Tensile Adhesion Strength of Atmospheric Plasma Sprayed MgAl₂O₄, Al₂O₃ Coatings

Abstract

This study analyses the distribution of stress during the testing of glued cylindrical specimens with thermally sprayed MgAl₂O₄, Al₂O₃ oxide coatings in order to evaluate the tensile adhesion strength. The set of studies that make up this work were conducted in order to evaluate the influence of the geometric parameters of cylindrical test specimens, 25 mm in diameter by 16–38.1 mm in height, on the measured tensile adhesion strength of the specimens. The stress and strain states inside the coating and at the coating-substrate interface were determined using the finite element modelling method. The debonding mechanisms, failure mode and influence of the coating microstructure on bond strength are also discussed. The finite element stress analysis shows a significant level of non-uniform stress distribution in the test specimens. The analysis of the results of the modelling stresses and strains using the finite element method for six types of cylindrical specimens, as well as the values obtained for the adhesion testing of MgAl₂O₄, Al₂O₃ coatings, show a need to increase the height of the standard cylindrical specimen (according to ASTM C633-13 (2021), GOST 9.304-87). The height should be increased by no less than 1.5–2.0 times to reduce the level of a non-uniform stress distribution in the separation area.

1. Introduction

Thermally sprayed coatings of MgAl₂O₄, Al₂O₃ are commonly used as a protective layer for parts that require strong insulating properties ($\mathsf{\rho }$ = 10⁻¹⁴−10⁻¹⁵ Ohm cm, $\mathrm{T}$ = 20 °C) under the conditions of high temperature, vacuums and radiation. Magnesium aluminate spinel compositions using aluminium oxide MgAl₂O₄ + 30%Al₂O₃, MgAl₂O₄ + 50%Al₂O₃, with a coating thickness of between 0.35 and 0.50 mm, offers satisfactory electrical insulation via high intensity neutron and gamma fluxes in vacuum, at temperatures ranging between 20 and 700 °C. The MgAl₂O₄, Al₂O₃ coatings are widely used as an electrical insulation layer for high-voltage electrical devices, channels and parts of magnetohydrodynamic (MHD) generators, sensors, the pads of reactors and International Thermonuclear Experimental Reactor (ITER) blanket modules. These types of coatings are predominantly deposited using atmospheric plasma spraying (APS) [1,2,3]. The performance of oxide ceramics that are applied to reactor facilities will be determined not only by a significant difference in the thermal expansion coefficients of the substrate and the coating material, but also by phase transformations at high temperatures and through exposure to radiation [4,5]. Compared to aluminium oxide, the cubic crystal lattice of spinel provides greater stability at high fluxes of gamma-neutron irradiation at high temperatures.

The problem of estimating the degree of a non-uniform stress-strain state at the interfacial debonding region of coating and over the thickness has practical applications that will allow us to better interpret the results and optimise the geometry parameters of standard cylindrical specimens, according to the international standards EN ISO 14916:2017, ASTM C633-13 (2021) and GOST 9.304-87, JIS H 8666:1994 [6,7,8,9]. In practical terms, the standards and methods used to estimate the bond strength of thermal spray coatings are reduced to two main approaches for assessing cylindrical and conical specimens under tensile loads (uniaxial stress state). Most methods used to estimate the adhesion strength of a coating to a substrate can also be divided into two main groups; namely, qualitative testing methods and quantitative testing methods [10,11,12,13,14,15].

The tensile adhesion strength of thermal spray coatings will depend on different mechanical characteristics, including crack resistance (fracture toughness), Young’s module, microhardness/hardness, compressive and tensile strength, types of residual stresses (quenching, peening, compressive [16,17]), differences in the coefficients of thermal expansion between the coating and the substrate as well as other features [10,11,18,19,20]. Parameters such as the roughness of the substrate after grit blasting, coating thickness, the composition of the coating, the tendency of the substrate/coating system to form chemical bonds and spraying methods (APS, HVOF, suspension-HVOF, DGS, LPCS [21,22,23], etc.) are critically important in terms of the adhesion strength. As such, there is no straightforward method for determining the adhesion strength of the part if it is operating under a complex stress state [24,25,26,27].

