Enhancement of Piezoelectric Response in Yttrium Aluminum Nitride (YxAl1‐xN) Thin Films

Alloying rare earth elements into aluminum nitride (AlN) thin films to increase the piezoelectric response has gained a lot of attention in the past few years. Many rare earth elements are investigated in which scandium alloying resulted in the highest piezoelectric response for AlN. At the same time, researchers have also theoretically explored yttrium alloying as a feasible and economical alternative to scandium. Herein, for the first time, experimentally, the increase of the piezoelectric response of sputter‐deposited YxAl1–xN thin films as a function of increasing yttrium concentration as predicted by density functional theory calculations is demonstrated. Using differently manufactured targets, YxAl1–xN thin films with four different yttrium alloying concentrations (9, 12, 15, and 20 at%). are synthesized. Detailed thin‐film analysis is carried out and the highest value of d33 measured is 12 pC N−1 for Y0.2Al0.8N, which is a 250% increase compared to pure AlN. Even more, the Young's modulus decreases with increasing yttrium concentration in excellent agreement with theoretical predictions. Finally, Y0.15Al0.85N and Y0.2Al0.8N layers show high crystalline stability in pure oxygen environment up to 800 °C, demonstrating high oxidation resistance even under harsh environmental conditions.

50%, [14] indicating that wurtzite Y x Al 1-x N with yttrium concentrations beyond those achievable with scandium are possible.Given the continuous increase of piezoelectric response with increasing scandium concentration in Sc x Al 1-x N up to the point of phase instability, wurtzite Y x Al 1-x N with x > 0.5 could feature even higher piezoelectric coefficients compared to the highest ones achieved with Sc x Al 1-x N.Besides this potential, yttrium is also significantly cheaper compared to scandium, making Y x Al 1-x N an economically much more viable option. [18]iven this prospect, theoretical ab initio calculations have been employed by multiple groups investigating the potential improvement of piezoelectric response in wurtzite Y x Al 1-x N.However, different groups have reported inconsistent results.Tholander et al. [20] have predicted values for d 33 in Y x Al 1-x N significantly below the values for scandium-doped AlN, which they obtained by including the significant volume mismatch between Y and Al atoms into their models.Compared to these results, studies by Manna et al. [21] and by Maryhofer et al. [22] both have predicted an increase of d 33 similar to scandium-alloyed AlN.The lack of experimental data and the disagreement of theoretical predictions have been a driving force for experimental studies of this promising material system.
Initially, cosputtering was used by Zukauskaite et al. [19] and Mayrhofer et al. [22] to grow Y x Al 1-x N thin films.Zukauskaite et al. [19] demonstrated for the first time the formation of wurtzite crystal growth for Y x Al 1-x N (x = 0.04, 0.13, 0.22) thin films, but did not investigate the piezoelectric response.Mayrhofer et al. [22] presented for the first time piezoelectric measurements of Y x Al 1-x N, but could not produce layers with a piezoelectric response matching any of the predictions.In fact, the piezoelectric coefficients were even lower than those of pure AlN thin films.A major reason for this low piezoelectric response was the incorporation of oxygen in the Y x Al 1-x N layer, which was most likely caused by oxygen contamination of the target.This highlights one of the major challenges when working with Y x Al 1-x N, namely, the high affinity of yttrium to oxidation, especially when compared to aluminum, resulting in yttrium oxide compounds (Y 2 O 3 ). [23]Looking at the Ellingham diagrams for oxidation of both aluminum and yttrium, the Gibbs free energy change for aluminum is in the range from -960 to -1020 kJ mol À1 at 200 °C and from -800 to -850 kJ mol À1 at 1000 °C [24-26] compared to -1130 to -1200 kJ mol À1 at 200 °C and from -970 to -1040 kJ mol À1 at 1000 °C for yttrium. [24,25,27]The significantly more negative values of the Gibbs free energy change indicate a higher oxidation affinity of yttrium compared to aluminum, resulting in the aforementioned challenges.
Recently, Schlögl et al. [28] demonstrated for the first time enhanced piezoelectric coefficients for Y 0.09 Al 0.91 N of d 33 ¼ 7.79 pm V À1 compared to 5.02 pm V À1 in pure AlN while still having roughly 2-3 at% oxygen in the layer.They used a target manufactured in a powder metallurgic sintering process, which is prone to oxygen inclusion in the target during target manufacturing.In the same year, Pandit et al. [29] were able to achieve Y 0.15 Al 0.85 N layers without any oxygen inclusion using an arc-melted alloy target and a seed layer deposition approach.This resulted in enhanced piezoelectric coefficients of d 33 ¼ 7.85 pC N À1 as well as significantly reduced full-width-at-half-maximum (FWHM) rocking curves of 2.7°for Y 0.15 Al 0.85 N compared to 5.73°for Y 0.09 Al 0.91 N by Schlögl et al. [28] These results demonstrate the importance of target quality and paved the way for a more systematic study of the impact of yttrium concentration on the properties of Y x Al 1-x N thin films.
In this work, Y x Al 1-x N thin films with yttrium concentrations up to 20 at% are deposited from carefully selected targets to avoid oxygen contamination and investigated regarding their microstructural, chemical, and mechanical properties using X-Ray diffraction (XRD), atomic force microscopy (AFM), scanning electron microscopy (SEM), energy-dispersive X-ray spectroscopy (EDX), high-resolution transmission electron microscopy (HRTEM), and nanoindentation.In addition, the piezoelectric coefficient (d 33 ) is measured.For the first time, we are able to show the increase of d 33 with increasing yttrium concentration, as predicted by Manna et al. [21] and by Maryhofer et al. [22] We therefore experimentally resolve the disagreement in theoretical predictions and are furthermore able to report the highest experimental value of the piezoelectric coefficient d 33 in Y x Al 1-x N so far with 12 pC N À1 .

