 7YSZ samples in Ref. [33] is 1.25 mm3/g, the steady erosion rate in erosion stage (iii) of the ‘as-coated’ coating system in this work is found to be 2.57 mm3/g. Though the erosion of the 7YSZ layer dominates at the erosion stage (iii), Al2O3 is still eroded to a lesser degree, which can be seen from a further growth of the erosion spot in this stage. This further enlargement of the erosion spot is similar to the ‘C1’ top view images after 360 s and 960 s of erosion in Fig. 7. This combined erosion is responsible for the different erosion rates. In the case of infiltrated 7YSZ, the crack is found to be initiated on the coating surface or feather arm and propagates into the next column. As a result, the crack propagates through several columns before material is removed. A special case is the partially infiltrated ‘C2’ sample, which is described in section 3.6 scenario (4) in more detail. 3.6. Erosion model -influence of the reactive layer and CMAS infiltration depth on erosion behavior of SPS alumina/EB-PVD 7YSZ TBC-

3.6.2. Scenario (2) - 25% infiltration of SPS alumina (CMAS 1) Scenario (2) was achieved in the case of C1, where a reaction layer and residual CMAS of 50–60 μm thickness has formed after CMAS infiltration. It was then successively removed by surface spallation during the initial erosion process by forming a crack network near the surface within the resulting residual CMAS or reaction layer, which is illustrated in Figs. 10 a and 17 scenario (2). After a certain erosive load, this network results in vertical cracking (Fig. 10 a). These cracks result in a fragmentation of the continuous reaction layer into many small areas. While a continuous layer causes a load distribution over a large surface

In the following, different scenarios are described which were observed experimentally. Respective erosion mechanisms mentioned in section 3.5, e.g. compaction damage (Fig. 9 a) and cracking of the 7YSZ layer (Fig. 10), are now described in detail. A model description of the four scenarios identified in this work is given in Fig. 17. 3.6.1. Scenario (1) - Non-infiltrated SPS alumina/EB-PVD 7YSZ system The ‘as-coated’/‘aged’ samples have no reaction layer and have shown a lower erosion resistance. A heat treatment at 1250 � C for 300 min has not caused a significant change in the microstructure of the 12

L. Steinberg et al.

Wear 438–439 (2019) 203064

Fig. 17. Erosion model showing the expected erosion behavior according to the CMAS-infiltration depth: SPS Al2O3 layer – gray area with white dots; EB-PVD 7YSZ layer – gray columns; CMAS infiltrated area – yellow area; Cracks in the coating – red lines. Corresponding relative erosion rate is represented by a color bar (low – green; medium – yellow; high – red). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

area during the particle load, fractured layers cannot deflect the load into other areas. This leads to higher pressure underneath the fractured regions. As a consequence, the already described compaction mecha­ nism takes place in the volume of the non-infiltrated alumina layer underneath. Subsequently, spallation of the fractured reaction layer occurs at a remaining reaction layer/residual CMAS thickness of about 30 μm (Fig. 10 b). This results in the remaining alumina layer failing within 60 s of further erosion (Figs. 4 and 11) and thus a transition of stage (i) towards stage (ii). Since the 7YSZ of ‘C1’ is not infiltrated, ‘C1’ alumina/7YSZ interface and 7YSZ of ‘C1’ have the same thermal and chemical history, as the ‘aged’ sample, the erosion process and the predominant erosion mech­ anisms of stages (ii) and (iii) are similar. This can be seen from their

steady state erosion rates of ES(iii) 4.36 mm3/g (Fig. 3).

aged

¼ 3.67 mm3/g and ES(iii)

C1

¼

3.6.3. Scenario (3) - complete infiltration of the SPS alumina/EB-PVD 7YSZ system (IVA) Compared to CMAS 1, IVA has only formed a thin un-protective re­ action layer allowing the melt infiltration through the entire alumina layer, which in turn infiltrated further through the 7YSZ layer. There­ fore, it is an example of scenario (3). Although the alumina could not act as an infiltration barrier, it is important to understand the erosion mechanism of this semi reacted/fully infiltrated layer. Especially, the compaction damage of the porous alumina layer is eliminated and the bond strength at the alumina/7YSZ interface is enhanced by an 13