Meanwhile, test data obtained using different methods are often difficult to compare and are therefore rarely used for analytical description. There are some basic disadvantages that make it unfeasible to apply standard tensile strength test methods to a cylindrical specimen for high strength thermally sprayed ceramic coatings (bond strength more than 100 MPa). These are, firstly, the insufficient adhesive strength of glue at room and high temperatures, and secondly, the penetration of the adhesive to the substrate when the coatings are thin.

In cases where it impossible to perform the glue test, it is possible to apply an alternative pin method (Figure 1). However, this also has some drawbacks due to the lack of precise interpretation of the data due to the small diameter of the conical pin separation area (${\mathrm{d}}_{\mathrm{pin}}$ = 2–2.5 mm) and the incomplete separation of the brittle and highly porous coating.

The application of a tensile adhesion test to a conical pin specimen may eliminate the influence of friction forces between the pin and the mating conical pair matrix (Figure 1). The end surfaces of the assembled pair must be carefully grounded before grit blasting and thermal spraying, and the absence of glue then makes it possible to investigate the adhesive strength of the thermally sprayed coatings at high temperatures.

Comparative studies [10,14,24,28] show a significant impact on stress distribution at the bonding interface of the main geometric parameters based on the ratio of height (length)—${\mathrm{h}}_{\mathrm{spec}}$ to depth of the threaded closed hole—${\mathrm{h}}_{\mathrm{thr}.\mathrm{holl}}$ and the nominal diameter—${\mathrm{d}}_{\mathrm{spec}}$ of a cylindrical specimen. The studies focus on investigating the influence of specimen height on tensile adhesion strength for both standard and elongated glued specimens with thermal spray coatings assumed by ASTM C633-13 (2021), GOST 9.304-87. The effect of various specimen heights on the tensile adhesion test results and the interface stress distribution at the debonding area will be presented and discussed in the following sections.

2. Experimental Procedures

2.1. Plasma Spraying Conditions

Magnesium aluminum spinel (magnesium aluminate) and aluminium oxide powders (MgAl₂O₄ GTV 40.70.1, GTV GmbH, Luckenbach, Westerwald, Germany, particle size 20−45 µm; Al₂O₃ Amperit 740.001, H.C. Starck, particle size 22−45 µm) were deposited onto austenitic stainless steel 316L(N)-IG substrates using atmospheric plasma spraying, through the YPY-8M (JSC “Electromekhanika”, Rzhev, Tver region, Russia [29]) technique, with a mounted low-capacity plasma torch. The spraying parameters for two types of oxide coatings and bond coatings are listed in

Table 1. Plasma spraying operating parameters.

Parameters	MgAl2O4, Coating	Al2O3, Coating	NiAl, Bond Coat
Current (A) × voltage [V] = power [kW]	330 × 50 = 16.50	320 × 50 = 16.00	330 × 48 = 15.84
Spray distance, [mm]	105		90
Feed powder, [g/min]	15–18		30
Velocity of travel plasma torch, [s/min]	2800		1600
Carrier gas type and flow rates, [L/min]	Ar: 25–30; N2: 2.0–2.4		Ar: 20–25; N2: 1.5–2.0
Transporting gas type and flow rate, [L/min]	Ar: 2.5−3.0
Number of passes	9–10		1–2

. Argon-nitrogen was used as plasma-forming mixture, and the constant flow rates are also listed in Table 1. The standard plasma spray conditions of the oxide coating, such as MgAl₂O₄, Al₂O_3, are described in various articles [22,24,30,31]. Each set of specimens was deposited by specially designed equipment to set specimens at the same spray distance (Table 1). The spraying of ceramic electrically insulating MgAl₂O₄, Al₂O₃ oxide coatings on an austenitic steel substrate of a 316L(N)-IG alloy, standard EN 10088-1:2005 was performed using atmospheric plasma spraying equipment by feeding powdered material into the column of the compressed arc discharge (the anode spot of plasma torch between cathode and anode).

For testing purposes, the coatings were deposited onto six cylindrical specimen types (${\mathrm{d}}_{\mathrm{spec}}$ = 25/25.4 mm in diameter; 3 standard and 3 elongated—heights ${\mathrm{h}}_{\mathrm{spec}}$ = 28–38.1 mm) by thickness ${\mathrm{h}}_{\mathrm{c}}$ from 0.4 to 0.5 mm with a bond coating of NiAl (powder: Metco 40NS, Oerlicon Metco, Wohlen, Switzerland, particle size 45−90 µm) (Figure 2). In order to isolate the uncoated surfaces of a cylindrical specimen from a stream of spray particles, a protective mask was applied. The first and third types of cylindrical specimen were manufactured in accordance with the standards ASTM C633-13 (2021), GOST 9.304-87, (Figure 2).

