*2.3. KOH Wet Etching*

2.3.1. Principle

The wet etching of III–V nitrides in general involves the formation of an oxide on the surface and the subsequent dissolution of the oxide [24]. The flowing reactions occur when AlN/Al1−xScxN is subjected to the alkaline environment [29]:

$$2\text{AlN} + 3\text{H}\_2\text{O} \quad \overset{\text{KOH}}{\rightarrow} \quad \text{Al}\_2\text{O}\_3 + 2\text{NH}\_3\tag{1}$$

$$\text{AlN} + 3\text{H}\_2\text{O} \quad \overset{\text{KOH}}{\rightarrow} \quad \text{Al(OH)}\_3 + \text{NH}\_3 \tag{2}$$

$$2\text{ScN} + 3\text{H}\_2\text{O} \quad \overset{\text{KOH}}{\rightarrow} \quad \text{Sc}\_2\text{O}\_3 + 2\text{NH}\_3\tag{3}$$

$$\text{ScN} + 3\text{H}\_2\text{O} \quad \overset{\text{KOH}}{\rightarrow} \text{Sc(OH)}\_3 + \text{NH}\_3\tag{4}$$

**Figure 2.** (**a**) Al1−xScxN and SiN<sup>x</sup> hard mask deposition; (**b**) photoresist patterning; (**c**) SiN<sup>x</sup> dry etch; (**d**) resist stripping and Al1−xScxN KOH wet etch. The opening area for the etching is 30 × 200 µm<sup>2</sup> . In this reaction, KOH acts as the catalyst that pushes the equation to the right side. Due to the origin of the reaction, N-polar AlN/Al1−xScxN are preferred to be etched as it is difficult for OH− to make contact with the Al/Sc atoms in the metal polar state because of repulsion from the negatively charged dangling nitrogen bonds [24].

*Micromachines* **2022**, *13*, x FOR PEER REVIEW 6 of 19

(**a**) (**b**)

Al0.64Sc0.36N film.

*2.2. Film Patterning*

illustrates the fabrication process.

**Figure 1.** (**a**) AFM of the Al0.64Sc0.36N Film. (**b**) Rocking curve measurement (Omega Scan) of an

To selectively etch AlN/Al1−xScxN, the film must be patterned so that only locations of interest would be exposed in the etchant. Silicon nitride (SiNx) was chosen as our hard mask, as it has an etch rate of 0.67 nm/min in 30% KOH at 80 °C [55], which is negligible compared to the etch rates of Al1−xScxN. A 200 nm SiN<sup>x</sup> film was deposited on top of the Al1−xScxN film in an Oxford Plasma Lab 100 Plasma-enhanced Chemical Vapor Deposition (PECVD) machine (Oxford Instruments Plasma Technology, Bristol, United Kingdom), and on the backside of the wafer as well to protect the Si substrate during the etch. Afterwards, photoresist MICROPOSIT® S1813 was spin-coated and exposed in a Karl Süss MA6 Mask Aligner (SÜ SS MicroTec SE, Garching, Germany) via contact lithography. The exposed film was developed in TMAH-0.26N developer, then transferred to an Oxford 80 plus Reactive-ion Etching (RIE) etcher (Oxford Instruments Plasma Technology, Bristol, United Kingdom). A 30 s O<sup>2</sup> descum was performed first to remove the remaining resist, followed by 3 min of SiN<sup>x</sup> etch with CHF3/O<sup>2</sup> mixture. Finally, the resist was stripped in MICROPOSIT® remover 1165 with ultrasonic bath at 60 °C and plasma cleaned. Figure 2

**Figure 2.** (**a**) Al1−xScxN and SiN<sup>x</sup> hard mask deposition; (**b**) photoresist patterning; (**c**) SiN<sup>x</sup> dry etch; (**d**) resist stripping and Al1−xScxN KOH wet etch. The opening area for the etching is 30 × 200 µm<sup>2</sup> . **Figure 2.** (**a**) Al1−xScxN and SiN<sup>x</sup> hard mask deposition; (**b**) photoresist patterning; (**c**) SiN<sup>x</sup> dry etch; (**d**) resist stripping and Al1−xScxN KOH wet etch. The opening area for the etching is 30 <sup>×</sup> <sup>200</sup> <sup>µ</sup>m<sup>2</sup> . difficult for <sup>−</sup> to make contact with the Al/Sc atoms in the metal polar state because of repulsion from the negatively charged dangling nitrogen bonds [24].

