Plasma-enhanced chemical vapor deposition (PECVD) is an excellent technique for fabricating antireflection films and optical filters. The PECVD technique can be used to prepare photoelectrical films of silicon nitride (SiNx)^[1,2], silicon oxynitride^[3,4], silicon carbonitride^[5,6], and fluorinated silica^[7,8]. These films are transparent in the visible (Vis) near-infrared region, exhibit good wear resistance, high stability, high surface hardness, and a high degree of densification, and thus are used in various applications^[9–13]. The PECVD technique is currently one of the preferred deposition methods for the SiNx films due to a lower thermal budget than other chemical vapor deposition techniques, which are known as a low-temperature (lower than 400°C) deposition technique. Further, the dielectric constant^[14] and chemical components^[15,16] of the thus-deposited SiNx films can be controlled readily by adjusting the deposition parameters; the variable range of the refractive index is 1.83–2.06, and the extinction coefficient is lower than 10−3^[17]. Therefore, the amorphous SiNx films were widely used in the field of optics; for instance, they can be used as optical waveguide materials in integrated optics^[18], biorecognition layers of biosensors in biological applications^[19], and antireflection films and passivation layers in photovoltaics for improving the photoelectric conversion efficiency^[20]. More importantly, the adjustable refractive index^[21], low-absorption, and amorphous state of the SiNx films make them particularly attractive for use as optical thin films, especially films that exhibit a gradient in their refractive index^[22]. Thus, SiNx films, which can be considered as a tailored high refractive index structure, are expected to play a crucial role in many applications and devices.

For the advantages mentioned above, SiNx films deposited by the PECVD technique, were chosen as a high refractive index material to design and to deposit complex optical filters with a gradient refractive index profile, such as broadband antireflection filters, notch filters, and laser protection filters. Besides, SiNx films can be deposited on a Si substrate with SiO2 nanostructures as an optical microstructure device to improve the imaging properties of the optical field. All of these applications are related to the effects of various substrates on SiNx films; this is why we need to do this study.

Several researchers have reported that the optical properties and microstructures of thin films are affected by the material properties of the substrate used^[23–26]. For example, the radio frequency (RF) magnetron sputtering technique was used to synthesize indium gallium nitride (InGaN) crystalline films on n- and p-type Si substrates. It was found that the choice of the substrate material had a significant effect on the optical, morphological, and structural characteristics of the InGaN films^[24]. Other researchers used the same method to deposit zinc oxide (ZnO) crystalline films onto Si, sapphire, polyethylene terephthalate (PET), and polypropylene carbonate substrates and found that the ZnO crystalline films formed on the PET substrates showed better optical and structural characteristics^[25]. These reports suggest that the material properties of the substrate used affect those of the deposited thin films, including their optical characteristics and structures. However, few researchers have studied the influence of substrates on the optical properties of amorphous SiNx films deposited by PECVD at present. Therefore, to improve the manufacturing accuracy of SiNx optical films and to clarify the differences in the characteristics of the films deposited on various substrates, it is necessary to study the effects of various substrates on the optical and structural characteristics of films.

In this Letter, the SiNx films were deposited by the PECVD technique using silane (SiH4) and ammonia (NH3) gases on Si, fused silica (FS), and glass substrates. The structural and optical properties of the films in the Vis near-infrared region were measured. We used the chemical bond configurations to analyze the changed refractive index of SiNx films. By analyzing these results, the effects of various substrates on the structural and optical properties of the films were clarified, which can be referenced by other researchers.