Experimental Section
Major sources for oxygen contamination in Y x Al 1-x N thin films are the target material and its quality, as well as the chamber walls and the purity of the gases used for the sputter process.Preliminary investigations and previous works [22,28,29] showed that especially the target as a source of oxygen contamination is dominating.To overcome this problem, we evaluated three different target manufacturing methods to achieve high target purity for different yttrium concentrations: Inlet targets, segmented targets (pizza-shaped slices of yttrium and aluminum) and arc-melted alloy targets 9% yttrium alloying was achieved with the help of the inlet target method.For this purpose, a 150 mm Al target was modified by carefully drilling 30 holes across the magnetron race track.All 30 holes were loaded with 99.99% pure yttrium inlets, as shown in the schematic of Table 1a.This target was mechanically workshopped at TU Wien and the 99.99% pure yttrium rods were procured from Edgetech Industries.
12% yttrium alloying into AlN was achieved with the help of a segmented target, as shown in Table 1b.For manufacturing of this target, different sizes of trapezoidal aluminum and yttrium pieces were bonded together.Sindlhauser Materials GmbH manufactured this target.In both cases (inlet and segmented), the use of large bulk volumes of Y compared to a finely grained powder helped to avoid oxygen contamination of the target.15% and 20% yttrium alloyed targets were achieved with vacuum arc melting technology, as shown in Table 1c,d.AJA International provided both targets.The 15% Y-Al alloy target had a diameter of 150 mm, whereas the 20% Y-Al alloy target had a diameter of 100 mm due to high-temperature demands during target casting, thus limiting the target size. [30]o estimate the oxygen concentration in the targets (mentioned in Figure 1), an Y x Al 1-x thin film with a thickness of 1 to 2 μm was sputter deposited in pure argon atmosphere after the in situ target cleaning step.Next, these thin films were investigated with EDX.The accelerating voltage was varied from 3 to 10 keV.Table 1 shows the corresponding EDX analysis of all four Y x Al 1-x thin films resulting from different targets, excluding the presence of oxygen within the measurement accuracy.
All thin films were deposited in a DC magnetron sputtering system LS730S from Von Ardenne on (100) p-type silicon wafers with 100 mm diameter.The Si wafers were cleaned with acetone and isopropanol prior to deposition.The deposition of Y x Al 1-x N thin films included two substrate pretreatment steps.
First, inverse sputter etching (ISE) is used prior to each deposition to clean the surface and remove the native oxide. [6]econdly, a pure, sputter-deposited AlN thin film is deposited as a seed layer for the subsequent deposition of Y x Al 1-x N. [29] AlN seed layers with a high degree of c-axis orientation provide the crystallographic information for the Y x Al 1-x N growth by exposing the regular wurtzite crystal lattice of AlN, thus providing a template for Y, Al, and N atoms to arrange themselves again in the preferred wurtzite lattice configuration.The parameters of the AlN seed layer and ISE are mentioned in Table 2.Even more, Table 3 shows the optimized sputtering parameters obtained after extensive experimental preinvestigation on all the respective sputter targets.

