**2.3 Indentation hardness**

170 Ceramic Coatings – Applications in Engineering

CoCr samples the effect of decarburization was examined by determining the intensity ratio of the strongest WC in relation to the W2C peak (IW2C(100)/IWC(100), see Fig. 3 b). Similar values of 0.22 in case of the fine and 0.17 for the coarser fractionated feedstock were obtained indicating a stronger decarburization when spraying the fine feedstock. Compared to the

**C1**

Fig. 2. Comparison of achievable microstructures using fine powder feedstock < 25 µm grain size (left) and conventional more coarsely fractionated feedstock -45+5/20 (right) for Cr2O3

(a), Cr3C2-NiCr (b) and WC-CoCr (c)

**F1** 

**F2** 

**F3** 

Cr3C2-NiCr samples these effects of phase chances were quite low.

One characteristic criterion determining the wear resistance of thermal sprayed coatings is the hardness, which is usually measured by indentation techniques. The Vickers hardness indentation test is well-established both in the course of the quality management of job shops as well as in the characterization of coatings reported in literature. Another technique is the superficial Rockwell hardness testing, by means of that the coatings can be analysed without metallographic preparation. To investigate the suitability of both methods and the influences on the measurement results, a cause-and-effect diagram was established for the indentation testing of thermal spray coatings (see **Figure 4**). The goal of the workings was the reduction of the variability of the measuring results to enhance the comparability.

Fig. 4. Cause-and-effect diagram of the indentation hardness measurement of coatings

Thermal Spraying of Oxide Ceramic and Ceramic Metallic Coatings 173

one. When increasing the loading force to 0.5 kp, the standard deviation is comparable low with 24 to 27. But the calculated mean value of 1327 HV0.5 is considerably higher than all values derived with 0.3 kp loading force. Nevertheless the comparison with other testing series showing less dense microstructures resulted in the conclusion, that the standard deviation of Vickers measurements is lowest with 0.3 kp loading force. When applying 0.1 kp, the indentation pits are too small to be analysed correctly, and the standard deviation rises againg. With the higher force of 0.5 increased cracking occurs due to not optimal cohesion of the coatings. Therefore it is the best solution to choose 0.3 kp loading force to

To investigate the necessary number of measurement repetitions in correlation to the porosity as weakening effect of the coatings cohesion, samples with extraordinary high and relative low porosity were measured 50 times with all techniques. The relative uncertainty of the derived mean value is plotted over the number of repetitions (see **Figure** 5). It is

*mean i*

(1)

1

*mean*

*s n*

3 10 20 30 40 50 Quantity of repititons

*s* (2)

(3)

 HR30N manual [2.5 %] HR30N digital [2.5%] HV 0.3 [2.5%] HR30N manual [16.0 %] HR30N digital [16.0%] HV 0.3 [16.0%]

*n x x* 

*s x* 

1

 

*<sup>i</sup> s*

*n*

Calculation of standard deviation:

 HR30N manual [4.0 %] HR30N digital [4.0%] HV 0.3 [4.0%] HR30N manual [17.0 %] HR30N digital [17.0%] HV 0.3 [17.0%]

**a b**

Calculation of mean standard deviation: *mean*

Calculation of relative uncertainty: *mean*

Fig. 5. Relative uncertainty of indentation hardness values for chromia (a) and Cr3C2-NiCr (b) coatings in relation to the measurement technique, the quantity of measuring repetitions

0,00

0,02

0,04

0,06

Relative uncertainty ε

0,08

0,10

As expected the relative uncertainty of the derived mean hardness value is significantly higher for the samples with high porosity compared to the more dense coatings. The values tend to remain static when more than approximately 20 repetitions are made, whereas this trend is reached after circa the half quantity of measurements when testing the denser

obtain results of high reproducibility.

and the coatings porosity given in brackets

3 10 20 30 40 50 Quantity of repititons

calculated as follows:

0,00

0,01

0,02

0,03

Relative uncertainty ε

0,04

0,05

A large number of predominantly oxide ceramic and cermet coating systems were investigated concerning the different sources of variation depicted in Figure 4. In the following especially the influence of the microstructure, the loading force, the inset type of hardness tester and the necessary quantity of measurement repetitions on one sample regarding the increase of the repeatability are discussed. In the following **Table 1** the indentation hardness values for the coatings shown in **Figure 2** are listed.