(1)

(2)

(3)

(4)

#### 2.3.2. Etching Process 2.3.2. Etching Process

A water bath was established for the etching to be conducted in a stable temperature environment. A trough was filled with deionized (DI) water and placed upon an Echo Thermal HP30 hotplate (Torrey Pines Scientific, Inc., Carlsbad, California). Around 300 mL of 30% KOH/diluted KOH was poured into a beaker, which was later transferred into the trough. A Teflon plate was added in between to avoid direct heating. Finally, a thermocouple was submerged into the KOH solution and connected to the hotplate through a proportional–integral–derivative (PID) control loop to adjust the solution temperature. With all these measures, the solution temperature was able to be stabilized within ±1 ◦C during etching. The test sample was cleaved into an 8 <sup>×</sup> 10 mm<sup>2</sup> die and clamped by a tweezer when dipped into the solution. When submerged in the KOH, samples experienced minimal agitation, and upon removal, they were rinsed under DI water and subsequently dried with N2. Finally, the sample was cleaved from the middle (Figure 3a) and a cross-section was imaged in a FEI Quanta 600 Environmental Scanning Electron Microscope (ESEM) (FEI Company, Hillsboro, OR, USA) (Figure 3b). A water bath was established for the etching to be conducted in a stable temperature environment. A trough was filled with deionized (DI) water and placed upon an Echo Thermal HP30 hotplate (Torrey Pines Scientific, Inc., Carlsbad, California). Around 300 mL of 30% KOH/diluted KOH was poured into a beaker, which was later transferred into the trough. A Teflon plate was added in between to avoid direct heating. Finally, a thermocouple was submerged into the KOH solution and connected to the hotplate through a proportional–integral–derivative (PID) control loop to adjust the solution temperature. With all these measures, the solution temperature was able to be stabilized within ±1 °C during etching. The test sample was cleaved into an 8 × 10 mm<sup>2</sup> die and clamped by a tweezer when dipped into the solution. When submerged in the KOH, samples experienced minimal agitation, and upon removal, they were rinsed under DI water and subsequently dried with N2. Finally, the sample was cleaved from the middle (Figure 3a) and a cross-section was imaged in a FEI Quanta 600 Environmental Scanning Electron Microscope (ESEM) (FEI Company, Hillsboro, United States) (Figure 3b).

#### 2.3.3. Data Interpretation 2.3.3. Data Interpretation

Three types of data were extracted from the SEM images: vertical etch depth, lateral etch length (undercutting) and the sidewall angle. The vertical etch depth is defined as the etching depth into the AlN/Al1−xScxN film from the bottom of the SiN<sup>x</sup> hard mask. It can be somewhat ambiguous, as hexagonal-shaped hillocks are known to form after KOH etching [29], which makes it difficult to identify the end point of the etch. Since the phenomenon is presumed to be defect-related in AlN [28,55], we conclude that the 'tip' of the pyramid acts as a mask in this process and thus the etch front should be read as the basal plane of the hexagonal pyramid. Figure 4a is a demonstration of how the etch depth was measured in case of the existence of the hillocks. The lateral etch length is defined as the etch length underneath the SiN<sup>x</sup> hard mask (Figure 4b). Finally, the sidewall angle is simply the angle between the sidewall and base plane. Three types of data were extracted from the SEM images: vertical etch depth, lateral etch length (undercutting) and the sidewall angle. The vertical etch depth is defined as the etching depth into the AlN/Al1−xScxN film from the bottom of the SiN<sup>x</sup> hard mask. It can be somewhat ambiguous, as hexagonal-shaped hillocks are known to form after KOH etching [29], which makes it difficult to identify the end point of the etch. Since the phenomenon is presumed to be defect-related in AlN [28,55], we conclude that the 'tip' of the pyramid acts as a mask in this process and thus the etch front should be read as the basal plane of the hexagonal pyramid. Figure 4a is a demonstration of how the etch depth was measured in case of the existence of the hillocks. The lateral etch length is defined as the etch length underneath the SiN<sup>x</sup> hard mask (Figure 4b). Finally, the sidewall angle is simply the angle between the sidewall and base plane.