Characterization Methods
A Malvern PANalytical X'pert PRO X-Ray powder diffractometer was used to determine the crystallographic orientation of the polycrystalline thin-film microstructure.The equipment provided a copper tube having a wavelength of X-Rays as 1.540598 Å for Kα 1 and 1.544426 Å for Kα 2 at the operating voltage of 40 kV and a current of 40 mA.For a first screening, a Bragg-Brentano scan was performed from 20 to 80°to identify the crystallographic phases in a straightforward approach.Additionally, rocking curves were also measured for the (002) peak at a fixed value of 2θ at the maximum peak located in the range from 34 to 38°.
To investigate the high-temperature resistance in a pure oxygen environment, high-temperature XRD measurements were performed for Y 0.15 Al 0.85 N and Y 0.2 Al 0.8 N up to 24 h.The equipment chamber was an Anton Paar HTK 1200 N. The oxygen pressure was 1 atmosphere and the oxygen flow rate was 0.4 L min À1 .Rocking curves were continuously measured during different temperature loads varying from 27 to 1200 °C.Given the higher oxidation affinity of yttrium compared to aluminum (see Section 1), it is reasonable to assume that a Table 2. Process parameters for both the deposition of an AlN seed layer and the ISE pretreatment. [29]verse higher ratio of yttrium-to-aluminum atoms (i.e., higher concentrations) will lead to a higher affinity of Y x Al 1-x N thin films toward oxidation.In addition, the atomic size of yttrium was nearly three times that of aluminum.Therefore, yttrium atoms will distort the crystal lattice of Y x Al 1-x N to a certain extent, resulting in an increased defect density in Y x Al 1-x N compared to pure AlN especially at higher concentrations.This could further increase the in-diffusion of oxygen into the layer during hightemperature exposure in oxidizing atmospheres.Given these considerations, only the samples with the highest concentrations were investigated, as they represented both the worst case in terms of oxidation affinity and the most promising layers for application of Y x Al 1-x N in MEMS; as for lower concentrations, the slight increase in piezoelectric response does not warrant the required effort.
An in-depth microstructural analysis was performed by cross-sectional HRTEM analysis.A FEI TECNAI F20 equipment was used for this purpose.An operating voltage of 200 kV was applied to capture the selected-area electron diffraction (SAED) imaging.The images were further processed with the software tool Digital Micrograph.
Standard investigations on, for example, film thickness and microstructure requesting a lower resolution, were performed with an SEM SU8030 from Hitachi applying acceleration voltage of 10 kV.The surface topology of the Y x Al 1-x N thin films was investigated using an AFM Dimension Edge from Bruker to determine the surface roughness and the mean lateral grain size of Y x Al 1-x N thin films.For all the samples, the mean grain size was determined by evaluating the valley-to-valley distance.An n-doped Si cantilever with a spring constant of 42 N m À1 was used, resulting in a resonant frequency of 320 kHz.The image processing and analysis were done with the help of Gwyddion.
Fischer-Cripps Laboratories UMIS nanoindenter equipped with a Berkovich diamond tip was used for nanoindentation.Thin films were subjected to a decrementing indentation force starting from 30 mN with a decrement of 0.5 mN per indent for 30 indents for every sample.The penetration depths for the indents were up to 300-400 nm resulting in negligible influence of the substrate as the thin films had a thickness of about 1 μm.
The piezoelectric constant d 33 was measured with a Piezometer PM300 from Piezo test Ltd.based on the direct loading method.A shadow mask was used for sample preparation enabling the patterning of Al contact pads with a diameter of 1000 μm (thickness: 300 nm).The backside of the silicon substrate was completely coated with an Al thin film again with a thickness of 300 nm.