The geometry of the second type of specimen (Figure 2) was altered to improve the ability of the process to remove the adhesive (glue) surge after curing and to save the possibility of further centring the specimen in the precision bushings (Figure 3). The design of the elongated specimen was changed in accordance with the ratio of the specimen’s height ${\mathrm{h}}_{\mathrm{spec}}$ to depth of the threaded closed hole ${\mathrm{h}}_{\mathrm{thr}.\mathrm{holl}}$ using the following equation:

$$\mathsf{\alpha } =\frac{{\mathrm{h}}_{\mathrm{thr}.\mathrm{holl}}}{{\mathrm{h}}_{\mathrm{spec}}}=0.4$$

(1)

The coefficient ratio of the specimen heights $\mathsf{\alpha }$ = 0.4 was chosen based on the results of similar studies [14].

Before the substrates were coat-sprayed, they were grit-blasted with 1 mm alumina grit at an air pressure of up to 4–5 kg/cm². The mean roughness of the substrate was approximately Rz 45 ± 5 μm.

2.2. Finite Element Analysis, Boundary Conditions

Considerable research has been performed to investigate the stress-strain distribution in oxide coating and substrate interface to develop six finite element models using the ANSYS R16.2 programme, (Ansys Inc., Canonsburg, PA, USA). The axisymmetric models included two identical cylindrical halves, bonded together with a ceramic coating.

In order to facilitate calculations, the adhesive layer (glue) was excluded from the models, and the primary objective of this study was to estimate the level of uniformity in the interface stress distribution. It was therefore possible to simplify the finite element models of the specimens.

The following boundary conditions and assumptions were made for the finite-element models:

• The contact between the substrate and coating was assumed as a rigid connection with a restriction on all degrees of freedom
• Axial symmetry was provided by a cyclic region in the form of a 1/4 model by imposing boundary conditions to the symmetry regions along the XOY and YOZ planes
• The material of the substrate and oxide coating possessed only elastic strain.

The finite element grid near the coating-substrate interface was more refined than in other areas of the model—approximately 0.1 mm (Figure 4). The results of a finite element analysis (FEA) of the models were compared with each other and with the results of the bond strength tests. The finite element models were loaded with a tensile load of $\mathrm{F}$ = 14,000 N to the threaded hole, as shown in Figure 4. The mechanical properties of the Al₂O₃ coating and substrate were taken from the material library of the ANSYS R16.2 programme.

Previous tests [31,32,33] revealed very similar mechanical and physical properties for the MgAl₂O₄, Al₂O₃ oxide coatings, which made it possible to perform FEA only for the Al₂O₃ coating.

2.3. Tensile Adhesion Strength Testing

In this investigation, the test method of estimation tensile adhesion strength was based on the ASTM C633-13 (2021), GOST 9.304-87 [6,9] standards. Six specimens (standard and elongated) were bonded to the same uncoated specimens (counterparts) using a single-part epoxy resin Permobond ES 550 (maximum tensile strength ${\mathsf{\sigma }}_{\mathrm{ten}}$ = 60 ± 3 GPa, maximum shear strength ${\mathsf{\tau }}_{\mathrm{sh}}$ = 41 ± 2 GPa by $\mathrm{T}$ = 20 °C). The alignment of the specimens was provided by centring bushings. The coatings produced were compared in terms of phase composition and microstructure, and the bond strength was also analysed at a temperature of $\mathrm{T}$ = 20 °C.

After thermal curing and cooling, the specimens were loaded with axial breaking force until debonding occurred at a traverse speed of 0.5 mm/min in the IR5143-200-11 tensile testing machine (JSC “Tochpribor”, Ivanovo, Russia [34]). The tensile adhesion strength of ${\mathsf{\sigma }}_{\mathrm{ten}}$ was calculated as the ratio between ultimate tensile force $\mathrm{F}$ divided by cross-sectional area $\mathrm{A}$ ($\mathrm{A}$ = 491 mm² for all testing results). Three identical tests were performed under the same conditions for each type of cylindrical specimen, using the MgAl₂O₄, Al₂O₃ plasma spray coating.

