Effect of SiO₂ Layer Thickness on SiO₂/Si₃N₄ Multilayered Thin Films

Abstract

Silicon nitride (Si₃N₄) materials are widely used in the electronics, optoelectronics, and semiconductor industries, with Si₃N₄ thin films exhibiting high densities, high dielectric constants, good insulation performance, and good thermal and chemical stability. However, direct deposition of Si₃N₄ thin films on glass can result in considerable tensile stress and cracking. In this study, magnetron sputtering was used to deposit a Si₃N₄ thin film on glass, and a silicon dioxide (SiO₂) thin film was introduced to reduce the Si₃N₄ binding force and stress. The effect of the SiO₂ layer thickness on the SiO₂/Si₃N₄ multilayered thin film was explored. The results indicated that the introduction of the SiO₂ layer could improve the weak adhesion characteristics of Si₃N₄ thin films. Moreover, sputtering the SiO₂ layer to 150 nm resulted in the highest hardness and transmittance of the SiO₂/Si₃N₄ multilayered thin films. The findings of this study lay a solid foundation for the application of Si₃N₄ thin films on glass.

1. Introduction

A silicon nitride (Si₃N₄) thin film is a type of compound thin film with a high density, high dielectric constant, good insulation performance, and good thermal and chemical stability, owing to which it is widely used in multiple fields [1,2,3]. For example, in the field of solar cells, it can be used as an anti-reflective film and passivation layer [4,5], and in the semiconductor field it can be used as a surface passivation protective film [6,7] and insulation layer [8,9,10,11]. Moreover, it can serve as a surface coating [12,13,14] in material surface modification applications, or for preparing multilayered and functional materials [15,16], among other applications. In recent years, the application of Si₃N₄ thin films in the field of mobile phones has gradually increased. For example, the excellent scratch resistance and light transmittance features of Honor Diamond Rhinoceros Glass have been attributed to the application of Si₃N₄ thin films and new coating processes.

You Daoming et al. [17] studied the effect of ion-assisted deposition on Si₃N₄ thin films using reactive magnetron sputtering and found that it had a considerable impact on their optical properties and elemental composition, with the refractive index decreasing from 2.87 to 2.17, and the extinction coefficient decreasing from 0.1535 to 0.0031. Zhang Yuntao et al. [14] analyzed the adhesion of Si₃N₄ thin films deposited on stainless steel surfaces by adding nickel, nickel–chromium, and aluminum oxide thin films as transition layers. Research has shown that when the transition layer was nickel or nickel–chromium, a large number of cracks appeared in the Si₃N₄ film, owing to the high thermal stress in the film. Moreover, aluminum oxide film exhibited better matching performance with 304 stainless steel, and its interface toughness was 51.2% higher than that of Si₃N₄ film. Li et al. [18] deposited Si₃N₄ thin films using low-pressure chemical vapor deposition and two-step deposition techniques, followed by separate annealing, effectively improving the uniformity of the Si₃N₄ thin films and reducing stress. Pariksha et al. [5] studied the optical properties of single- and double-layer amorphous Si₃N₄ using plasma-enhanced chemical vapor deposition (PECVD) technology and thin films with different stoichiometries for photovoltaic applications. Research has shown that double-layer SiN_x exhibits better antireflection performance over a wider wavelength range than single-layer SiN_x. Kuo et al. [19] studied a three-layer SiO_x/SiO_xN_y/SiN_x antireflective coating on textured crystalline silicon solar cells to reduce the surface reflectivity of the crystalline silicon substrate and improve the step coverage of the texture structure. Li et al. [20] used intermediate-frequency magnetron sputtering to design SiN_x: H/SiN_x/SiO_xN_y multilayer films for silicon crystal solar cells. It was observed that the reflectivity decreased in the 30–1100 nm wavelength range. Moreover, the reflectivity of the silicon wafers under single-layer Si₃N₄ films was 6.58% when the nitrogen-to-argon ratio (R) was 0.03, and 5.50% when the ratio was 0.05.

