Effect of temperature and deposition power on microstructure and properties of magnetron sputtered thin AlN coatings

1. Introduction

Aluminum nitride (AlN) coatings are used in micro and nanosensory applications^1,2 due to low coefficient of thermal expansion, high thermal conductivity,^3,4,5 resistivity and dielectric properties,³ excellent piezoelectric properties.^6,7,8 In the last decade, AlN has been widely used as a component for the construction of biosensors.^1,9

With the development of micro-, nano- and biosensors, the requirements for the structure and properties of the materials used in the industry are increasing. When using AlN as a piezoelectric material in sensors, surface roughness, mechanical properties and stoichiometric composition play a vital role. Thus, high roughness leads to a decrease in the piezoelectric properties of devices based on surface acoustic waves (SAW) due to poor propagation of the acoustic waves over the surface of the piezoelectric element. The usage of AlN for constructing contact pads or electrodes presents demands on the wear-resistant properties of the coatings in addition to their roughness and mechanical properties. The structure (including roughness), chemical composition and properties of AlN coatings depend on the application method and parameters.^{7,10,11,12,13,14,15,16} The use of AlN coatings in micro and nanosensory applications requires studying the microstructure and properties at the micro and nanoscale. For this purpose, high-precision probe methods such as atomic force microscopy (AFM) and nanoindentation are widely used.^17,18,19

The purpose of this work was to establish the influence of magnetron sputtering parameters (temperature, power and deposition time) on the microstructure, roughness, mechanical, microtribological and conductive properties of the AlN coatings for practical application in micro and nanosensory devices.

2. Materials and Methods

AlN coatings were obtained by reactive DC magnetron sputtering of Al using the VSM 100 (ROBVAC, Russia) magnetron system on silicon substrates of (100) orientation at a constant pressure of 0.78Pa in the closed chamber of the system. The samples were sputtered in the power stabilization mode, which was 80, 100 and 150W at temperatures of $20^{\circ }$C and $100^{\circ }$C (Table 1). The nitrogen flow was constant for all coatings and amounted to 2 sccm (Table 1). The deposition was carried out at a constant rotation speed of the samples. Before each deposition session, the target was cleaned by its sputtering with 100W power maintained in an inert Ar atmosphere for 15min. After deposition, the coated samples were cleaved along a previously applied notch on the back side of the Si substrate for the cross-section formation.

The thickness and microstructure of the coatings were further determined using a Crossbeam 340 scanning electron microscope (SEM, Carl Zeiss Micrography, Germany). During the study, the samples were fixed in a special holder so that the surface of the cross-section was located normally under the electron beam. The studies were carried out using an Everhart–Thornley secondary electron detector with extra high tension (EHT) of 3kV. After this, the samples were glued to the positioning table using conductive tape and the microstructure of their surface was studied, the EHT was 3kV as well.

The stoichiometric composition and chemical analysis of the coatings were assessed using an EDX Oxford X-Max 80 (Oxford Instruments, UK) with 200x magnification and a voltage of 20kV, from the sample surface over the entire area with a wide coverage (approximately $\sim 150$–200$\mu$m by $\sim 400$–500$\mu$m), as well as from individual contrast crystals pointwise.

The microstructure, roughness parameters and mechanical properties were determined using Hysitron 750Ubi (Bruker, USA) nanoindentation system. A diamond conical indenter with a radius of curvature of 226nm and an angle of $60^{\circ }$ at the apex was used. The mechanical properties were determined at a constant load of 100$\mu$N and the results were averaged for nine indentations for each sample. The indentation depth did not exceed the 1/10 of the coating thickness. The microtribological properties were determined using the same device and indenter. To conduct microtribological tests (nanoscratch testing), the nanoindenter was equipped with a two-dimensional transducer. Tests were carried out using specified functions of load and scratch length. Multi-cycle tests consisted of applying three scratches with a length of 5$\mu$m in 100 cycles of 5s/cycle (total path 500$\mu$m, duration 500s) was carried out with a load of 100$\mu$N for all samples. The average value of the friction coefficient was determined over 100 cycles. The specific volumetric wear ($\omega )$ after 100 cycles was calculated from the volume of material removed during the friction test divided by the product of load and distance.

The specific surface electrical resistivity $\rho$ was determined according to the following formula :

where d is coating thickness, m; $R_{\square }$ is surface resistance [$\mathrm{\Omega }∕\square$] (Ohm per square). Surface resistance $R_{\square }$ was measured using the four-probe method on the IUS-3 (SamaraPribor, Russia) unit.