3. Results and Discussion

3.1. Results of Tensile Bond Strength of Standard and Elongated Specimens

Verification of the FEA data was performed by comparing them with the summary results of the tensile adhesion testing. The bond strengths of the APS ceramic coatings—MgAl₂O₄, Al₂O₃ for both standard and elongated specimens—are shown in

Table 2. Summary results of tensile bond strength measurements for APS—MgAl₂O₄, Al₂O₃ and finite element surface stresses on standard and elongated specimens.

Type of Specimen/Test Method	Type of Coating	Thickness $h_c$ [μm]	Roughness $Ra$ [μm]	Tensile Bond Strength $\overline{{\mathsf{\sigma }}_{\mathrm{ten}}}\pm \delta$ [MPa]	Relative Error of Tensile Bond Strength [%]	Finite Element Average Equivalent Stress $\overline{{\mathsf{\sigma }}_{\mathrm{eq}}}$ [MPa]	$\mathsf{\beta }$
Type 1, standard/GOST 9.304-87 [9]	Al2O3_APS	420 ± 35	38 ± 10	30.1 ± 1.6	5	60	6.7
	MgAl2O4_APS	406 ± 27	33 ± 6	28.8 ± 1.9	7	-	-
Type 4, elongated/GOST 9.304-87 [9]	Al2O3_APS	484 ± 21	29 ± 4	25.0 ± 2.6	10	45	1.4
	MgAl2O4_APS	446 ± 45	48 ± 5	22.0 ± 4.3	19	-	-
Type 2, modified/an in-house test method	Al2O3_APS	424 ± 38	33 ± 8	23.2 ± 3.2	14	45	3.1
	MgAl2O4_APS	484 ± 18	31 ± 4	26.4 ± 1.6	6	-	-
Type 5, elongated/an in-house test method	Al2O3_APS	426 ± 36	31 ± 6	19.4 ± 0.5	2	44	1.8
	MgAl2O4_APS	440 ± 29	30 ± 2	20.8 ± 2.0	10	-	-
Type 3, standard/ASTM C633-13 (2021) [6]	Al2O3_APS	460 ± 32	37 ± 8	27.8 ± 0.6	2	43	4.9
	MgAl2O4_APS	430 ± 30	38 ± 4	25.1 ± 3.2	13	-	-
Type 6, elongated/ASTM C633-13 (2021) [6]	Al2O3_APS	472 ± 12	33 ± 3	20.7 ± 1.7	8	44	1.7
	MgAl2O4_APS	450 ± 31	37 ± 3	18.4 ± 1.5	8	-	-

. Before testing the tensile strength of the coating thickness, the ${\mathrm{h}}_{\mathrm{c}}$ and roughness Ra were estimated (Table 2). In the present investigation, the adhesive failure mode occurred at the bond coating-substrate interface [30] for all of the tensile tests of the MgAl₂O₄, Al₂O₃ coatings, without the visible penetration of epoxy resin into the coating-substrate interface. The trend of decreased adhesion strength for elongated specimens correlated well with the FEA results.

The results of the tensile adhesion strength tests showed a low level of bond strength for applied APS technology for MgAl₂O₄, Al₂O₃ coatings [3,10,22,24]. This may be related to the spraying equipment used, the construction of the plasma torch, the residual stress distribution, the fatigue crack propagation, the pore size and distribution [35], or the rate of particle velocity and temperature.

The high level of stress discontinuity at the coating-substrate interface might help explain the increased value of the adhesion strength (Table 2). As the surface morphology images in Figure 5 show, the low level of bond strength conforms well to the results of the metallographic analysis of the scanning electron microscope (SEM) images of the cross sections and surface morphologies of the coatings (large, moulded agglomerates, a large number of pores and other defects).