L. Steinberg et al.

Wear 438–439 (2019) 203064

additional chemical component (Fig. 17 scenario (3)) due to the infiltration. The erosive process begins with surface spallation, the same mech­ anism already mentioned for ‘C1’ (Fig. 12 a). Due to this erosion mechanism, the fully infiltrated alumina layer undergoes a successive removal up to the alumina/7YSZ interface (Fig. 12 c). No vertical cracks in the alumina layer could be observed as seen in the case of C1. This is due to the absence of a porous alumina layer with a low Young’s modulus as shown in Fig. 5. Therefore, the stiffness of Al2O3 layer is higher compared to ‘C1’. Hence, ES(i) IVA is smaller than ES(i) C1, whereby the higher hardness value of the infiltrated ‘IVA’ layer also has a positive influence on the higher erosion resistance (Fig. 5). The interface between Al2O3 and 7YSZ was strengthened by an additional chemical reaction where a possible dissolution and reprecipitation process of 7YSZ in to the melt takes place. The measured nano-indentation values also show a clear strengthening of the me­ chanical properties at this interface (H(ii) IVA ¼ 25.5 GPa; E(ii) IVA ¼ 329 GPa compared to H(ii) C1 ¼ 10 GPa; E(ii) C1 ¼ 110 GPa). As a result, sudden spallation of the alumina layer at the interface was prevented. Fig. 12 b shows that cracks do not propagate along the interface, but through the alumina layer and 7YSZ columns tips equally. The subsequent erosion of the fully infiltrated 7YSZ layer in stage (iii) was already described in Ref. [33]. In this case, the 7YSZ coating now behaves like a continuum material, which means that individual columns are no longer cracked, but the cracks can propagate through several columns. Material is only removed when the end of the crack reaches a surface. As already described in Ref. [33] and shown in Fig. 3, the erosion resistance was usually improved for infiltrated 7YSZ. During the erosion test the erosion rate of ‘IVA’ in stage (i) and (iii) hardly change (ES(i) IVA ¼ 1.65 mm3/g; ES(iii) IVA ¼ 1.13 mm3/g) as shown in Fig. 3 a. In comparison to previous results of full infiltrated ‘Feathery’ 7YSZ TBC (Es ¼ 6.95 g/kg) [33] the complete infiltration of the 7YSZ/Al2O3 coating system with IVA (Es(iii) IVA ¼ 4.72 g/kg) reduced the coating mass loss based erosion rate by 32%. However, in realistic conditions established thermal gradient would not allow a complete infiltration in the TBC unless the engine runs extremely hot.

compared to the other scenarios. Since the alumina/7YSZ interface was partially infiltrated and chemically reacted with CMAS, the interface could not be entirely excluded as a weak point during erosive loading. Fig. 14 shows that in the alumina layer around 50 μm deep vertical cracking occurs, which is comparable to that of ‘C1’. Furthermore, in the non-infiltrated areas the initiation of cracks took place directly at the interface region. However, the EB-PVD 7YSZ columns tips were not damaged. Due to the higher adhesion of the infiltrated areas (as explained in section 3.3), alumina islands with increased erosive resistance (see Fig. 14) were formed during the erosion process. In stage (iii) a mixed erosion process occurs which, according to the knowledge of the authors, has not yet been described in the literature. While crack growth in the non-infiltrated areas is limited to individual columns (Fig. 15 a), in infiltrated areas cracks can propagate through several columns (Fig. 15 b). The horizontal cracks in the infiltrated areas quickly reach the vertical column gap in the non-infiltrated area, leading directly to material removal. This in combination with the strong degradation of 7YSZ by CMAS 2 described in Ref. [33] leads to the high erosion rate of the infiltrated 7YSZ coating (Fig. 15 c). The stereomi­ croscope image in Fig. 16 a) shows the erosion spot of ‘C2’ after 660 s of erosion. The non-infiltrated 7YSZ layer (white) is honeycombed with infiltrated 7YSZ areas (brown). Fig. 16 b) shows a confocal microscope image of the same sample after 660 s of erosion. The comparison of those images indicates that the erosion depth of infiltrated 7YSZ areas is higher compared to non-infiltrated 7YSZ. The partial infiltration of CMAS, which had led to a lowest erosion rate (ES(i)) of all tested alumina layers in stage (i), leads to a highest erosion rate of all samples (ES(iii)) in stage (iii) and an early failure of the coating system (Fig. 3 a). 3.6.5. Erosion model of SPS Al2O3/EB-PVD 7YSZ system It was shown that the CMAS infiltration depth and thickness of the reaction layer has a strong influence on the erosion behavior of sacrifi­ cial top coats. In order to test the erosion behavior with respect to the reaction layer thickness, a variety of CMAS compositions with different melting points and viscosities have been used in this study. This approach has the further advantage that the different reaction behaviors of the individual CMAS compositions with the sacrificial layer can also be evaluated. A similar study of such chemical effects was studied in literature [36,60]. The schematic erosion model of the coating system depending on the CMAS infiltration depth/reaction layer formation is presented in Fig. 17.