1Techniques:

1 = Superficial Rockwell HR15N manual

2 = Superficial Rockwell HR15N digital

3 = Vickers HV0.3

Table 1. Comparison of indentation hardness values derived by superficial Rockwell and Vickers testing

The mean values were derived from 10 measurements for each sample and measurement technique. The results of the measurements of different experimental series were investigated regarding their distribution and the appearance of outliers using the span of standard deviation and Grubb´s test. In most cases the values are not normal distributed, but also hardly any outlier can be detected. Therefore the goal was chosen to reduce the standard deviation of the measurements as the repeatability between different operators, hardness testing devices etc. is expected to increase with decreasing standard deviation. For first evidence the standard deviations of the first 5 measurements and of 10 measurements were compared to get information about the necessary number of measurements to receive robust results.

The results of the two different types of Rockwell hardness testers (one manual Wilson device and a digital type STRUERS DuraJet with closed loop control of the applied force) do not differ very much. The standard deviation is lower than approximately 4 % of the mean measured value and is often higher when it was calculated from ten values instead of the first five ones. This might be due to influences of the microstructure on the results like unmelted particles in the case of chromia and the bimodal hardness distribution of the cermet type coatings. Furthermore the derived mean indentation hardness value is comparable for both the fine and the coarse fractioned feedstock. In case of the Vickers testing the same effect was established. As the Vickers measurements were performed by a minor experienced operator, the tests were repeated by another more experienced person. For the sample C1 a significant lower value of 1124 HV0.3 with comparable standard deviation values of 68 and 79, respectively, were derived. In a further series on the same sample the standard deviation could be reduced significantly to 27 for both 5 and 10 measurements by excluding nonuniformly shaped indentation pits showing different lengths of the two diagonals. The mean value of 1152 HV0.3 seems to be the most reliable

A large number of predominantly oxide ceramic and cermet coating systems were investigated concerning the different sources of variation depicted in Figure 4. In the following especially the influence of the microstructure, the loading force, the inset type of hardness tester and the necessary quantity of measurement repetitions on one sample regarding the increase of the repeatability are discussed. In the following **Table 1** the

indentation hardness values for the coatings shown in **Figure 2** are listed.

**Fraction/** 

1Techniques:

3 = Vickers HV0.3

Vickers testing

robust results.

1 = Superficial Rockwell HR15N manual 2 = Superficial Rockwell HR15N digital

**Feedstock Cr2O3 Cr3C2-NiCr WC-CoCr** 

**Sample** F1 C1 F2 C2 F3 C3 **Technique1)** 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 **Mean** 89,7 85,9 1222 87,4 90,2 1424 87,9 87,2 1032 85,6 85,8 867 87,5 86,8 1172 88,2 85,3 1245 **SD 5 values** 1,1 1,4 45 2,6 1,9 74 1,8 2,3 31,8 2,0 3,9 76 2,3 2,3 26,5 4,6 1,9 148,5 **SD 10 values** 1,0 1,7 37 2,3 2,0 81 2,0 2,5 57,3 1,8 3,5 73 1,9 2,9 84,3 3,4 2,9 142

Table 1. Comparison of indentation hardness values derived by superficial Rockwell and

The mean values were derived from 10 measurements for each sample and measurement technique. The results of the measurements of different experimental series were investigated regarding their distribution and the appearance of outliers using the span of standard deviation and Grubb´s test. In most cases the values are not normal distributed, but also hardly any outlier can be detected. Therefore the goal was chosen to reduce the standard deviation of the measurements as the repeatability between different operators, hardness testing devices etc. is expected to increase with decreasing standard deviation. For first evidence the standard deviations of the first 5 measurements and of 10 measurements were compared to get information about the necessary number of measurements to receive

The results of the two different types of Rockwell hardness testers (one manual Wilson device and a digital type STRUERS DuraJet with closed loop control of the applied force) do not differ very much. The standard deviation is lower than approximately 4 % of the mean measured value and is often higher when it was calculated from ten values instead of the first five ones. This might be due to influences of the microstructure on the results like unmelted particles in the case of chromia and the bimodal hardness distribution of the cermet type coatings. Furthermore the derived mean indentation hardness value is comparable for both the fine and the coarse fractioned feedstock. In case of the Vickers testing the same effect was established. As the Vickers measurements were performed by a minor experienced operator, the tests were repeated by another more experienced person. For the sample C1 a significant lower value of 1124 HV0.3 with comparable standard deviation values of 68 and 79, respectively, were derived. In a further series on the same sample the standard deviation could be reduced significantly to 27 for both 5 and 10 measurements by excluding nonuniformly shaped indentation pits showing different lengths of the two diagonals. The mean value of 1152 HV0.3 seems to be the most reliable one. When increasing the loading force to 0.5 kp, the standard deviation is comparable low with 24 to 27. But the calculated mean value of 1327 HV0.5 is considerably higher than all values derived with 0.3 kp loading force. Nevertheless the comparison with other testing series showing less dense microstructures resulted in the conclusion, that the standard deviation of Vickers measurements is lowest with 0.3 kp loading force. When applying 0.1 kp, the indentation pits are too small to be analysed correctly, and the standard deviation rises againg. With the higher force of 0.5 increased cracking occurs due to not optimal cohesion of the coatings. Therefore it is the best solution to choose 0.3 kp loading force to obtain results of high reproducibility.