XRD: Bragg Brentano and Rocking Curves
Prior to Y x Al 1-x N thin-film deposition, Bragg-Brentano measurements are used to verify that wurtzite highly c-axis oriented AlN with a predominant (0002) peak at 36°is indeed synthesized as seed layer.°C in 100% oxygen environment.Figure 6 shows the FWHM of rocking curves for both thin films as a function of temperature.Both layers show a constant FWHM up to 800 °C.Above 900 °C, increased oxygen diffusion into the film occurs, leading to an amorphization of the layer.Gillinger et al. [31] observed similar behavior for pure AlN, where he showed the stability of AlN up to 1000 °C.Similar investigations were performed for ScAlN, [32] demonstrating its crystalline stability up to high temperatures of 1000 °C. [32]A significant novelty of this work is the  °C in typical CMOS semiconductor fabrication process after metallization, this temperature stability of YAlN thin films is sufficiently high for guaranteeing structural and chemical integrity of this material system.The oxidation of Y 0.2 Al 0.8 N starts at slightly lower temperatures compared to Y 0.15 Al 0.85 N (as can be observed from the disappearance of the 0002 reflex at lower temperatures in Figure 4 and 5).One possible explanation is the difference in stress as indicated by the shifted 0002 peak positions (see Figure 2) between both layers, but further studies are required to fully understand this behavior.

Scanning Electron Microscopy and Atomic Force Microscopy (SEM and AFM)
SEM and AFM characterization are performed and a typical resentative of each Y concentration is given in Figure 7.The SEM images are captured under an angle of 45°to give an overview of both the columnar microstructure and surface morphology.Independent of the Y concentration with the exception of 12%, the microstructure of the Y x Al 1-x N thin films replicates the columnar morphology of the AlN seed layer.Figure 7 shows the growth of larger triangular grains in Y 0.12 Al 0.82 N, which is also reflected in the high FWHM values determined from the corresponding rocking curves.At the same time, lower FWHM values correlate with the finely grained columnar growth of Y 0.09 Al 0.91 N, Y 0.15 Al 0.85 N, and Y 0.2 Al 0.8 N. The most homogeneous finely grained microstructure is achieved using the arcmelted targets, indicating that this fabrication method is most promising.The inlet target produces a finely grained microstructure with some larger grains, translating to a larger FWHM.Of all the targets, the pizza-shaped target used for Y 0.12 Al 0.82 N has the most inhomogeneous distribution of aluminum and yttrium on the target, whereas the alloy targets are the most homogeneous.This could be one of the main reasons for the large difference in morphology between the different layers.
Along with the SEM imaging, Figure 7 also shows the surface topography of the samples applying AFM analysis.The scan area of the sample is 0.5 Â 0.5 μm 2 .The RMS roughness values of Y x Al 1-x N range from 0.9 to 1.7 nm as compared to 0.8 nm determined as reference from pure AlN with a similar film thickness.The mean lateral grain size measured for Y x Al 1-x N is between 30 and 60 nm in contrast to 15 nm determined from the AlN reference sample that can be observed in Table 4.
According to Akhiyama et al., [33] a lower lateral grain size plays a major role in achieving high piezoelectric response in ScAlN thin films.A smaller grain size results in low surface roughness, thus leading to enhanced piezoelectric coefficients.

High Resolution Transmission Electron Microscopy (HRTEM)
In addition to the SEM and AFM investigations, the Y 0.15 Al 0.85 N sample was subjected to HRTEM analysis.Figure 8a shows the cross-section image of the Y 0.15 Al 0.85 N sample, whereas 8b indicates the position in the middle section of the Y 0.15 Al 0.85 N film where the SAED pattern was taken, which is presented in Figure 8c.The sharp and bright spots of the (002) crystal orientation indicate the exceptional wurtzite crystal quality of the Y 0.15 Al 0.85 N thin film, thus confirming the XRD results shown in Figure 2 and 3. [29] In addition, a high-resolution image of the oriented grains of Y 0.15 Al 0.85 N shows a superior wurtzite crystal, given in Figure 8d, proving their high quality.