3.6.4. Scenario (4) - Uneven infiltration of the SPS alumina/EB-PVD 7YSZ system (CMAS 2) In scenario (4) parts of the coating system were completely infil­ trated and in other parts, a partial infiltration around 50 μm deep into the alumina layer was found (Fig. 13 and Table 4). This inhomogeneous infiltration occurs due a combination of CMAS properties and the microstructural effects of the coating. On one hand, C2 infiltrates the coating faster than C1 due to its lower viscosity, not offering enough time for the formation of infiltration hindering reaction products. On the other hand, alumina layer exhibits a high and discontinuous pore network. This unevenly porous microstructure causes an inhomoge­ neous infiltration, where highly porous areas were infiltrated stronger. Due to the reduced viscosity the effect may be more pronounced than with a comparably higher-viscous C1 [60]. The reaction layer thickness thus varied between 50 μm and 90 μm and influences the erosion pro­ cess, which is illustrated in Fig. 17 scenario (4). The first stage of the erosive process was characterized by surface spallation similar to the case for the ‘C1’ and ‘IVA’ samples, where the reaction layer was gradually removed. Although, some non-infiltrated alumina areas lie below the reaction layer, no compaction damage was observed. Since these areas are comparatively small (a few 100 μm in lateral dimension at a maximum), a supporting effect during the particle impact is provided by the incompressible, infiltrated areas surrounding the non-infiltrated areas (Fig. 13). Furthermore, cracks that were initiated in the infiltrated area of the alumina layer could not continue to grow in to the non-infiltrated area due to the high porosity that drastically lowers the stresses at the crack tip. This leads to a lim­ itation of crack growth. As a consequence of these two effects the lowest ES (ES(i) C2 ¼ 1.25 mm3/g) in stage (i) was achieved in this case

4. Conclusions SPS Al2O3 deposited on top of EB-PVD 7YSZ TBCs were infiltrated with CMAS/VA and the erosion properties were studied Various erosion mechanisms were observed depending on the SPS alumina/EB-PVD 7YSZ microstructure as well as on the different CMAS infiltration depths. The erosion process could be divided in three distinct stages that are characterized by different erosion mechanisms. This includes stage (i) surface erosion and initial spallation of SPS alumina layer, stage (ii) mixed erosion mode of SPS alumina spallation and EB-PVD 7YSZ erosion, and stage (iii) dominated by EB-PVD 7YSZ erosion followed by delamination and failure of the TBC. An erosion model was derived which describes the erosion behavior and the underlying mechanisms according to the depth of CMAS-infiltration in the different stages. Those include: (1) Erosion of non-infiltrated coating systems (e.g. ‘as-coated’) � Compaction damage and initial spallation of porous alumina layer within first 15 s � Individual column cracking of 7YSZ layer (2) Erosion of coating systems with a continuous reaction layer on top after CMAS infiltration and a remaining porous alumina layer underneath (e.g. ‘C1’) 14

L. Steinberg et al.

Wear 438–439 (2019) 203064

� Successive removal of residual CMAS and reaction layer by surface spallation � At a certain remaining layer thickness, vertical cracking of the reaction layer followed by compaction damage underneath the layer � Formation of horizontal delamination cracks in the porous alumina coating � Further erosion process comparable to non-infiltrated coating system (3) Erosion of 100% CMAS infiltrated coating system (e.g. ‘IVA’) � Successive removal of alumina and 7YSZ layer by surface spallation � Mechanical interlocking and chemical bonding improves the adhesion between alumina and 7YSZ layer � Highest erosion resistance of all tested systems (4) Erosion after non-uniform CMAS infiltration (e.g. ‘C2’) � Only surface spallation and no compaction damage even in non-infiltrated alumina areas � Local alteration of porous non-infiltrated alumina and fully infiltrated coatings systems initially retards erosion damage � Crack growth in the non-infiltrated 7YSZ areas is limited to the individual columns � Horizontal cracks in the infiltrated areas reach the vertical column gaps in the non-infiltrated areas � Increased erosion rate of non-uniform infiltrated 7YSZ areas leading to early failure