Based on nanoscratching using the nanoindenter (load of 100$\mu$N), the fracture toughness $K_{IC}$ of AlN coatings was calculated. For this purpose, the following formula was used²⁰ :

where $F_{\mathrm{\text{eq}}}$ is driving force of fracture, H; d is depth of the track, m; w is width of the track, m.²⁰ Driving force of fracture $F_{\mathrm{\text{eq}}}$ was calculated as in Ref. 19 : where $F_T$ is vertical (normal) load and $F_V$ is horizontal (lateral) force (N), held during the nanoscratching process.²⁰

To calculate the scratch hardness $H_s$ of the material, the formula was used²¹ :

where A — horizontal projection of the contact area resisting horizontal force, m².

Since the projection of the indenter contact area was hemispherical in shape, then the following formula was used²¹ :

3. Results and Discussion

SEM studies of the surface of the resulting coatings demonstrated a significant effect of temperature and deposition power on the structure of the surface and cross-section of AlN coatings (Figs. 1 and 2). High temperature ($200^{\circ }$C) in combination with a power of 150W led to a formation of granular, porous, advanced microstructure (Fig. 1(c)), thus a dendritic growth pattern of the coating on the cross-section was observed (Fig. 1(f)). A similar granular structure of the coating surface was obtained in Ref. 22 at a deposition pressure of 0.2–0.55Pa and with increasing pressure the grain size decreased.

Reducing the temperature to $20^{\circ }$C at the same power reduced both the grain size and porosity; on the cross-section, there was a dendritic and columnar growth pattern of the coating (Figs. 1(a) and 1(d)). The combination of a temperature of $100^{\circ }$C and a power of 150W allowed for obtaining a crystalline microstructure and a columnar cross-sectional shape (Figs. 1(b) and 1(e)). The crystallite size varied in the range of 150–900nm.

Reducing the temperature and deposition power to $20^{\circ }$C and 80–100W (Fig. 2) resulted in a fine-crystalline microstructure and a columnar cross-section of the coatings. A power of 80W leads to the formation of irregularly-shaped crystallites with a size of 20–90nm and triangular crystallites with a size of 100–150nm on the surface (Fig. 2(a)). The observation revealed columnar and uneven throughout the entire thickness of the coating transverse structure (Fig. 2(d)). Increasing the power to 100W with other unchanged parameters leads to an increase in the size of irregular crystallites to 50–200nm and their size distribution (Fig. 2(b)) and the disappearance of triangular crystallites. The cross-section with a columnar structure became more uniform with clear boundaries between each column (Fig. 2(e)). The columnar structure of AlN coating formation was also reported in Ref. 23.

Reducing deposition time to 20min at a temperature of $20^{\circ }$C and a power of 100W reduced the crystallite size to 20–100nm, the crystallite size distribution became more uniform, and the structure became more compact (Fig. 2(c)). The cross-section also revealed a columnar structure with unclear boundaries between the columns (Fig. 2(f)). The results of the stoichiometric composition observation are shown in Table 2 and Fig. 3. No strict dependence of the stoichiometric composition on precipitation parameters was revealed. It should be noted that at $20^{\circ }$C, 100W, 20min of deposition, a coating with the maximum nitrogen content was obtained compared to other coatings within the study group — 29.71at.% N (sample No. 6).

The analysis of the results demonstrated that the surface roughness significantly depends on the size of the structures on the surface of the coatings (Fig. 3). Within the course of experiments, high temperature led to an increase in surface roughness (Fig. 3(a)). On a sample deposited at $200^{\circ }$C, the roughness parameters Ra and Rq were 178 and 216nm, respectively. The granular surface and dendritic growth form of coatings had high roughness values.