For glass substrates, the direct sputtering of Si₃N₄ thin films can result in poor adhesion and internal stress, which can be effectively resolved by designing multilayer films [5,14,20,21,22,23]. The introduction of multilayer films can compensate for the physical performance mismatch between the thin film and substrate in contrast to single-layer films. This study regulated the adhesion and stress of Si₃N₄ thin films by introducing silica thin films [24]. The physical properties of the multilayered film obtained by adjusting the thickness of the SiO₂ layer were used to evaluate the relationship between SiO₂ layer thickness and performance, advancing our understanding of the properties of Si₃N₄ thin films and improving device manufacturing processes.

2. Experimental Methods

2.1. Preparation of SiO₂/Si₃N₄ Thin Films

In this study, a magnetron sputtering device (PVD 500) was used to prepare Si₃N₄ and silicon dioxide (SiO₂) thin films. The target materials were SiO₂ (99.999%) and Si₃N₄ (purity 99.999%) ceramic targets, with argon (Ar) as the working gas. By adjusting the magnetron sputtering time of the SiO₂ layer, SiO₂/Si₃N₄ multilayered films with varying SiO₂ layer thicknesses were deposited. The deposition parameters are listed in

Table 1. Deposition parameters of SiO₂/Si₃N₄ multilayered thin films.

No.	SiO2		Si3N4		Pressure (Pa)	Flow (sccm)	Temperature (°C)
	Power (W)	Time (min)	Power (W)	Time (min)
S0	/	/	150	30	1	45	200
S40	150	40	150	30	1	45	200
S60	150	60	150	30	1	45	200
S80	150	80	150	30	1	45	200
S100	150	100	150	30	1	45	200

.

2.2. Characterization of Thin-Film Structure and Properties

The crystal structure, texture, and stress of the materials were determined by qualitative and quantitative analyses performed using X-ray diffraction (XRD, Rigaku, JPN). The Verios G4 UC Thermoscience field-emission scanning electron microscopy device (SEM, Thermo Fisher Scientific, Czech) was used to observe and analyze the surface morphology of the thin film. Moreover, the chemical bond structures of various elements in the thin films were analyzed using an Axis Supra X-ray photoelectron spectrometer (KRATOS, UK).

A light source emitted light of a certain wavelength, some of which were absorbed by the sample. The intensity of the transmitted light was measured to determine the absorption of specific wavelengths by the sample and the transmittance of the thin film using the Lambda 750s ultraviolet and visible spectrophotometer (PerkinElmer, USA). The testing wavelength range was 300–800 nm with a step size of 1 nm.

Using an HVS-1000B Vickers (USA) hardness tester, a diamond indenter of a specific shape was pressed onto the surface of the thin film at a specified speed under a predetermined load, and the load was removed after maintaining it for a specific period. Subsequently, the hardness of the thin film was calculated by measuring the size of the indentation and in conjunction with the applied load. The angle of the pressure head was 136°, the applied load was 500 GF, and the load application time was 10 s.

During the nanoscratch test (using an APEX-type micromechanical testing system produced by the Center for Tribology, USA), a needle was scratched on the sample surface, and the applied load was gradually increased. Critical state data of the samples were collected using sensors and microscopes to determine the adhesion of the thin film.

3. Results and Discussion

3.1. Sputtering Rate Testing of SiO₂ Thin Films

SiO₂ thin films were deposited by magnetron sputtering, and the thickness of the thin films was measured using a step gauge to determine the relationship between their thickness and the sputtering time. The results are shown in Figure 1.

3.2. X-ray Diffraction (XRD)

Figure 2 shows the XRD patterns of samples S₀ and S₈₀ in the SiO₂/Si₃N₄ multilayered films, where curve a corresponds to sample S₀ and curve b corresponds to sample S₈₀. No obvious diffraction peak existed in curve a, indicating that there was no crystalline Si₃N₄ in the film. Moreover, there was an obvious Mantou peak of Si₃N₄ in the range of 15–30° at 2θ, indicating that the Si₃N₄ film was amorphous Si₃N₄. Curve b was similar to curve a in that there was no obvious diffraction peak, and a Mantou peak was evident, indicating that there was no crystalline Si₃N₄ in the SiO₂/Si₃N₄ multilayered film, and the Si₃N₄ film was amorphous Si₃N₄. The results of the remaining groups of samples were similar to those shown in curve b.