**Figure 4.** (**a**) Al0.95Sc0.05N etched for 10 s in 30% KOH at 45 °C. (**b**) Al0.85Sc0.15N etched for 2.5 min in 30% KOH at 65 °C. **Figure 4.** (**a**) Al0.95Sc0.05N etched for 10 s in 30% KOH at 45 ◦C. (**b**) Al0.85Sc0.15N etched for 2.5 min in 30% KOH at 65 ◦C.

#### **3. Results and Analysis 3. Results and Analysis**

*3.1. Etch Result with 30% KOH at 45 °C 3.1. Etch Result with 30% KOH at 45* ◦*C*

Due to the limited film thickness and the vastly different etch rates of films with varying Sc concentrations, it was not practical to use the same etch time when determining the vertical etch rate. Instead, films with lower Sc concentration were etched for a shorter amount of time. The following table (Table 4) illustrates the time used in each case: Due to the limited film thickness and the vastly different etch rates of films with varying Sc concentrations, it was not practical to use the same etch time when determining the vertical etch rate. Instead, films with lower Sc concentration were etched for a shorter amount of time. The following table (Table 4) illustrates the time used in each case:

**Table 4.** Scandium concentration vs. etch time (short etch time in 30% KOH at 45 °C). **Table 4.** Scandium concentration vs. etch time (short etch time in 30% KOH at 45 ◦C).


vertical etch rate measured due to the availability of the sample at the time this experiment was conducted. For the etched samples, cross-section images were taken from several spots to avoid local variations. Figure 5 summarizes the vertical and lateral etch rates vs. Sc concentration for etching in 30% KOH at 45 °C. It should be noted that not all samples fabricated, as mentioned in Section 2, had their vertical etch rate measured due to the availability of the sample at the time this experiment was conducted. For the etched samples, cross-section images were taken from several spots to avoid local variations. Figure 5 summarizes the vertical and lateral etch rates vs. Sc concentration for etching in 30% KOH at 45 ◦C.

It should be noted that not all samples fabricated, as mentioned in Section 2, had their

**Figure 5.** (**a**). Vertical etch rate of Al1−xScxN in 30% KOH at 45 °C. (**b**) Lateral etch rate of Al1−xScxN in 30% KOH at 45 °C (short etch time). **Figure 5.** (**a**). Vertical etch rate of Al1−xScxN in 30% KOH at 45 ◦C. (**b**) Lateral etch rate of Al1−xScx<sup>N</sup> in 30% KOH at 45 ◦C (short etch time).