To investigate the necessary number of measurement repetitions in correlation to the porosity as weakening effect of the coatings cohesion, samples with extraordinary high and relative low porosity were measured 50 times with all techniques. The relative uncertainty of the derived mean value is plotted over the number of repetitions (see **Figure** 5). It is calculated as follows:

$$\text{Calculation of standard deviation:} \quad s = \sqrt{\frac{\sum\_{i=1}^{n} (\mathcal{X}\_{\text{mean}} - \mathcal{X}\_i)}{n - 1}} \tag{1}$$

$$\text{Calculation of mean standard deviation: } \text{S}\_{\text{mean}} = \frac{s}{\sqrt{n}} \tag{2}$$

$$\text{Calculation of relative uncertainty:} \quad \varepsilon = \frac{\mathcal{S}\_{mean}}{\mathcal{X}\_{mean}} \tag{3}$$

Fig. 5. Relative uncertainty of indentation hardness values for chromia (a) and Cr3C2-NiCr (b) coatings in relation to the measurement technique, the quantity of measuring repetitions and the coatings porosity given in brackets

As expected the relative uncertainty of the derived mean hardness value is significantly higher for the samples with high porosity compared to the more dense coatings. The values tend to remain static when more than approximately 20 repetitions are made, whereas this trend is reached after circa the half quantity of measurements when testing the denser

Thermal Spraying of Oxide Ceramic and Ceramic Metallic Coatings 175

The results of the wear tests after 2500 cycles showed different results for each spray feedstock material. The Cr2O3 coatings regardless whether based on fine or standard powder fractions showed almost no volume loss. According to optical micrographs of the wear scars a tribofilm was formed consisting of plastically deformed debris and splats. This tribofilm was smoother than the original surface and was placed slightly above the mean line of the unworn surface protecting the surface from further wear, see **Figure 6a**. During the experiments of Cr3C2-NiCr samples measurable wear scars were formed, **Figure 6b**. The volume loss was higher on the coating based on the fine powder (0.33 mm3) whereas the wear rate of the standard sample was a bit lower (0.24 mm3). Apparently the above-mentioned phase changes which occurred while processing the fine Cr3C2-NiCr powder influenced the wear behaviour negatively. On the contrary the WC-CoCr coatings regardless whether based on fine or standard powder fractions did not suffer a measurable volume loss. In fact the sintered WC6Co ball was abraded instead of the coatings. This is probably due to the surface finish of the coatings as

fine and carbide rich hard grooves (see **Figure 3c)** abraded the Co-matrix of the ball.

Fig. 6. Optical micrographs of wear scars after 2500 cycles in ball on disk-tests

Fine Cr2O3, Cr3C2-NiCr, and WC-CoCr feedstock with grain sizes below 25 µm were processed in order to investigate the spraying of near net shape coatings. The characteristics of the coatings based on fine powders were analysed and compared to standard coatings based on -45+5/20 µm powder fractions. Compared to standard coatings it was possible to improve the key coating characteristics porosity, surface roughness and corrosion resistance significantly. Other coating properties like hardness or wear resistance showed comparable behaviour as that of standard samples. In case of spraying cermet feedstock, especially Cr3C2-NiCr, optimized parameter sets are necessary to control decarburization and

Thermal Spraying is an indirect process, where only the basic conditions can be controlled by altering the process parameters. A deterministic control of the transfer of heat and kinetic energy to the feedstock particle is not possible. Due to the vast variety of process parameters some time said to be more than one hundred (Lugscheider & Bach, 2002) - sophisticated approaches of designed experiments are a good tool to both understand the complex interdependencies between the parameters and to optimize coatings properties due to the demands. In the following the basic considerations and the proof of suitability of statistical design of experiments are given for controlling and optimization of thermal spraying processes.

a) Cr2O3 validation sample (E1) b) Cr3C2-NiCr validation sample (E2) c) WCCoCr validation sample (E3)

**3. Design and optimization** 

**2.5 Conclusion** 

oxidation.

coatings. Furthermore not the same degree of certainty can be reached when testing coatings with the high porosity. In further workings this tool will be investigated to work out a measuring concept to classify the reliability of indentation hardness testing of thermal sprayed coatings.