[8] R.G. Wellman, M.J. Deakin, J.R. Nicholls, The effect of TBC morphology on the erosion rate of EB PVD TBCs, Wear 258 (2005) 349–356, https://doi.org/10.1016/ j.wear.2004.04.011. [9] R.G. Wellman, M.J. Deakin, J.R. Nicholls, The effect of TBC morphology and aging on the erosion rate of EB-PVD TBCs, Tribol. Int. 38 (2005) 798–804, https://doi. org/10.1016/j.triboint.2005.02.008. [10] S. Rezanka, D.E. Mack, G. Mauer, D. Sebold, O. Guillon, R. Vaßen, Investigation of the resistance of open-column-structured PS-PVD TBCs to erosive and hightemperature corrosive attack, Surf. Coat. Technol. 324 (2017) 222–235, https:// doi.org/10.1016/j.surfcoat.2017.05.003. [11] M.P. Schmitt, D.E. Wolfe, D. Zhu, AmarendraK. Rai, R. Bhattacharya, Multilayered thermal barrier coating architectures for high temperature applications, in: D. Zhu, H.-T. Lin, Y. Zhou, T. Hwang, M. Halbig, S. Mathur (Eds.), Ceramic Engineering and Science Proceedings, John Wiley & Sons, Inc., Hoboken, NJ, USA, 2012, pp. 1–18, https://doi.org/10.1002/9781118217474.ch1. (Accessed 20 June 2016). [12] R.J.L. Steenbakker, R.G. Wellman, J.R. Nicholls, Erosion of gadolinia doped EBPVD TBCs, Surf. Coat. Technol. 201 (2006) 2140–2146, https://doi.org/10.1016/j. surfcoat.2006.03.022. [13] F. Cernuschi, L. Lorenzoni, S. Capelli, C. Guardamagna, M. Karger, R. Vaßen, K. von Niessen, N. Markocsan, J. Menuey, C. Giolli, Solid particle erosion of thermal spray and physical vapour deposition thermal barrier coatings, Wear 271 (2011) 2909–2918, https://doi.org/10.1016/j.wear.2011.06.013. [14] F. Cernuschi, C. Guardamagna, S. Capelli, L. Lorenzoni, D.E. Mack, A. Moscatelli, Solid particle erosion of standard and advanced thermal barrier coatings, Wear 348–349 (2016) 43–51, https://doi.org/10.1016/j.wear.2015.10.021. [15] M.P. Schmitt, A.K. Rai, D. Zhu, M.R. Dorfman, D.E. Wolfe, Thermal conductivity and erosion durability of composite two-phase air plasma sprayed thermal barrier coatings, Surf. Coat. Technol. 279 (2015) 44–52, https://doi.org/10.1016/j. surfcoat.2015.08.010. [16] M.P. Schmitt, B.J. Harder, D.E. Wolfe, Process-structure-property relations for the erosion durability of plasma spray-physical vapor deposition (PS-PVD) thermal barrier coatings, Surf. Coat. Technol. 297 (2016) 11–18, https://doi.org/10.1016/ j.surfcoat.2016.04.029. [17] D. Shin, A. Hamed, Influence of microstructure on erosion resistance of plasma sprayed 7YSZ thermal barrier coating under gas turbine operating conditions, Wear 396–397 (2018) 34–47, https://doi.org/10.1016/j.wear.2017.11.005. [18] N.A. Fleck, Th Zisis, The erosion of EB-PVD thermal barrier coatings: the competition between mechanisms, Wear 268 (2010) 1214–1224, https://doi.org/ 10.1016/j.wear.2009.12.020. [19] X. Chen, M.Y. He, I. Spitsberg, N.A. Fleck, J.W. Hutchinson, A.G. Evans, Mechanisms governing the high temperature erosion of thermal barrier coatings, Wear 256 (2004) 735–746, https://doi.org/10.1016/S0043-1648(03)00446-0. [20] S. Mahade, N. Curry, S. Bj€ orklund, N. Markocsan, P. Nyl� en, R. Vaßen, Erosion performance of gadolinium zirconate-based thermal barrier coatings processed by suspension plasma spray, J. Therm. Spray Technol. 