Intuitively, the vertical etch rate matches our expectation: a steady decline with increasing Sc concentration. The Al0.80Sc0.20N has an average vertical etch rate of 7.58 nm/s, and the Al0.64Sc0.36N has an average vertical etch rate of 3.68 nm/s. Compared to the values of 3.59 nm/s and 2.77 nm/s obtained by K. Bespalova et al. [25] and S. Fichtner et al. [23] respectively, they are not exact matches but in the same order of magnitude. It should be noted that as per the view of A. Ababneh et al. [56], the etch rate of AlN is a strong function of sputtering conditions; thus, this study can be used to predict the trend instead of the absolute etch rate when applied to films deposited under different conditions. The lateral etch rate has several distinctive features requiring further examination. First, the standard deviation of the lateral etch rate is considerably larger than that for the vertical etch. This is due to the finite amount of lateral etching performed. As exhibited in Figure 4a, the film has a lateral etch length of only a few dozen nanometers in some cases, which makes it difficult to accurately measure. Secondly, the etch rate of Al0.72Sc0.28N does not fit in the line. Lastly, the lateral etch rate begins to increase for concentrations in excess of 20% Sc. Intuitively, the vertical etch rate matches our expectation: a steady decline with increasing Sc concentration. The Al0.80Sc0.20N has an average vertical etch rate of 7.58 nm/s, and the Al0.64Sc0.36N has an average vertical etch rate of 3.68 nm/s. Compared to the values of 3.59 nm/s and 2.77 nm/s obtained by K. Bespalova et al. [25] and S. Fichtner et al. [23] respectively, they are not exact matches but in the same order of magnitude. It should be noted that as per the view of A. Ababneh et al. [56], the etch rate of AlN is a strong function of sputtering conditions; thus, this study can be used to predict the trend instead of the absolute etch rate when applied to films deposited under different conditions. The lateral etch rate has several distinctive features requiring further examination. First, the standard deviation of the lateral etch rate is considerably larger than that for the vertical etch. This is due to the finite amount of lateral etching performed. As exhibited in Figure 4a, the film has a lateral etch length of only a few dozen nanometers in some cases, which makes it difficult to accurately measure. Secondly, the etch rate of Al0.72Sc0.28N does not fit in the line. Lastly, the lateral etch rate begins to increase for concentrations in excess of 20% Sc.

To further explore the lateral etching, a second experiment was conducted in which the lateral etch length was extracted over a longer etching period. The number of Sc concentrations in this study was also expanded for higher resolution as shown in Table 5: To further explore the lateral etching, a second experiment was conducted in which the lateral etch length was extracted over a longer etching period. The number of Sc concentrations in this study was also expanded for higher resolution as shown in Table 5:

**Table 5.** Scandium concentration vs. etch time (long etch in 30% KOH at 45 °C). **Table 5.** Scandium concentration vs. etch time (long etch in 30% KOH at 45 ◦C).


Before the long-span etching was performed, a linearity check was conducted to make sure that the etch does not change during the etching process. The film tested was Al0.64Sc0.36N and it was subjected to etches of 60 s, 80 s, 300 s, 1200 s and 2400 s, respectively. A linear regression was performed, and the lateral etching proved to be highly linear with the R<sup>2</sup> = 0.9968. The fact that the same etch rate was retained after 40 min indicates that the etch is reaction-limited, e.g., not confined by mass transferring, and matches the description of AlN etching in KOH given by Mileham et al. [26]. Before the long-span etching was performed, a linearity check was conducted to make sure that the etch does not change during the etching process. The film tested was Al0.64Sc0.36N and it was subjected to etches of 60 s, 80 s, 300 s, 1200 s and 2400 s, respectively. A linear regression was performed, and the lateral etching proved to be highly linear with the R<sup>2</sup> = 0.9968. The fact that the same etch rate was retained after 40 min indicates that the etch is reaction-limited, e.g., not confined by mass transferring, and matches the description of AlN etching in KOH given by Mileham et al. [26] (Figure 6).

**Figure 6.** (**a**)**.** Lateral etch length with respect to etch time of Al0.64Sc0.36N in 30% KOH at 45 °C. (**b**) Lateral etch rate of Al1−xScxN in 30% KOH at 45 °C (long etch time). **Figure 6.** (**a**). Lateral etch length with respect to etch time of Al0.64Sc0.36N in 30% KOH at 45 ◦C. (**b**) Lateral etch rate of Al1−xScxN in 30% KOH at 45 ◦C (long etch time).