SiNx thin films were deposited by the RF-PECVD technique on a p-type crystalline (100) single-polished Si substrate with a thickness of 500 μm (10mm×10mm), FS substrate with a thickness of 2 mm (Φ25mm), and glass substrate with a thickness of 1 mm (Φ30mm). All the experiments were performed in a class 1000 cleanroom. The substrates were cleaned with a mixture of ethanol and ether (3:1) and then heated for 10 min. Finally, the dried substrates were placed in the vacuum chamber of the coating system. The coating system used was a typical flat-plate PECVD (RF 13.56 MHz) system (PD-220 N™, SAMCO). The upper plate of the chamber had many holes with different sizes and is called the shower plate. All reactive gases were fully mixed through the shower plate before they entered the chamber. Figure 1 shows a schematic diagram of the RF-PECVD coating machine. Before the depositing process, the base pressure of the chamber was 8×10−3Pa, and all the substrates were heated at least 40 min, which means that all the substrates were heated uniformly, avoiding the thermal gradient effects caused by the difference in the sample thickness. The films were deposited at a temperature of 250°C, chamber pressure of 100 Pa, deposition time of 6 min, and RF power of 200 W, and the flow rates of SiH4 (10% in Ar) and NH3 (purity, 99.999%) were 60 sccm (standard cubic centimeters per minute). For convenience, the SiNx thin films formed at Si, glass, and FS substrates were named SiNx1, SiNx2, and SiNx3, respectively.

X-ray diffraction (XRD) using a diffractometer (Advance-D8, Bruker) equipped with a CuKα radiation source having λ=1.5406×10−10m was performed to determine the crystalline states of the films. The X-ray photoelectron spectroscopy (XPS) (K-Alpha, Thermo Fisher) was used to analyze the relative atomic composition and the chemical bonding configuration formed within the various films. XPS measurements were performed using monochromatized X rays from an AlKα source (1486.6 eV). The sample surface was etched using a 2 keV Ar ion beam with a current of 10 mA applied for 30 s. The spectra were analyzed and processed using the Thermo Avantage software. All the XPS spectra were corrected for any charging effects by fixing the C 1s binding energy (BE) at 284.8 eV and were subjected to Shirley background subtraction. The experimental data were evaluated using the Powell fitting algorithm and Gauss–Lorentz peak shape. The BE scale was referenced from the BE database, which can provide enough information about the chemical composition of the samples.

An ultraviolet (UV)-Vis spectrophotometer (Lambda 950, Perkin Elmer) was used to measure the transmittances of the films formed on the glass and FS substrates. The refractive indices, extinction coefficients, and physical thicknesses of the films were determined from a spectroscopic ellipsometer (M-2000UI, J. A. Woollam) using the Cauchy model. The optical film design software TFCalc was used to establish the “air/films/substrate/air” model for fitting the transmittance values of the thin films formed on Si substrates.

The XRD patterns of the non-stoichiometric SiNx thin films formed on different substrates are shown in Figs. 2(a) and 2(b). Figure 2(a) shows the XRD patterns of the SiNx1 thin films formed on the p-type crystalline (100) single-polished Si substrates. It can be seen that all the samples exhibit a strong peak at the same 2θ value. However, this peak is not attributable to the SiNx1 films but is related to the Si substrate (100) itself. Figure 2(b) shows the XRD patterns of the SiNx thin films formed on the glass and FS substrates. No peak related to a crystal plane was observed; this was in keeping with previous reports^[27]. In addition, it is known that as-deposited PECVD films often contain substantial amounts of bonded hydrogen, and that these films are more properly described as hydrogenated SiNx films^[28,29]. In this regard, PECVD SiNx films are both structurally undefined (amorphous) and chemically undefined (non-stoichiometric)^[2,30,31]. Therefore, it is confirmed that the films deposited on all three substrates were amorphous.

Table 1 shows the relative atomic concentrations of elemental Si and N in the SiNx films. The concentrations are the semi-quantitative test results; the uncertainty is ±4% for XPS results because of some variations in the composition depending on the depth of the analysis within the layer. It can be found that the relative atomic concentrations of Si and N of all the SiNx films are similar, and the concentration of elemental Si in the SiNx1 is slightly higher than that of other elements.

In order to analyze the chemical bond configurations of amorphous and non-stoichiometric SiNx films, two models are commonly used to describe the bonds: random bonding model (RBM) and random mixture model (RMM). RBM and RMM are based on the Si-centered tetrahedrons with in-plane triply coordinated N as bonding structures^[32].