Electromechanical Characterization
Young's modulus is an important parameter of piezoelectric thin films.Ab initio calculations by Mayrhofer et al. [22] and Manna et al. [21] predicted that Y x Al 1-x N becomes softer with increasing yttrium concentration.This mainly originates from the large size of the Y atom when incorporated into the AlN wurtzite crystal by replacing an Al atom.Similar findings are reported about ScAlN, as shown in Figure 9. Starting with around 350 GPa (theoretical) for pure AlN, the value of C 33 for Y 0.09 Al 0.91 N is 292 GPa and  Figure 8. HRTEM analysis of Y 0.15 Al 0.85 N. [29] continuously decreases by %30% to a value of 203 GPa for Y 0.2 Al 0.8 N in good agreement with theoretical predictions.Mayrhofer et al. and Manna et al. theoretically predicted the increase in the piezoelectric coefficients of Y x Al 1-x N with increasing yttrium concentration. [21,22]The predicted increase of piezoelectric coefficient is similar to ScAlN, as shown by experimental results of the Akiyama group. [14]In contrast, Tholander et al. [20] theoretically predicted lower piezoelectric coefficients for Y x Al 1- x N, as presented in Figure 10.
In this work, for the first time, a constant increase in the value of d 33 of Y x Al 1-x N is measured for all investigated alloying concentrations from 9 to 20%, as shown in Figure 10.Y 0.2 Al 0.8 N layers show a value for d 33 of 12 pC N À1 (AE2%), which is the highest value ever recorded for Y x Al 1-x N up to now, resulting in an increase of about 250% compared to pure AlN.This high value of d 33 is achieved by ensuring oxygen-free Y x Al 1-x N layers and by minimizing the rocking curve FWHM using an AlN seed layer approach and careful deposition parameter optimization.Minimizing the FWHM is essential given that AlN [34] and ScAlN [35] [14] at similar alloying concentrations as shown in Figure 10.

Conclusion
In this work, we have successfully and for the first time demonstrated by experimental investigations the enhancement of the piezoelectric coefficient (d 33 ) with increasing yttrium alloying into sputter-deposited AlN.For the four different alloying concentrations investigated, d 33 was increased for the highest yttrium concentration of 20 at% up to 12 pC N À1 (Y 0.2 Al 0.8 N), which is a substantial 250% increase compared to the piezoelectric coefficient of pure AlN and the highest value ever recorded for Y x Al 1-x N up to now.The increase in piezoelectric response also perfectly matched theoretical predictions by DFT calculations from Manna et al. and Mayrhofer et al. while providing experimental evidence against prediction by Tholander et al.Even more, experimental investigations for the Young's modulus also show an excellent agreement with calculations, thus proving that the Y x Al 1-x N (0 ≤ x ≤ 0.2) thin films get softer with increasing yttrium alloy concentration.Through our collaborative research efforts, we unveil insights into the as-of-now poorly investigated material Y x Al 1-x N.These findings emphasize the strong potential of Y x Al 1-x N as a replacement for Sc x Al 1-x N in future PiezoMEMS devices.
2) CT-Th: Theoretical work on Y x Al 1-X N by Christopher Tholander, [20] 3) SM-Th: Theoretical work on Y x Al 1-X N by Surkriti Manna, [21] 4) AlN ref: measured reference sample of AlN, 5) PM-Exp: Experimental work on Y x Al 1-X N from Patrick Mayrhofer, [22] 6) MS-Exp: Experimental work on Y x Al 1-X N from Matthias Schlögl, [28] 7) MA-ScAlN: Experimental study on Sc x Al 1-x N by Morito Akhyiama [14] (values are interpolated, and 8) SP Exp: This work.targeting higher yttrium alloying concentrations above 25%.Up to now, arc melting has produced targets with the highest quality, but due to metallurgical constraints it is currently limited to concentrations up to 20%.For Y x Al 1-x N to become an even more feasible alternative to Sc x Al 1-x N, especially at higher concentrations, significant research efforts have to be put into the development of target manufacturing toward highly pure, oxygen-free Y x Al 1-x materials.

Figure 1 .
Figure 1.Schematic drawings of different types of targets used for the sputter deposition of Y x Al 1-x N thin films.a) Inlet target: Y 9 Al 91 , b) pizza-shaped segmented target: Y 12 Al 88 , c) 150 mm arc-melted alloy target: Y 15 Al 85 , d) 100 mm arc-melted alloy target: Y 20 Al 80 .

Figure 2
shows the Bragg-Brentano scans of all Y x Al 1-x N layers after deposition on the seed layer, demonstrating a strong (0002) peak in all films, thus indicating the presence of a c-axis-oriented wurtzite microstructure in the Y x Al 1-x N layers.The slight shift of the (0002) peaks from 36°to lower values in Y x Al 1-x N compared to pure AlN can be explained by the increased stress generated in Y x Al 1-x N thin films.The FWHM of the rocking curves in Y x Al 1-x N varies from 2.2 to 3°, as shown in Figure 3. 2.2°as the lowest value for sputter-deposited Y 0.2 Al 0.8 N is quite comparable to FWHM values obtained from pure, sputter-deposited AlN being in the range from 1.5 to 2°.