The long etch time returned similar results compared with the short etch time. The etch rate experienced a transition where it reaches the lowest at x = 0.125. However, discrepancies still exist for the etch rates of x = 0.28, 0.30, 0.34 and 0.38 when compared to other films with Sc concentration x > 0.125. One explanation might be that these films were deposited in different batches as opposed to the rest of the films. As stated above, what exhibits as different etch rates here might be the result of the different chamber condition when the films were being sputtered. It was apparent that even though they do not fit in the line made from the x = 0.125, 0.15, 0.20, 0.25, 0.32 and 0.36 films, they were able to constitute their own trend line with almost has the same slope. Deeper examination revealed that the etch rate of the x = 0.28 film was lower than expected on two different films deposited in different batches, and further studies should be carried out to better understand these subtle trends in lateral etch rate. The long etch time returned similar results compared with the short etch time. The etch rate experienced a transition where it reaches the lowest at x = 0.125. However, discrepancies still exist for the etch rates of x = 0.28, 0.30, 0.34 and 0.38 when compared to other films with Sc concentration x > 0.125. One explanation might be that these films were deposited in different batches as opposed to the rest of the films. As stated above, what exhibits as different etch rates here might be the result of the different chamber condition when the films were being sputtered. It was apparent that even though they do not fit in the line made from the x = 0.125, 0.15, 0.20, 0.25, 0.32 and 0.36 films, they were able to constitute their own trend line with almost has the same slope. Deeper examination revealed that the etch rate of the x = 0.28 film was lower than expected on two different films deposited in different batches, and further studies should be carried out to better understand these subtle trends in lateral etch rate.

For most Sc concentrations, the sidewall angle of films remains invariant throughout the etching. As demonstrated in Figure 7, an extra-long submerge of the sample does not change the sidewall angle by a visible amount, and for the changes that could be measured, it can be attributed to the tilting of the sample itself during imaging. For most Sc concentrations, the sidewall angle of films remains invariant throughout the etching. As demonstrated in Figure 7, an extra-long submerge of the sample does not change the sidewall angle by a visible amount, and for the changes that could be measured, it can be attributed to the tilting of the sample itself during imaging.

The sidewall is a reflection of the crystal structure of the Al1−xScxN films. As per the findings of W. Guo et al., due to the energy difference between crystalline planes, the c-plane 0001 will be etched first prior to the deterioration of the 1011 planes [31]. Hence, during the anisotropic etching process, the exposed h1123i slip edges between the boundary of the 1011 planes forms the facets that follow the 1212 of the hexagonal crystal structure, behind which lateral etching ceases advancing (Figure 8).

**55.8 °**

**56.7 °**

**Figure 7.** (**a**) Al0.90Sc0.10N etched for 20 s; (**b**) Al0.75Sc0.25N etched for 60 s; (**c**) Al0.64Sc0.36N etched for 80 s; (**d**) Al0.90Sc0.10N etched for 10 min; (**e**) Al0.75Sc0.25N etched for 5 min; (**f**) Al0.64Sc0.36N etched for 20 min. All etching was performed in 30 wt % KOH at 45 °C. The sidewall angle is preserved after the long etch. **Figure 7.** (**a**) Al0.90Sc0.10N etched for 20 s; (**b**) Al0.75Sc0.25N etched for 60 s; (**c**) Al0.64Sc0.36N etched for 80 s; (**d**) Al0.90Sc0.10N etched for 10 min; (**e**) Al0.75Sc0.25N etched for 5 min; (**f**) Al0.64Sc0.36N etched for 20 min. All etching was performed in 30 wt% KOH at 45 ◦C. The sidewall angle is preserved after the long etch.

*Micromachines* **2022**, *13*, x FOR PEER REVIEW 11 of 19

**Figure 8.** (**a**). Exposed 〈1123〉 slip edge in an Al0.95Sc0.05N film etched in 30% KOH at 45 °C for 10 min; (**b**) the sidewall that follows the {1212} facet created by the boundary between the {1011} planes of the same etch condition for 5 min. **Figure 8.** (**a**). Exposed h1123i slip edge in an Al0.95Sc0.05N film etched in 30% KOH at 45 ◦C for 10 min; (**b**) the sidewall that follows the 1212 facet created by the boundary between the 1011 planes of the same etch condition for 5 min. **Figure 8.** (**a**). Exposed 〈1123〉 slip edge in an Al0.95Sc0.05N film etched in 30% KOH at 45 °C for 10 min; (**b**) the sidewall that follows the {1212} facet created by the boundary between the {1011} planes of the same etch condition for 5 min.