3.3. Scanning Electron Microscopy (SEM)

Figure 3 shows the surface SEM morphologies of the various SiO₂/Si₃N₄ multilayered films. The conductivity of the SiO₂/Si₃N₄ multilayered films is poor, and measurements using field-emission SEM can lead to charge accumulation on the film surface, resulting in poor image quality. Consequently, before the measurement process, the sample surface was treated with a gold spray to improve its conductivity. A conductive adhesive was also used to connect the sample to the metal base to ensure that charges were smoothly transferred from the sample surface.

Figure 3 shows that the surfaces of samples S₄₀, S₆₀, S₈₀, and S₁₀₀ are uneven. By comparison, sample S₀ is relatively flat. This may be owing to the addition of the SiO₂ transition layer, which increases the thickness of the film and gradually increases the unevenness of sputtering, or the introduction of oxygen atoms, which leads to oxygen doping in the Si₃N₄ film and increases defects and stresses in the film. These defects and stresses can damage the continuity and flatness of the film, resulting in surface undulations and unevenness. The surface of sample S₀ was more uniform with a small number of voids, while the surface of sample S₁₀₀ had aggregates of different sizes, forming island clusters of all sizes with a large number of voids. The surfaces of samples S₄₀ and S₆₀ only had a small number of tiny aggregates, and there were fewer voids compared to sample S₁₀₀. The surface of sample S₈₀ appeared smooth and uniform, without obvious voids. This may be because an increase in the thickness of the film may have altered the stress transfer mechanism between the film and substrate, leading to this phenomenon.

3.4. X-ray Photoelectron Spectroscopy (XPS)

According to Table 1, there is only one situation for the outermost layer of SiO₂/Si₃N₄ multilayered film, which is that the Si₃N₄ target material is deposited for 30 min under the following conditions: sputtering power 150 W, sputtering pressure 1 Pa, gas flow rate 45 sccm, and substrate temperature 200 °C. Consequently, sample S₈₀ was selected as the representative sample for XPS analysis and compared with sample S₀.

Figure 4 shows the characteristic peaks and fitting of the Si 2p spectrum. The three peaks are the Si–N bond, Si–O bond, and Si–Si bond, with an electron binding energy of 102.2 ± 0.1 eV for the Si–N bond, 105.2 ± 0.1 eV for the Si–O bond, and 99.7 ± 0.1 eV for the Si–Si bond. By calculating the proportion of the peak intensity of each fitted peak relative to the whole, the specific proportions of Si–N, Si–O, and Si–Si bonds in the thin film can be obtained. The results showed that the content of Si–N bonds in sample S₈₀ was 91.4% and that of Si–Si bonds was 7.8%. Compared to sample S₀, the content of Si–N bonds decreased slightly, the content of Si–Si bonds increased considerably, and the content of Si-O bonds disappeared.

3.5. Optical Performance Analysis

The wide bandgap of Si₃N₄ thin films provides them with good optical transparency. Consequently, they are widely used in optical devices. Figure 5 shows the UV–vis transmittance spectrum of the tested sample. Within the visible wavelength range, the optical transmittance of the SiO₂/Si₃N₄ multilayered film was considerably improved compared with that of the single-layer Si₃N₄ film, with the highest optical transmittance reaching 94.5% at approximately 500 nm. This may be because the introduction of the silica layer can effectively reduce light reflection and increase light transmission, as the reflectivity of light is related to the difference in the refractive index between the two media—that is, the smaller the difference, the lower the reflectivity. Consequently, the SiO₂ transition layer can reduce the difference in the refractive index between the Si₃N₄ film and the substrate or other media, thereby reducing reflection losses.

Samples S₄₀–S₁₀₀ showed that as the thickness of the silica layer increases, the optical transmittance of the SiO₂/Si₃N₄ multilayered film exhibits a trend of first increasing and then decreasing. Based on the interference and diffraction properties of light, when light passes through an interface, it is reflected at that interface, forming a superposition of multiple coherent waves, which can lead to energy loss and light interference. This is because when light is incident on the SiO₂/Si₃N₄ multilayered film, it is reflected and transmitted to the surface of each layer of the film. These reflected and transmitted light waves interfere multiple times within the double-layer film to form complex interference patterns. The optical thickness of the SiO₂ layer in sample S₈₀ was closer to a quarter wavelength. Based on the principle of light interference, the phase difference of the reflected light on the upper and lower surfaces of the SiO₂ film is one cycle, and the interference is eliminated. Consequently, the optical transmittance of sample S₈₀ was the highest. In addition, SiON is generated between the film layers, and the refractive index matched with SiON can lead to higher transmission [25]. The lowest transmittance of sample S₁₀₀ may be due to its uneven film surface, which leads to light scattering and reduced transmittance.