26 (2017) 108–115, https:// doi.org/10.1007/s11666-016-0479-4. [21] L. Yang, H.L. Li, Y.C. Zhou, W. Zhu, Y.G. Wei, J.P. Zhang, Erosion failure mechanism of EB-PVD thermal barrier coatings with real morphology, Wear 392–393 (2017) 99–108, https://doi.org/10.1016/j.wear.2017.09.018. [22] S. Kr€ amer, J. Yang, C.G. Levi, Infiltration-inhibiting reaction of gadolinium zirconate thermal barrier coatings with CMAS melts, J. Am. Ceram. Soc. 91 (2008) 576–583, https://doi.org/10.1111/j.1551-2916.2007.02175.x. [23] S. Kr€ amer, J. Yang, C.G. Levi, C.A. Johnson, Thermochemical interaction of thermal barrier coatings with molten CaO–MgO–Al2O3–SiO2 (CMAS) deposits, J. Am. Ceram. Soc. 89 (2006) 3167–3175, https://doi.org/10.1111/j.15512916.2006.01209.x. [24] S. Kr€ amer, S. Faulhaber, M. Chambers, D.R. Clarke, C.G. Levi, J.W. Hutchinson, A. G. Evans, Mechanisms of cracking and delamination within thick thermal barrier systems in aero-engines subject to calcium-magnesium-alumino-silicate (CMAS) penetration, Mater. Sci. Eng. A 490 (2008) 26–35, https://doi.org/10.1016/j. msea.2008.01.006. [25] R. Naraparaju, J.T. Gomez Chavez, U. Schulz, C.V. Ramana, Interaction and infiltration behavior of Eyjafjallaj€ okull, Sakurajima volcanic ashes and a synthetic CMAS containing FeO with/in EB-PVD ZrO2-65 wt% Y2O3 coating at high temperature, Acta Mater. 136 (2017) 164–180, https://doi.org/10.1016/j. actamat.2017.06.055. [26] G. Pujol, F. Ansart, J.-P. Bonino, A. Mali�e, S. Hamadi, Step-by-step investigation of degradation mechanisms induced by CMAS attack on YSZ materials for TBC applications, Surf. Coat. Technol. 237 (2013) 71–78, https://doi.org/10.1016/j. surfcoat.2013.08.055. [27] H. Peng, L. Wang, L. Guo, W. Miao, H. Guo, S. Gong, Degradation of EB-PVD thermal barrier coatings caused by CMAS deposits, Prog. Nat. Sci.: Mater. Int. 22 (2012) 461–467, https://doi.org/10.1016/j.pnsc.2012.06.007. [28] M.H. Vidal-Setif, N. Chellah, C. Rio, C. Sanchez, O. Lavigne, Calcium–magnesium–alumino-silicate (CMAS) degradation of EB-PVD thermal barrier coatings: characterization of CMAS damage on ex-service high pressure blade TBCs, Surf. Coat. Technol. 208 (2012) 39–45, https://doi.org/10.1016/j. surfcoat.2012.07.074. [29] C.G. Levi, J.W. Hutchinson, M.-H. Vidal-S�etif, C.A. Johnson, Environmental degradation of thermal-barrier coatings by molten deposits, MRS Bull. 37 (2012) 932–941, https://doi.org/10.1557/mrs.2012.230. [30] E. Bohorquez, B. Sarley, J. Hernandez, R. Hoover, L. Tetard, R. Naraparaju, U. Schulz, S. Raghavan, Investigation of the effects of CMAS-infiltration in EB-PVD 7% yttria-stabilized zirconia via Raman spectroscopy, in: 2018 AIAA/ASCE/AHS/ASC