The thickness of the coatings depends on changes in temperature (Fig. 3(b)) at $20^{\circ }$C, 150W and 40min the thickness was 1100nm, at $100^{\circ }$C, 150W and 40min the thickness was 788nm, and at $200^{\circ }$C, 150W and 40min the thickness was 1000nm (Fig. 3(b)). The roughness on a crystalline surface depends on the size of the crystallites — an increase in the average size leads to an increase in roughness. When comparing coating samples deposited with varying power, roughness increased as power changed from 80W to 100W (Fig. 3(c)). The thickness also increased with increasing power, from 585nm to 850nm (Fig. 3(d)). The same phenomena occurred with the surface roughness and thickness values of coatings with increasing deposition time (Figs. 3(e) and 3(f)) — an increase in Ra, Rq and thickness. Of the entire group of compared samples, the coating sample deposited at $20^{\circ }$C, 100W and 20min had the lowest roughness (Ra 4.8nm and Rq 6.2nm) and thickness 320nm (Figs. 3(e) and 3(f)). The change in roughness was obtained in Refs. 2, 9 and 10 when changing the coating deposition parameters. The granular microstructure, surface development, porosity and dendritic cross-sectional shape of deposited coatings lead to reduced mechanical properties (Fig. 4(a)). The elastic modulus and microhardness values for coatings obtained at $20^{\circ }$C and $200^{\circ }$C with a constant power of 150W were significantly lower compared to other coatings within the compared group (Figs. 4(a), 4(c) and 4(e)). In other coatings of crystalline microstructure and columnar cross-section shape, the mechanical properties were higher and depended on changes in the size of the crystallites — with a decrease in the size of the crystallites, the mechanical properties increased. Thus, the highest values of elastic modulus (97GPa) and microhardness (1.8GPa) were obtained on the coating deposited at $20^{\circ }$C, 100W, 20min. It should be noted that the smallest crystallite size was obtained on this coating (Fig. 2(c)) with a uniform size distribution over the surface. Increasing the deposition power and time reduced the overall mechanical properties (Figs. 4(c) and 4(e)).

The values of the mechanical properties of the studied coatings were lower compared to the ones in the literature sources.^24,25,26 This can be explained by the low-nitrogen content in the composition, as well as the geometry of the indenter tip during nanoindentation. Thus, low microhardness was obtained in AlN coatings with 28 at.% N₂, described in Ref. 25 and amounted to 2GPa.

Microtribological tests differ from standard ones in contact areas and contact stresses.^27,28 A multi-cycle test (100 cycles) was carried out on the coatings with a reciprocating motion of a spherical diamond indenter across the surface. For each cycle, the friction coefficient was determined. In Figs. 4(b), 4(d) and 4(f), the average values of the friction coefficient over 100 cycles were demonstrated. After 100 test cycles with a total distance of 500$\mu$m on each coating (Fig. 5), the microtribotest area was visualized (Fig. 6). From the images obtained, the width and depth of each track were determined, and then the specific volumetric wear was calculated (Figs. 4(b), 4(d) and 4(f)).

Based on the results of determining microtribological properties, the relationship and influence of roughness and mechanical properties of deposited coatings on the value of the friction coefficient was established. The coefficient of friction of coatings deposited with increasing temperature increases from 0.065 to 0.118 (Figs. 4(b) and 5). Deposition of coatings at a temperature of $20^{\circ }$C lead to a decrease in the friction coefficient (Figs. 4(d) and 4(f)) to the range of 0.039–0.049 due to a decrease in roughness and an increase in mechanical properties.

The specific volumetric wear completely correlated with the friction coefficient (Figs. 4(b), 4(d) and 4(f)). The lowest friction coefficient (0.039) and specific volumetric wear ($0.8\cdot 10^{-12}$m³/$\mathrm{\text{N}}\cdot \mathrm{\text{m}}$) were obtained for the coating deposited at $20^{\circ }$C, 100W, 20min. This coating had the lowest roughness value (Fig. 3(e)) and the highest mechanical properties (Fig. 4(e)).

The level of conductivity of the coatings (a parameter important for micro- and nanosensory activity) was assessed using the four-probe method and measuring the specific surface resistance (Table 3). The change in such resistance on the coatings under study depended significantly on the ratio of aluminum and nitrogen. It was clearly observed that as the nitrogen content increases, the surface resistivity increases as well (Table 3).

When the temperature changed to $200^{\circ }$C, a coating with 21.16at.% N₂ was obtained, on which the highest resistivity was 0.709$\mu \mathrm{\Omega }\cdot \mathrm{\text{m}}$. The dependence of the resistivity on the chemical composition can be traced on all the coatings under study, except for the coating deposited at $20^{\circ }$C, 100W, 20min. This coating at the highest nitrogen content (29.71 at.%) almost did not change its specific surface resistance compared to the coating deposited at $20^{\circ }$C, 100W and 50min. The absence of a change in resistance on this coating can be explained by a significant decrease in the crystallite size.

In this work, additional studies were carried out on the microhardness and fracture toughness of the coatings,¹⁸ determined by nanoscratch testing (Table 3). These parameters were determined using formulas (2) and (4). A correlation has been established between the values of microhardness determined by nanoindentation and nanoscratch testing methods (Table 3). The results demonstrated higher microhardness values obtained using the nanoscratch testing method. According to Ref. 21, the microhardness determined by the scratch test method can be higher compared to the microhardness obtained by the nanoindentation method, because when scratching, a lateral frictional force acts on the indenter, which must be overcome. The highest microhardness value using the nanoscratch test method was 5.56GPa and was obtained on a coating deposited at $20^{\circ }$C, 100W and 20min.