3.6. Vickers Hardness Analysis

Figure 6 shows the Vickers hardness values of the SiO₂/Si₃N₄ multilayered thin films. It is evident that the Vickers hardness of the SiO₂/Si₃N₄ multilayered thin film with an added silica layer is slightly lower than that of the single-layer Si₃N₄ thin film. This may be due to the introduction of silica layers, which cause thin-film erosion, resulting in an increase in impurities and defects in the Si₃N₄ layer and a decrease in hardness.

Figure 7 shows the surface-indentation morphologies of the SiO₂/Si₃N₄ multilayered thin films. None of the five sample groups exhibited any damage, and there were no pores or cracks. This indicates that all five samples have high hardness and good toughness and could withstand the pressure in microhardness testing without damage. From samples S₄₀–S₁₀₀, it is evident that the thickness of the silica layer increases. The hardness shows a trend of first increasing and then decreasing. As the thickness of the SiO₂ film increases, the number of oxygen atoms in the Si₃N₄ film gradually increases. A small amount of oxygen atom doping enables the Si₃N₄ film to better bond with the SiO₂-film interface, thereby playing a repair and compensation role. When excessive oxygen doping leads to an increase in the number of defects, the brittleness of the film may increase, resulting in a decrease in its impact and bending resistance. This implies that when subjected to external forces, the film is more prone to fracture or fragmentation, resulting in a decrease in its mechanical properties.

3.7. Analysis of Nanoscratch Performance

The bonding strength between the thin film and substrate is a key indicator for evaluating the performance of thin films in tribological applications. The scratch test—which is one of the most commonly used and effective methods for testing hard films—plays an important role in evaluating the film’s performance.

Figure 8 shows the critical load corresponding to each sample, and Figure 9 shows the variation curves of the load and friction coefficient with scratch distance for samples S₀-S₁₀₀ under the same indenter curvature radius. As shown in the figure, compared to single-layer Si₃N₄ thin films, the addition of SiO₂ thin films as a transition layer increased the bonding force between the film and the substrate. This is because SiO₂ thin films have similar chemical compositions and structures to glass, which makes it easy for the two to form stable chemical bonds at the interface, achieve good interface bonding, reduce interface defects, and enhance their bonding force. As a transition layer, SiO₂ can also play a buffering role and alleviate the stress that can occur during the preparation of Si₃N₄ films. Moreover, the bonding ability and compatibility between SiO₂ and Si₃N₄ are excellent, which can help achieve good interface bonding during the preparation process, reduce interface defects, and improve the overall performance of the SiO₂/Si₃N₄ multilayered films.

As shown in Figure 8, with an increase in the thickness of the SiO₂ transition layer, the critical load of the multilayered film shows a trend of first increasing and then decreasing. This may be because as the thickness of the SiO₂ layer increases, the doping of a small amount of oxygen atoms promotes bonding between the Si₃N₄ film and the SiO₂ film, and the bonding force between the films increases. However, with increasing oxygen doping, the increase in defects can lead to a decrease in the critical load of the film.

4. Conclusions

The addition of a SiO₂ transition layer to the SiO₂/Si₃N₄ multilayered film increased the critical load, indicating an improvement in the film–substrate bonding force and a reduction in internal stress. Moreover, the hardness of the film decreased whereas the transmittance increased. As the thickness of the silica layer increased, the hardness and critical load of the multilayered film first increased and then decreased. The maximum hardness of the multilayered film was 1168.8 HV and the maximum critical load was 7.09 N. This was because oxygen atom doping enabled the Si₃N₄ film to better combine with the SiO₂ film interface, thereby playing a repair and compensation role. As the thickness of the SiO₂ layer increased, the transmittance of the SiO₂/Si₃N₄ multilayered film first increased and then decreased. After 80 min, the maximum transmittance was 94.5%. The thickness of the SiO₂ layer was 150 nm, which is closer to a quarter-wavelength thickness. Based on the interference principle, it could be inferred that interference cancellation occurs, leading to the maximum transmittance of the sample.