Figure 4
and 5 show the Bragg-Brentano scans for Y 0.15 Al 0.85 N and Y 0.2 Al 0.8 N at temperatures from 27 to 1200

Figure 3 .
Figure 3. FWHM values determined from rocking curves of Y x Al 1-x N thin films with different Y concentrations (x = 9, 12, 15, and 20 at%).The inserted lines serve as guide to the eyes.

Figure 4 .
Figure 4. High-temperature measurement cycle from 27 to 1200 °C in Bragg-Brentano configuration for a Y 0.15 Al 0.85 N thin film in pure oxygen atmosphere.

Figure 5 .
Figure 5. High-temperature measurement cycle from 27 to 1200 °C in Bragg-Brentano configuration for a Y 0.2 Al 0.8 N thin film in pure oxygen atmosphere.

Figure 6 .
Figure 6.Results of the rocking curve measurements for Y 0.15 Al 0.85 N and Y 0.2 Al 0.8 N thin films in pure oxygen atmosphere as a function of postdeposition annealing temperature.The inserted lines serve as guide to the eyes.

Figure 7 .
Figure 7. SEM and AFM images of Y 0.09 Al 0.91 N, Y 0.12 Al 0.88 N, Y 0.15 Al 0.85 N, and Y 0.2 Al 0.8 N.
have previously shown a strong correlation of improved d 33 with a lower FWHM.All measured d 33 values perfectly match to the theoretical predictions by Manna et al. and Mayrhofer et al. while providing strong experimental evidence against the theoretical predictions by Tholander et al. when comparing the interpolation of the value at 50% yttrium concentration and experimental data.Further theoretical calculations at the concentrations used in this work are required to finally resolve the discrepancy in DFT results in literature using the measurements of this work to immediately test future theoretical predictions against experimental data.From an experimental point of view, the d 33 values of Y x Al 1-x N (9,12, 15, and 20 at %) excellently match the d 33 experimental values of Sc x Al 1-x N by Akiyama

Figure 9 .
Figure 9. Theoretical and experimental study of Young's modulus (C 33 ) of Y x Al 1-x N as compared to the state of the art of experimental results from ScAlN thin films.Notations used: 1) PM-Th: Theoretical work on Y x Al 1-X N by Patrick Mayrhofer,[22] 2) CT-Th Theoretical work on Y x Al 1-X N by Christopher Tholander,[20] 3)SM-Th: Theoretical work on Y x Al 1-X N by Surkriti Manna,[21] 4) ScAlN*: Experimental investigation of Young's modulus of ScAlN[36,37] (values are interpolated), and 5) SP-Exp: This work.

Figure 10 .
Figure 10.Theoretical and experimental piezoelectric coefficient (d 33 ) response of Y x Al 1-x N as compared the experimental study of Sc x Al 1-x N from Morito Akhiyama.Notations used: 1) PM-Th: Theoretical work on Y x Al 1-X N by Patrick Mayrhofer,[22] 2) CT-Th: Theoretical work on Y x Al 1-X N by Christopher Tholander,[20] 3) SM-Th: Theoretical work on Y x Al 1-X N by Surkriti Manna,[21] 4) AlN ref: measured reference sample of AlN, 5) PM-Exp: Experimental work on Y x Al 1-X N from Patrick Mayrhofer,[22] 6) MS-Exp: Experimental work on Y x Al 1-X N from Matthias Schlögl,[28] 7) MA-ScAlN: Experimental study on Sc x Al 1-x N by Morito Akhyiama[14] (values are interpolated, and 8) SP Exp: This work.

Table 1 .
EDX analysis of sputter-deposited Y-Al thin films from different Y-Al targets listed in Figure 1.

Table 3 .
Sputter parameters for Y x Al 1-x N thin-film deposition.

Table 4 .
RMS roughness values and grain size of Y 0.09 Al 0.91 N, Y 0.12 Al 0.88 N, Y 0.15 Al 0.85 N, and Y 0.2 Al 0.8 N as compared to AlN.AlN Y 0.09 Al 0.91 N Y 0.12 Al 0.88 N Y 0.15 Al 0.85 N Y 0.2 Al 0.8 N