The fracture toughness of a coating characterizes the ability of a coating to resist external mechanical loads and shows the limit after which the coating begins to fail irreversibly.^18,19,20 Fracture toughness is quantitatively determined by the critical fracture intensity coefficient $K_{IC}$ — the higher values are usually preferable for the applications. As a result of determining $K_{IC}$, it was found that the coating deposited at $20^{\circ }$C, 100W and 20min also has the highest fracture toughness (Table 2). The $K_{IC}$ value for it is 0.947$\mathrm{\text{MPa}}\cdot {\mathrm{\text{m}}}^{1∕2}$. This result is explained by the high mechanical and microtribological properties and low values of the roughness parameters.

4. Conclusions

As a result of this research, the influence of deposition parameters (temperature, power and time) and the stoichiometric composition of thin AlN coatings (thickness from 320nm to 1100nm) on the microstructure, mechanical and microtribological properties was established. The coatings were deposited using direct current reactive magnetron sputtering. The temperature (from $20^{\circ }$C to $200^{\circ }$C), power (from 80W to 150W) and time (from 20min to 50min) of deposition varied. The following research methods were used: Scanning electron microscopy with EDX, nanoindentation, nanoscratch testing and the four-probe method.

High power (150W) and temperature ($200^{\circ }$C) lead to the formation of coatings with high-roughness values, and low mechanical and microtribological properties due to the formation of a coarse-grained structure, high porosity and dendritic growth pattern on the cross-sections of the coating. Reducing the temperature to $20^{\circ }$C and power to 80–100W allowed obtaining a fine-crystalline structure with crystallites evenly and compactly distributed over the surface, as well as low roughness. Such coatings have higher mechanical and microtribological properties.

Surface resistivity was lower on coatings with fine crystalline structure and correlated with the nitrogen content of the coating. When the temperature changed to $200^{\circ }$C, a coating with 21.16 at.% N₂ was obtained, for which the highest resistivity was 0.709$\mu \mathrm{\Omega }\cdot m$. The coating at the highest nitrogen content (29.71 at.%) almost did not change its specific surface resistance (0.432$\mu \mathrm{\Omega }\cdot \mathrm{\text{m}}$) compared to the coating deposited at $20^{\circ }$C, 100W and 50min, which was associated with a significant decrease in the crystallite size.

A correlation has been established between the values of nanoindentation-derived microhardness and nanoscratch testing methods. The highest microhardness value using the nanoscratch test method was 5.56GPa and was obtained on a coating deposited at $20^{\circ }$C, 100W and 20min.

The coating deposited at $20^{\circ }$C, 100W and 20min also demonstrated the highest fracture toughness $K_{\text{IC}}$ 0.947$\mathrm{\text{MPa}}\cdot {\mathrm{\text{m}}}^{1∕2}$ due to high mechanical properties, low roughness and microtribological properties.

From the point of view of the optimal combination of microstructure, mechanical, microtribological properties and electrical resistivity, the most suitable for practical use in micro- and nanosensory applications is an AlN coating with a thickness of 320nm and 29.71at.% N₂, deposited at $20^{\circ }$C, 100W and 20min. This coating has the highest mechanical properties, low roughness and specific surface resistance, as well as low coefficient of friction and specific volumetric wear compared to all coatings studied.

Acknowledgments

Lapitskaya V., Khabarava A. and Chizhik S. acknowledge the support of the Belarus Republican Foundation for Fundamental Research (Grant Nos. T23RNF-132 and T17KIG-009), Nikolaev A. acknowledges the support of the Russian Science Foundation (Grant No. 23-49-10062, https://rscf.ru/project/23-49-10062/). Coating deposition was conducted in the Resource Center for Collective Usage, Research and Education Center “Materials”, Don State Technical University.

ORCID

Vasilina Lapitskaya  https://orcid.org/0000-0002-3245-5945

Andrey Nikolaev  https://orcid.org/0000-0003-3491-4575

Anastasiya Khabarava  https://orcid.org/0000-0002-8780-7717

Evgeniy Sadyrin  https://orcid.org/0009-0000-2227-1299

Sergei Aizikovich  https://orcid.org/0000-0002-2756-5752

Sergei Chizhik  https://orcid.org/0000-0002-5301-0195

Weifu Sun  https://orcid.org/0000-0002-7263-3435