Moreover, for a Hexagonal Close-Packed (HCP) unit cell with an axis length *c*/*a*, the sidewall angle, *θ* (Figure 9), of the {1212} facets as a function of lattice length can be calculated as = arctan(/): Moreover, for a Hexagonal Close-Packed (HCP) unit cell with an axis length *c*/*a*, the sidewall angle, *θ* (Figure 9), of the 1212 facets as a function of lattice length can be calculated as *θ* = arctan(*c*/*a*): Moreover, for a Hexagonal Close-Packed (HCP) unit cell with an axis length *c*/*a*, the sidewall angle, *θ* (Figure 9), of the {1212} facets as a function of lattice length can be calculated as = arctan(/):

**Figure 9.** Crystal planes in Al1−xScxN HCP lattice. **Figure 9.** Crystal planes in Al1−xScxN HCP lattice. **Figure 9.** Crystal planes in Al1−xScxN HCP lattice.

against the experimental value (Figure 10).

Using the *c/a* data from the work presented by Ö sterlund et al. [45], and considering the isotropic etching of the {1011} plane, the theoretical sidewall angle *φ* can be plotted against the experimental value (Figure 10). Using the *c/a* data from the work presented by Ö sterlund et al. [45], and considering the isotropic etching of the {1011} plane, the theoretical sidewall angle *φ* can be plotted Using the *c*/*a* data from the work presented by Österlund et al. [45], and considering the isotropic etching of the 1011 plane, the theoretical sidewall angle *ϕ* can be plotted against the experimental value (Figure 10).

**Figure 10.** Comparison of experimental and theoretical sidewall angle vs. Sc concentration. **Figure 10.** Comparison of experimental and theoretical sidewall angle vs. Sc concentration.

The absolute value of the experimental and theoretical angles follows the same trajectory with a parabolic downtrend with increasing Sc concentration until the Sc concentration exceeds 40%. We hypothesize that the side profile is the result of both anisotropic and isotropic etching. At lower Sc concentration, the etch rate on the c-plane {0001} is significantly higher, the momentum of the etch is downwards, and thus the side wall creates a facet that follows the {1212} of the hexagonal crystal structure. At a higher Sc concentration, the low vertical etch rate slows down the descending penetration, allowing the etchant to further react with the sidewall planes already exposed. Therefore, the closer the {1011} planes are to the surface, the more they are etched away; as a result, the sidewall angle becomes lower than that predicted solely based on the anisotropic crystal etching, i.e., lower than *θ*. This can be partially verified by some of the abnormal points on the graph, most of which have a small lateral etch rate (e.g., 12.5% and 42% Sc), which prevents the etching of their corresponding {1011} planes. The absolute value of the experimental and theoretical angles follows the same trajectory with a parabolic downtrend with increasing Sc concentration until the Sc concentration exceeds 40%. We hypothesize that the side profile is the result of both anisotropic and isotropic etching. At lower Sc concentration, the etch rate on the c-plane 0001 is significantly higher, the momentum of the etch is downwards, and thus the side wall creates a facet that follows the 1212 of the hexagonal crystal structure. At a higher Sc concentration, the low vertical etch rate slows down the descending penetration, allowing the etchant to further react with the sidewall planes already exposed. Therefore, the closer the 1011 planes are to the surface, the more they are etched away; as a result, the sidewall angle becomes lower than that predicted solely based on the anisotropic crystal etching, i.e., lower than *θ*. This can be partially verified by some of the abnormal points on the graph, most of which have a small lateral etch rate (e.g., 12.5% and 42% Sc), which prevents the etching of their corresponding 1011 planes.