3.2. Assessment of Stress Distribution during Tensile Testing

The equivalent stress distribution and total strain showed in the three standard and three elongated specimens are shown in Figure 6, Figure 7 and Figure 8. A calculation of the stress distribution at the interface of three standard specimens showed a high level of non-uniformity over the debonding area. For example, the maximum value of equivalent stress came at the outer edge of the ASTM C633-13 (2021) standard specimen (type 3, Figure 2)—${\mathsf{\sigma }}_{\mathrm{eq}}^{\max }$ = 71.5 MPa and the minimum value occurs at the centre—${\mathsf{\sigma }}_{\mathrm{eq}}^{\min }$ = 14.5 MPa (Figure 8). The international standards procedure assumes that the stress is uniform based on testing a cylindrical specimen. The most uneven distribution of the equivalent stress was found at the interface of the GOST 9.304-87 standard specimen (type 1, Figure 2)—the maximum ${\mathsf{\sigma }}_{\mathrm{eq}}^{\max }$ = 59.6 MPa occurs between the centre and the outer edge of the specimen, while the minimum ${\mathsf{\sigma }}_{\mathrm{eq}}^{\min }$ = 9 MPa occurs at the centre of the specimen (Figure 6). The maximum value of the equivalent stress can therefore be taken as the average equivalent of the interface for the standard specimen type 1—${\mathsf{\sigma }}_{\mathrm{eq}}^{\max }=\overline{{ \mathsf{\sigma }}_{\mathrm{eq}}}$.

The average equivalent stress $\overline{{\mathsf{\sigma }}_{\mathrm{eq}}}$ = 43 MPa (Figure 8) showed by FEA is 155% higher than the average tensile adhesion strength ${\mathsf{\sigma }}_{\mathrm{ten}}$ = 27.8 ± 0.6 MPa for Al₂O₄ (type 3, see Table 2) assumed by the ASTM C633-13 (2021) procedure.

The level of non-uniformity stress distribution in the test specimen was estimated using the following equation:

$$\mathsf{\beta } =\frac{{\mathsf{\sigma }}_{\mathrm{eq}}^{\max }}{{\mathsf{\sigma }}_{\mathrm{eq}}^{\min }},$$

(2)

The non-uniform stress distribution will lead to the misinterpretation of the adhesion tensile testing results [14]. Significant excess of peak stress values at the outer edge, compared with average stress distribution, means that the inherent determination of high or low values of adhesion strength can be assumed by the ASTM C633-13 (2021), GOST 9.304-87 test procedure. The result implies that the bond strength depends not only on the geometry of a specimen [14,16], but also on the degree of plasticity and brittleness of the coating (brittle or ductile fracture).

The main reasons for the appearance of significant non-uniform interface stress distribution should also be mentioned. These are the appearance of additional shear stress at the outer edge of the coating (the Edge effect), and the decrease in the stiffness of the specimen due to the close proximity of the threaded hole in relation to the detachment surface of the coating.

The impact of the Edge effect can be minimized by increasing the diameter of the cylindrical specimen, up to ${\mathrm{d}}_{\mathrm{spec}}$ = 40–60 mm, as well as the length of a cylindrical specimen. The distance from the threaded closed holes to the plane face of a cylindrical specimen with a diameter of ${\mathrm{d}}_{\mathrm{spec}}$ = 25/25.4 mm should be enough to minimise the stiffness reduction. Applying elongated specimens with lengths of ${\mathrm{h}}_{\mathrm{spec}}$ = 38–50 mm and more [9] will lead to more uniform stress distribution.

The finite element analysis and laboratory testing of the elongated cylindrical specimens with a diameter of ${\mathrm{d}}_{\mathrm{spec}}$ = 25/25.4 mm confirmed that the results of estimating bond strength using a standard cylindrical specimen assumed by ASTM C633-13 (2021), GOST 9.304-87 can be significantly higher than the actual tensile adhesion strength [2]. With the exception of the GOST 9.304-87 standard specimen, the equivalent stress peaks were not identified at the centre of the specimens, but rather, at the outer edge in all models.

4. Conclusions

According to the results of this scientific study and the analysis of the existing literature, we were able to make three conclusions. Firstly, the significant peak stress at the outer edge of the specimen, compared to the average stress calculated by tensile testing, implies that the bond strength should be estimated using elongated specimens, according to the ASTM C633-13 (2021), GOST 9.304-87 test procedure. Secondly, compared to conventional tests (EN ISO 14916:2017, ASTM C633-13 (2021), GOST 9.304-87), the accuracy of tensile bond strength estimation may be higher when the new geometry of a cylindrical specimen is applied. Thirdly, to provide a more accurate method, researchers should carry out finite element analysis and an epoxy tensile adhesion test using elongated cylindrical specimens using different diameters, ${\mathrm{d}}_{\mathrm{spec}}$ = 20–60 mm (using the procedure defined in ASTM C633-13 (2021), GOST 9.304-87).