Overall, infiltrated alumina layers have shown increased erosion resistance compared to the as coated samples. . Nevertheless, a higher degree of CMAS infiltration is assumed to reduce the strain tolerance of columnar EB-PVD 7YSZ which influences the lifetime of the TBC. Declarations of interest None. Acknowledgments The authors thank the Deutsche Forschungsgemeinschaft (DFG) for funding this project (project no. LE 1373/34-1, SCHU 1372/5-1). The technical assistance of P. Lutze at the TU Dresden as well as helpful technical discussions with M. Thorhauer and S. Heinze are gratefully acknowledged. The authors also express their gratitude to J. Brien, A. Handwerk and D. Peters from DLR for producing of the EB-PVD 7YSZ layers, as well as for technical support and advise. References [1] N.P. Padture, M. Gell, E.H. Jordan, Thermal barrier coatings for gas-turbine engine applications, Science 296 (2002) 280–284. [2] A.G. Evans, D.R. Clarke, C.G. Levi, The influence of oxides on the performance of advanced gas turbines, J. Eur. Ceram. Soc. 28 (2008) 1405–1419, https://doi.org/ 10.1016/j.jeurceramsoc.2007.12.023. [3] R. Naraparaju, P. Mechnich, U. Schulz, G.C. Mondragon Rodriguez, The accelerating effect of CaSO4 within CMAS (CaO–MgO–Al2O3–SiO2) and its effect on the infiltration behavior in EB-PVD 7YSZ, J. Am. Ceram. Soc. 99 (2016) 1398–1403, https://doi.org/10.1111/jace.14077. [4] R.M. Leckie, S. Kr€ amer, M. Rühle, C.G. Levi, Thermochemical compatibility between alumina and ZrO2–GdO3/2 thermal barrier coatings, Acta Mater. 53 (2005) 3281–3292, https://doi.org/10.1016/j.actamat.2005.03.035. [5] R.G. Wellman, J.R. Nicholls, A review of the erosion of thermal barrier coatings, J. Phys. D Appl. Phys. 40 (2007) R293–R305, https://doi.org/10.1088/00223727/40/16/R01. [6] R. Wellman, J.R. Nicholls, A mechanism for the erosion of EB PVD TBCs, in: Materials Science Forum, Trans Tech Publ, 2001, pp. 531–538. http://www.scien tific.net/MSF.369-372.531.pdf. (Accessed 14 June 2016). [7] R.G. Wellman, J.R. Nicholls, K. Murphy, Effect of microstructure and temperature on the erosion rates and mechanisms of modified EB PVD TBCs, Wear 267 (2009) 1927–1934, https://doi.org/10.1016/j.wear.2009.04.002.