### *3.2. Etch Results with 30 wt% KOH at 65* ◦*C and 10 wt% KOH at 65* ◦*C*

*3.2. Etch Results with 30 wt % KOH at 65 °C and 10 wt % KOH at 65 °C* The experiment was also carried out at an elevated temperature and with lower KOH concentration to rule out any possible interference to the outcome except for the intrinsic material properties. At elevated temperature, the etching was performed with a short etching time of 150 s, except for Al0.875Sc0.125N, which was etched for 450 s. Etching with lower KOH concentration was performed for 20 min for all Sc alloying concentrations. As a result of the faster etching rate, the vertical etch rate could not be measured. The lateral The experiment was also carried out at an elevated temperature and with lower KOH concentration to rule out any possible interference to the outcome except for the intrinsic material properties. At elevated temperature, the etching was performed with a short etching time of 150 s, except for Al0.875Sc0.125N, which was etched for 450 s. Etching with lower KOH concentration was performed for 20 min for all Sc alloying concentrations. As a result of the faster etching rate, the vertical etch rate could not be measured. The lateral etch rate and sidewall angle of the experiment are presented below (Figure 11):

etch rate and sidewall angle of the experiment are presented below (Figure 11):

**Figure 11.** (**a**) Lateral etch rate at elevated temperature and lower etchant concentration. (**b**) Sidewall angle at elevated temperature and lower etchant concentration. **Figure 11.** (**a**) Lateral etch rate at elevated temperature and lower etchant concentration. (**b**) Sidewall angle at elevated temperature and lower etchant concentration.

A trend consistent with the etching studies reported above was also observed here, where the lowest etch rate was found to be at x = 0.125 and a decreasing sidewall angle was observed with increasing Sc concentration. Under the SEM, white clusters can be seen occasionally near the etch front, which we assume to be the unsoluable reaction by-product Sc(OH)3. Further research still needs to be conducted on the effect of its presence. A trend consistent with the etching studies reported above was also observed here, where the lowest etch rate was found to be at x = 0.125 and a decreasing sidewall angle was observed with increasing Sc concentration. Under the SEM, white clusters can be seen occasionally near the etch front, which we assume to be the unsoluable reaction by-product Sc(OH)3. Further research still needs to be conducted on the effect of its presence.

#### *3.3. Formation of Vertical Sidewall in Al0.875Sc0.125N 3.3. Formation of Vertical Sidewall in Al0.875Sc0.125N*

While differences can be observed across the spectrum, the lateral etch rate unanimously reaches its lowest point when the Sc concentration is at 12.5%. An inspection of the etch results have shown that during the etching, a more vertical side wall can be formed, as shown in Figure 12. While differences can be observed across the spectrum, the lateral etch rate unanimously reaches its lowest point when the Sc concentration is at 12.5%. An inspection of the etch results have shown that during the etching, a more vertical side wall can be formed, as shown in Figure 12.

**Figure 12.** Lateral etch of Al0.875Sc0.125N in (**a**) 30% KOH at 45 °C for 10 min; (**b**–**d**) 10% KOH at 65 °C for 20 min. **Figure 12.** Lateral etch of Al0.875Sc0.125N in (**a**) 30% KOH at 45 ◦C for 10 min; (**b**–**d**) 10% KOH at 65 ◦C for 20 min.

