Introduction

Ceramic matrix composites consisting of an AlN matrix and refractory metal additive, such as Mo, have been explored for applications requiring high thermal conductivity, good mechanical strength, and tailorable electromagnetic properties [1,2,3,4]. Polycrystalline AlN, having thermal conductivities ranging from 80 to 200 W/m-k, depending on composition and microstructure [1, 5, 6], serves as the matrix of the composite. The electromagnetic properties of the composite are tuned by altering the metal additive concentration.

The bulk electrical and mechanical properties of AlN–Mo composites have been studied in detail by Khan et al. for molybdenum additive volume fractions ranging from 5 to 35 vol% [1,2,3,4]. More recently, dielectric properties of AlN:Mo composites with Mo concentrations of around 20 vol% have been studied kHz–MHz regimes [7, 8]. AlN–Mo ceramic matrix composites intended for use as electromagnetic susceptor materials for mm-wavelength radiation [9,10,11] require much smaller Mo concentrations, ranging from around 0.25 to 4 vol%. Previous studies have focused on thermal [12] and electromagnetic properties [11, 13] of AlN–Mo ceramics with Mo concentrations in the 0.25–4 vol% regime; however, to better understand design trades for electromagnetic heat exchangers, for which these materials are intended [9,10,11, 14], mechanical properties data are also necessary. While other recent work has begun to address mechanical properties of AlN:Mo with high Mo loading (> 20 vol% Mo) and various additives, such as BN [15], the present study focuses on comparisons of fracture toughness for the 0.25–4 vol% Mo formulations of AlN–Mo (with carbon additive) studied previously in Refs. [11, 12].

Materials and methods

A total of seven material samples were used in fracture toughness testing. The first six AlN samples contained 5.0 vol% Y2O3, 0.5 vol% carbon, and Mo concentrations of 0.0 vol%, 0.25 vol%, 0.5 vol%, 1.0 vol%, 2.0 vol%, 3.0 vol%, and 4.0 vol%. The yttria (Y2O3) is added as a sintering aid and the carbon is added to the composite formulations to help eliminate oxygen in the AlN and Mo powders to increase thermal conductivity [16]. These carbon-doped (C-doped) samples were formed using hot pressing in an inert gas atmosphere at 1900 °C, as described in Refs. [11, 12]. The seventh sample, included for comparison, was a commercially available ceramic, ST-200 AlN, from Sienna Technologies (Woodinville, WA, USA). The ST-200 contains around 5.0 vol% yttria as a sintering additive. The Authors note that the AlN:Mo composite samples used in the present study are from the same material batches previously characterized microstructurally and compositionally in Refs. [11, 12].

Test samples were prepared in two orientations by diamond machining and polishing to enable indentation both parallel to the pressing direction (A-orientation) and perpendicular to the pressing direction (B orientation). The authors note that because the commercial ST-200 AlN is pressurelessly sintered, only one test orientation was prepared.

Indentation tests were carried out at room temperature (20 °C) with a Vickers diamond indenter under 20 N load using a Lloyd Instruments universal mechanical tester. At least five indentations per sample were made. The indent dimensions and crack lengths were measured using a stage micrometer under an Olympus light microscope.

Fracture toughness (\({K}_{IC}\)) values were calculated using the Anstis formula [17],

$${K}_{IC}=0.0226\left(\frac{{E}^{1/2}{P}^{1/2}a}{{C}^{3/2}}\right),$$
(1)

where \(P\) is the indentation load, \(a\) is the half length of the indent, \(C\) is the half length of the crack, and \(E\) is the elastic modulus. The elastic moduli of all composites were assumed to be 320 GPa in the calculations of \({K}_{IC}\), which is representative of both pure AlN [5, 6, 18] and Mo [19, 20] at room temperature; note that Mo constitutes only a small fraction of the overall composite volume.

Results and discussion

Figure 1a and b shows light microscope images of indentations and indentation-induced cracks in C-doped AlN–Mo (2.0 vol% Mo) composites in the A and B orientations, respectively. The background microstructure shows the composites are well densified. It is also evident that large Mo particles and Mo particle agglomeration are present (as noted previously in Ref. [11]). The dry powder blending technique is responsible for Mo particle agglomeration. Figure 1a also features an example of a crack terminating within a Mo particle.

Fig. 1
figure 1

Images of indentation-induced microcracking in the 2 vol% Mo AlN–Mo composite. The AlN matrix appears as gray and the Mo particles appear as the lighter white color. a Image of an indentation parallel to the pressing direction (A-orientation). Note a crack (arrow) is stopped by a large Mo particle. b Image of an indentation perpendicular to pressing direction (B-orientation). Images were taken at ×50 magnification; note that the figure display size is not to scale and is for illustrative purposes only

Full size image

A second set of light microscope images of indentations and indentation-induced cracks in C-doped AlN–Mo (4.0 vol% Mo) composites in the A and B orientations are provided in Fig. 2a and b, respectively. Figure 2a shows another instance of a crack terminating in a Mo particle, while Fig. 2b shows a set of cracks which are deflected upon intersection with a Mo particle and then partially encircle the particle before reradiating.