15

L. Steinberg et al.

[31] [32] [33]

[34] [35] [36]

[37]

[38] [39] [40] [41] [42] [43] [44]

[45]

[46]

[47] [48]

[49] [50]

[51] [52]

[53]

Wear 438–439 (2019) 203064

Structures, Structural Dynamics, and Materials Conference, American Institute of Aeronautics and Astronautics (n.d.) doi:10.2514/6.2018-0096.. R.G. Wellman, J.R. Nicholls, Erosion, corrosion and erosion–corrosion of EB PVD thermal barrier coatings, Tribol. Int. 41 (2008) 657–662, https://doi.org/10.1016/ j.triboint.2007.10.004. R. Wellman, G. Whitman, J.R. Nicholls, CMAS corrosion of EB PVD TBCs: identifying the minimum level to initiate damage, Int. J. Refract. Metals Hard Mater. 28 (2010) 124–132, https://doi.org/10.1016/j.ijrmhm.2009.07.005. L. Steinberg, R. Naraparaju, M. Heckert, C. Mikulla, U. Schulz, C. Leyens, Erosion behavior of EB-PVD 7YSZ coatings under corrosion/erosion regime: effect of TBC microstructure and the CMAS chemistry, J. Eur. Ceram. Soc. 38 (2018) 5101–5112, https://doi.org/10.1016/j.jeurceramsoc.2018.06.039. W.C. Hasz, M.P. Borom, C.A. Johnson, Protected thermal barrier coating composite with multiple coatings, US5914189A, https://patents.google.com/patent/US59 14189A/en, 1999. (Accessed 17 July 2018). A.K. Rai, R.S. Bhattacharya, D.E. Wolfe, T.J. Eden, CMAS-resistant thermal barrier coatings (TBC), Int. J. Appl. Ceram. Technol. 7 (2010) 662–674, https://doi.org/ 10.1111/j.1744-7402.2009.02373.x. R. Naraparaju, R.P. Pubbysetty, P. Mechnich, U. Schulz, EB-PVD alumina (Al2O3) as a top coat on 7YSZ TBCs against CMAS/VA infiltration: deposition and reaction mechanisms, J. Eur. Ceram. Soc. 38 (2018) 3333–3346, https://doi.org/10.1016/j. jeurceramsoc.2018.03.027. Mechnich Peter, Wolfgang Braue, D.J. Green, Volcanic ash-induced decomposition of EB-PVD Gd2Zr2O7 thermal barrier coatings to Gd-oxyapatite, zircon, and Gd, Fezirconolite, J. Am. Ceram. Soc. 96 (2013) 1958–1965, https://doi.org/10.1111/ jace.12251. C. Jiang, Low Thermal Conductivity YSZ-Based Thermal Barrier Coatings with Enhanced CMAS Resistance, University of Connecticut, 2015. A. Aygun, Novel Thermal Barrier Coatings (TBCs) that Are Resistant to High Temperature Attack by Cao-MgO-Al2O3-SiO2 (CMAS) Glassy Deposits, Ohio State University, 2008. K.-I. Lee, Volcanic Ash Degradation on Thermal Barrier Coatings and Preliminary Fabrication of Protective Coatings, University of Manchester, 2014. A. Aygun, A.L. Vasiliev, N.P. Padture, X. Ma, Novel thermal barrier coatings that are resistant to high-temperature attack by glassy deposits, Acta Mater. 55 (2007) 6734–6745, https://doi.org/10.1016/j.actamat.2007.08.028. J.M. Drexler, A.D. Gledhill, K. Shinoda, A.L. Vasiliev, K.M. Reddy, S. Sampath, N. P. Padture, Jet engine coatings for resisting volcanic ash damage, Adv. Mater. 23 (2011) 2419–2424, https://doi.org/10.1002/adma.201004783. S. Kr€ amer, J. Yang, C.G. Levi, Infiltration-inhibiting reaction of gadolinium zirconate thermal barrier coatings with CMAS melts, J. Am. Ceram. Soc. 91 (2008) 576–583, https://doi.org/10.1111/j.1551-2916.2007.02175.x. A.R. Krause, H.F. Garces, B.S. Senturk, N.P. Padture, 2ZrO2⋅Y2O3 thermal barrier coatings resistant to degradation by molten CMAS: Part II, interactions with sand and fly ash, J. Am. Ceram. Soc. 97 (2014) 3950–3957, https://doi.org/10.1111/ jace.13209. U. Schulz, W. Braue, Degradation of La2Zr2O7 and other novel EB-PVD thermal barrier coatings by CMAS (CaO–MgO–Al2O3–SiO2) and volcanic ash deposits, Surf. Coat. Technol. 235 (2013) 165–173, https://doi.org/10.1016/j. surfcoat.2013.07.029. S. Mahade, N. Curry, S. Bj€ orklund, N. Markocsan, P. Nyl� en, R. Vaßen, Erosion performance of gadolinium zirconate-based thermal barrier coatings processed by suspension plasma spray, J. Therm. Spray Technol. 26 (2017) 108–115, https:// doi.org/10.1007/s11666-016-0479-4. M. Munro, Evaluated material properties for a sintered alpha-alumina, J. Am. Ceram. Soc. 80 (1997) 1919–1928, https://doi.org/10.1111/j.1151-2916.1997. tb03074.x. M.E. Murphy, M.T. Laugier, B.D. Beake, D. Sutton, S.B. Newcomb, The effects of C ion implantation on the near surface microstructure and properties of alpha alumina, J. Mater. Sci. 37 (2002) 2053–2059, https://doi.org/10.1023/A: 1015267702791. R.G. Wellman, A. Dyer, J.R. Nicholls, Nano and Micro indentation studies of bulk zirconia and EB PVD TBCs, Surf. Coat. Technol. 176 (2004) 253–260, https://doi. org/10.1016/S0257-8972(03)00737-0. C. Mercer, J.R. Williams, D.R. Clarke, A.G. Evans, On a ferroelastic mechanism governing the toughness of metastable tetragonal-prime (t’) yttria-stabilized zirconia, Proc. R. Soc. A Math. Phys. Eng. Sci. 463 (2007) 1393–1408, https://doi. org/10.1098/rspa.2007.1829. M. Fujikane, D. Setoyama, S. Nagao, R. Nowak, S. Yamanaka, Nanoindentation examination of yttria-stabilized zirconia (YSZ) crystal, J. Alloy. Comp. 431 (2007) 250–255, https://doi.org/10.1016/j.jallcom.2006.05.058. P. Mohan, B. Yao, T. Patterson, Y.H. Sohn, Electrophoretically deposited alumina as protective overlay for thermal barrier coatings against CMAS degradation, Surf. Coat. Technol. 204 (2009) 797–801, https://doi.org/10.1016/j. surfcoat.2009.09.055. X. Zhang, K. Zhou, W. Xu, B. Chen, J. Song, M. Liu, In situ synthesis of α-alumina layer on thermal barrier coating for protection against CMAS (CaO–MgO–Al2O3–SiO2) corrosion, Surf. Coat. Technol. 261 (2015) 54–59, https://doi.org/10.1016/j.surfcoat.2014.11.064.