Although (a) was etched with different parameters compared to (b)(c)(d)—plus, (b)(c)(d) were etched for almost the same time—we conjecture that these images are demonstrating the transient response of the same etching dynamics. The KOH etch is slowed significantly at the Si {111} plane and the Al0.875Sc0.125N {1011} plane due to the low etch rate. Because of the energy differences between removing {1011} Al0.875Sc0.125N and {111} Si , the KOH slowly etches Si {111} until it comes into contact with the Al0.875Sc0.125N {1100} planes, which requires lower energy to react than the {1011} planes. As a result, the etchant begins to remove and simultaneously etch Si {111} and {1100} AlScN. By the time the entire {1100} planes were exposed, a vertical sidewall was formed. It is possible that the slow lateral etch rate is necessary but not sufficient for the exposure of the {1100} AlScN to occur. This has been demonstrated with the lateral etching of Al0.85Sc0.15N and Al0.72Sc0.28N, which have slightly higher lateral etch rates than Al0.875Sc0.125N. As shown below (Figure 13), the etchant may preferentially etch Si {111} instead of m-plane Al1−xScxN. Although (a) was etched with different parameters compared to (b–d)—plus, (b–d) were etched for almost the same time—we conjecture that these images are demonstrating the transient response of the same etching dynamics. The KOH etch is slowed significantly at the Si {111} plane and the Al0.875Sc0.125N 1011 plane due to the low etch rate. Because of the energy differences between removing 1011 Al0.875Sc0.125N and {111} Si, the KOH slowly etches Si {111} until it comes into contact with the Al0.875Sc0.125N 1100 planes, which requires lower energy to react than the 1011 planes. As a result, the etchant begins to remove and simultaneously etch Si {111} and 1100 AlScN. By the time the entire 1100 planes were exposed, a vertical sidewall was formed. It is possible that the slow lateral etch rate is necessary but not sufficient for the exposure of the 1100 AlScN to occur. This has been demonstrated with the lateral etching of Al0.85Sc0.15N and Al0.72Sc0.28N, which have slightly higher lateral etch rates than Al0.875Sc0.125N. As shown below (Figure 13), the etchant may preferentially etch Si {111} instead of m-plane Al1−xScxN.

**Figure 13.** (**a**) Al0.85Sc0.15N etched for 20 min in 10 wt % KOH, etch front of AlScN approaching Si {111}; (**b**) KOH preferentially etches Si {111} instead of {1212} AlScN; (**c**) Al0.72Sc0.28N etched for 20 min in 10 wt % KOH, etch front of AlScN aligned with Si {111}; (**d**) KOH preferential etching of Si {111} instead of {1212} AlScN. All etches were performed at 60 °C. **Figure 13.** (**a**) Al0.85Sc0.15N etched for 20 min in 10 wt% KOH, etch front of AlScN approaching Si {111}; (**b**) KOH preferentially etches Si {111} instead of 1212 AlScN; (**c**) Al0.72Sc0.28N etched for 20 min in 10 wt% KOH, etch front of AlScN aligned with Si {111}; (**d**) KOH preferential etching of Si {111} instead of 1212 AlScN. All etches were performed at 60 ◦C.

The {1100} plane etching has been reported before in single-crystal GaN [57] and Al1−xGaxN [58], and as per the findings of W. Chen et al. [57], the preference of its etching in GaN is a result of its smaller dangling bond density, which makes it more stable in KOH than the {1011} plane. Therefore, one explanation might be that the Al0.875Sc0.125N film has the highest activation energy in terms of the lateral etching, which was calculated to be 23.14 kcal/mol based on the data available. Nevertheless, this is one of the limited examples where this is reported in sputtered Al1-xScxN, and more research needs to be conducted to reveal the mechanism behind the vertical sidewall formation. This method, combined with BCl3/Cl<sup>2</sup> dry etching, could potentially be applied in fabricating vertical side walls using a two-step fabrication process, which will benefit the research and production of LWR, laser mirrors, UV LEDs and a variety of MEMS devices. The 1100 plane etching has been reported before in single-crystal GaN [57] and Al1−xGaxN [58], and as per the findings of W. Chen et al. [57], the preference of its etching in GaN is a result of its smaller dangling bond density, which makes it more stable in KOH than the 1011 plane. Therefore, one explanation might be that the Al0.875Sc0.125N film has the highest activation energy in terms of the lateral etching, which was calculated to be 23.14 kcal/mol based on the data available. Nevertheless, this is one of the limited examples where this is reported in sputtered Al1−xScxN, and more research needs to be conducted to reveal the mechanism behind the vertical sidewall formation. This method, combined with BCl3/Cl<sup>2</sup> dry etching, could potentially be applied in fabricating vertical side walls using a two-step fabrication process, which will benefit the research and production of LWR, laser mirrors, UV LEDs and a variety of MEMS devices.