Fig. 2
figure 2

Images of indentation-induced microcracking in the 4 vol% Mo AlN–Mo composite. The AlN matrix appears as gray and the Mo particles appear as the lighter white color. a Image of an indentation parallel to the pressing direction (A-direction). Note a crack (a-1) is stopped by a large Mo particle. b Image of an indentation perpendicular to pressing direction (B-direction). Note that cracks (b-1), (b-2), and (b-3) run around Mo particles. Images were taken at ×50 magnification; note that the figure display size is not to scale and is for illustrative purposes only

Full size image

Toughening by particles may be brought about by (i) crack deflection either by compressive stress field or by weak interfaces leading to increases in the tortuosity of crack path, (ii) crack arrest by plastic deformation of ductile particles, and (iii) crack bridging. These phenomena are described in the literature [21,22,23] and generally contribute to measured increases in fracture toughness.

Figure 3 provides a plot of the measured indentation fracture toughness values for the ceramic samples calculated using the Anstis equation, which is described in the previous section. The Anstis equation is known to underestimate the \({K}_{IC}\) [24]; however, it is good for comparison purposes. The calculated value of \({K}_{IC}\) for the ST-200 AlN ceramic (no carbon), 2.07 ± 0.24 MPa m1/2, was found to be close (within 2%) to previously published values of similar pressurelessly sintered AlN compositions measured using the same method [25]. For reference, Vickers-indentation fracture toughness measurements for hot-pressed AlN compositions with [26] and without [4] Y2O3 (both without carbon additives) have been reported as 2.4 ± 0.3 MPa m1/2 and 2.3 MPa m1/2, respectively. Indentations described in Ref. [26] were performed in the B-orientation.

Fig. 3
figure 3

Fracture toughness as a function of Mo concentration for ceramic compositions characterized in this study. Error bars represent one standard deviation from the mean. Plotted data is provided in a tabular format in the Appendix

Full size image

The fracture toughness of the hot-pressed carbon-doped AlN with 0 vol% Mo, plotted in Fig. 3, was found to be below that of the measured ST-200 and that of the hot-pressed AlN examples from the literature. The reduced fracture toughness correlates to the inclusion of the carbon additive (0.5 vol%). Kurokawa et al. state that excess carbon addition (> 1 wt%) significantly lowered thermal conductivity and inhibits densification of AlN ceramics [16]. While the present ceramic compositions contain carbon additive levels that are below Kurokawa et al.’s stated threshold (and thus exhibit good thermal properties and densification [11, 12]), deleterious effects on the ceramic mechanical properties may still be realized.

Interpolation of the data on fracture toughness as a function of Mo loading in the AlN–Mo system described by Khan et al. [4] suggests that, at very low Mo volume fractions (0.25–0.5 vol%), increases in fracture toughness of 3.3–6.5% could reasonably be expected. For the present AlN–Mo system with carbon additives, the B-orientation fracture toughness measurements are found to have increases within this regime (3.7–5.5%); however, the A-orientation measurements show increases in fracture toughness between 10 and 18%.

With continued increases in Mo loading fractions, both the A-orientation and B-orientation fracture toughness generally trend upward, as would be expected from the results of Ref. [4], noting that the B-orientation fracture toughness was found to increase at a faster rate than that of the A-orientation fracture toughness. At loading fractions of 2.0 vol% Mo, the B-direction fracture toughness exceeds the A-orientation fracture toughness, and at 3.0 vol% Mo loading, the B-direction fracture toughness exceeds that of the ST-200 AlN.

From Ref. [11] it is known that, for the level of carbon loading used in the present material system (0.5 vol%), added Mo was found to react with carbon to form Mo2C. This Mo2C phase forms primarily as thick shells surrounding Mo particles in the sintered AlN–Mo ceramic composite [11]. Further, a substantial metallic Mo phase is not detected in the sintered AlN–Mo until loading fractions of 0.5 vol% Mo are exceeded [12]. It is possible that the increase in fracture toughness observed in the 0.25 vol% and 0.5 vol% Mo composites resulted from the removal of elemental carbon from the AlN matrix as it reacted to form Mo2C, localized at the Mo particle sites. Although the addition of low volume fractions of Mo correlates to an observed increase in fracture toughness, it would not be expected to contribute significantly to particle strengthening of the composite as the coefficient of thermal expansion (CTE) and elastic modulus of Mo are very close to that of AlN.

Conclusions

The fracture toughness of AlN ceramic matrix composites containing carbon (0.5 vol%), yttria (5.0 vol%), and Mo (0.0–4.0 vol%) was evaluated using Vickers indentation. The results of these measurements were compared to a carbon-free, commercial AlN (ST-200 ALN) as well as to AlN and AlN–Mo compositions from the literature. The presence of added carbon was found to correlate with a 21% reduction in fracture toughness of the 0.0 vol% Mo carbon-containing sample, compared to the ST-200 AlN, in both the A-orientation (indentation parallel to the pressing direction) and the B-orientation (indentation perpendicular to the pressing direction). Mo additions at small loading fractions (~ 0.25 vol%) were found to exhibit greater-than-expected increases in fracture toughness in the A-orientation, when compared to literature data on AlN–Mo composites. This increase in fracture toughness correlates to the removal of elemental carbon in the AlN matrix through reaction with the Mo additive, forming Mo2C, localized at the Mo particle sites. Further increasing Mo loading was observed to result in generally increasing fracture toughness values, as would be expected from literature data, noting that B-orientation fracture toughness was found to increase at a faster rate than that of the A-orientation fracture toughness.

For AlN–Mo ceramic matrix composites, the 3.0 vol% Mo regime has been previously identified as a region of interest for specific microwave-absorbing heat exchanger configurations [9, 10]. The observed ~ 15% difference in fracture toughness in the A-orientation and B-orientation at loading fractions in this Mo loading regime would therefore be an important design consideration when choosing brazing configurations for ceramic-to-metal interfaces in such a structure.