[54] X. Zhang, K. Zhou, W. Xu, B. Chen, J. Song, M. Liu, In situ synthesis of α-alumina layer on thermal barrier coating for protection against CMAS (CaO–MgO–Al2O3–SiO2) corrosion, Surf. Coat. Technol. 261 (2015) 54–59, https://doi.org/10.1016/j.surfcoat.2014.11.064. [55] L.-M. Berger, F.-L. Toma, A. Potthoff, Thermal spraying with suspensions-an economic spray process, Therm. Spray Bull. 6 (2013) 98–101. [56] F.-L. Toma, A. Potthoff, L.-M. Berger, C. Leyens, Demands, potentials, and economic aspects of thermal spraying with suspensions: a critical review, J. Therm. Spray Technol. 24 (2015) 1143–1152, https://doi.org/10.1007/s11666-015-02747. [57] C.S. Ramachandran, V. Balasubramanian, P.V. Ananthapadmanabhan, Erosion of atmospheric plasma sprayed rare earth oxide coatings under air suspended corundum particles, Ceram. Int. 39 (2013) 649–672, https://doi.org/10.1016/j. ceramint.2012.06.077. [58] A. Flores Renteria, A small-angle scattering analysis of the influence of manufacture and thermal induced morphological changes on the thermal conductivity of EB-PVD PYSZ thermal barrier coatings. http://darwin.bth.rwth -aachen.de/opus3/volltexte/2007/1790/pdf/Flores_Renteria_Arturo.pdf, 2007. (Accessed 14 July 2016). [59] R. Naraparaju, M. Hüttermann, U. Schulz, P. Mechnich, Tailoring the EB-PVD columnar microstructure to mitigate the infiltration of CMAS in 7YSZ thermal barrier coatings, J. Eur. Ceram. Soc. 37 (2017) 261–270, https://doi.org/10.1016/ j.jeurceramsoc.2016.07.027. [60] C. Mikulla, R. Naraparaju, U. Schulz, F.-L. Toma, M. Barbosa, C. Leyens, Investigation of CMAS resistance of sacrificial suspension sprayed alumina topcoats on EB-PVD 7YSZ layers, J. Therm. Spray Technol. (2019). [61] R. Naraparaju, U. Schulz, P. Mechnich, P. D€ obber, F. Seidel, Degradation study of 7wt.% yttria stabilised zirconia (7YSZ) thermal barrier coatings on aero-engine combustion chamber parts due to infiltration by different CaO–MgO–Al2O3–SiO2 variants, Surf. Coat. Technol. 260 (2014) 73–81, https://doi.org/10.1016/j. surfcoat.2014.08.079. [62] R. Naraparaju, J.J. Gomez Chavez, P. Niemeyer, K.-U. Hess, W. Song, D. B. Dingwell, S. Lokachari, C.V. Ramana, U. Schulz, Estimation of CMAS infiltration depth in EB-PVD TBCs: a new constraint model supported with experimental approach, J. Eur. Ceram. Soc. 39 (2019) 2936–2945, https://doi.org/10.1016/j. jeurceramsoc.2019.02.040. [63] D.N. Boccaccini, A.R. Boccaccini, Dependence of Ultrasonic Velocity on Porosity and Pore Shape in Sintered Materials (n.d.)(6).. [64] A. Nagarajan, Ultrasonic study of elasticity-porosity relationship in polycrystalline alumina, J. Appl. Phys. 42 (1971) 3693–3696, https://doi.org/10.1063/ 1.1659671. [65] M. Asmani, C. Kermel, A. Leriche, M. Ourak, Influence of porosity on Young’s modulus and Poisson’s ratio in alumina ceramics, J. Eur. Ceram. Soc. 21 (2001) 1081–1086, https://doi.org/10.1016/S0955-2219(00)00314-9. [66] V.L. Wiesner, N.P. Bansal, Mechanical and thermal properties of calcium–magnesium aluminosilicate (CMAS) glass, J. Eur. Ceram. Soc. 35 (2015) 2907–2914, https://doi.org/10.1016/j.jeurceramsoc.2015.03.032. [67] B.-K. Jang, H. Matsubara, Hardness and Young’s modulus of nanoporous EB-PVD YSZ coatings by nanoindentation, J. Alloy. Comp. 402 (2005) 237–241, https:// doi.org/10.1016/j.jallcom.2005.02.064. [68] A. Rico, J. G� omez-García, C.J. Múnez, P. Poza, V. Utrilla, Mechanical properties of thermal barrier coatings after isothermal oxidation, Surf. Coat. Technol. 203 (2009) 2307–2314, https://doi.org/10.1016/j.surfcoat.2009.02.035. [69] E.P. Busso, Z.Q. Qian, M.P. Taylor, H.E. Evans, The influence of bondcoat and topcoat mechanical properties on stress development in thermal barrier coating systems, Acta Mater. 57 (2009) 2349–2361, https://doi.org/10.1016/j. actamat.2009.01.017. [70] R.W. Jackson, E.M. Zaleski, D.L. Poerschke, B.T. Hazel, M.R. Begley, C.G. Levi, Interaction of molten silicates with thermal barrier coatings under temperature gradients, Acta Mater. 89 (2015) 396–407, https://doi.org/10.1016/j. actamat.2015.01.038. [71] L.-M. Berger, D. Schneider, M. Barbosa, R. Puschmann, Laser acoustic surface waves for the non-destructive characterisation of thermally sprayed coatings, Therm. Spay Bull. 64 (2012) 56–64. [72] C.K. Lin, C.C. Berndt, Measurement and analysis of adhesion strength for thermally sprayed coatings, JTST 3 (1994) 75–104, https://doi.org/10.1007/BF02649003. [73] J. Comyn, Surface treatment and analysis for adhesive bonding, Int. J. Adhesion Adhes. 10 (1990) 161–165, https://doi.org/10.1016/0143-7496(90)90099-J. [74] D. Schneider, R. Puschmann, M. Barbosa, L.-M. Berger, Evaluating porosity in thermal-sparyed Al2O3 coatings by Laser-Acoustics, in: Proceedings of the Twenty Fourth International Conference on Surface Modification Technologies, Dresden, 2010, pp. 435–448. [75] C.-J. Li, A. Ohmori, Relationships between the microstructure and properties of thermally sprayed deposits, J. Therm. Spray Technol. 11 (2002) 365–374, https:// doi.org/10.1361/105996302770348754. [76] H.E. Eaton, R.C. Novak, Particulate erosion of plasma-sprayed porous ceramic, Surf. Coat. Technol. 30 (1987) 41–50, https://doi.org/10.1016/0257-8972(87) 90006-5.

16