Figures 3(a) and 3(b) are the Si 2p and N 1s high-resolution XPS spectra of the SiNx films, respectively, and G represents the glass substrate. Figure 3(a) shows the BE values of the Si 2p main peaks for all the SiNx films are greater than that for Si (99.60 eV)^[33] and less than that for Si3N4 (102.40 eV)^[34], which means the Si–N chemical bond configurations with different types and amounts of Si-centered tetrahedron bonds were formed inside the films^[35]. The XPS Si 2p spectra do not exhibit a low-energy peak near 99.60 eV related to elemental Si, suggesting that no Si–Si bonds or Si clusters were formed in these films. Besides, the results in Fig. 3(a) show that the BE value of the Si 2p main peak of the SiNx films deposited on the Si, FS, and glass substrates decreased gradually from 101.90 eV to 101.70 eV, indicating that the relative concentration of elemental N in the SiNx1 films was higher than that in other films^[36]. Figure 3(b) shows that the BE values of elemental N 1s main peaks for all the SiNx films varied from 397.50 eV to 397.86 eV, which correspond to the BE values for different non-stoichiometric SiNx films^[37]. The atomic-level non-equilibrium mixtures were generated in the plasma of the reactive gas during the deposition process, further, in order to determine the types of the Si–N bonds formed in the samples, the typical Si-centered tetrahedron Si–SivN(4−v) (v=0, 1, 2, 3, 4) bond configurations were used for analysis. The chemical bond configurations based on Si-centered tetrahedrons were most likely to be formed inside the SiNx films that are shown in Table 2^[32,38].

Figure 3(a) shows that the main compounds of the SiNx films formed on FS, glass, and Si substrates were gradually changed from SiN (101.70 eV) to Si3N4 (102.40 eV). The BE values of two subpeaks in Si 2p high-resolution spectra of the SiNx1 films were 101.30 eV (P11) and 102.34 eV (P12). For the SiNx2 and SiNx3 films, the BE values of two subpeaks were 101.30 eV (P21), 102.10 eV (P22), 100.60 eV (P31), and 101.90 eV (P32). It is indicated that more Si–Si3N, Si–Si2N2, and Si–SiN3 bonds were formed in the SiNx2 and SiNx3 films, while the Si–N4 bonds were formed in the SiNx1 films. In addition, the BE values of the two subpeaks in the SiNx3 films deposited on FS substrates are smallest, indicating that the Si-centered tetrahedral bonds with higher Si concentration were formed in the SiNx3 films.

The XPS results show that various Si-centered tetrahedral bonds were formed by the elemental Si and N in the SiNx films, which corresponded to various compounds (i.e., Si–N4 for Si3N4 and Si–SiN3 for SiN). Different compounds exhibit different dielectric constants and densities. As a result, the refractive indices and film thicknesses of the SiNx films were different. Compared with the SiNx2 and SiNx3 films deposited on the glass and FS substrates, the SiNx1 films deposited on the Si substrates displayed Si-centered tetrahedral structures with a higher N content, because of which their properties were similar to those of Si3N4 films. This meant that, under the deposition parameters, the refractive index of the SiNx1 films formed on the Si substrates was greater than those of the films formed on the FS and glass substrates; Ref. [39] shows the same results.

Figure 6 shows the measured transmittance spectra of the substrates and SiNx films coincided at wavelengths that are integer multiples of λ0/2, where λ0 is the reference wavelength, indicating that the SiNx films formed on the glass and FS substrates did not exhibit absorption^[43]. This was consistent with the fact that the extinction coefficients of the films were close to zero. In addition, comparing with the measured transmittance, the calculated transmittance [Simulated (Si)] was shifted slightly. This was mainly owing to the fact that the film thickness of the films formed on various substrates was different; further, the depths of valleys differed due to the differences in the refractive indices. The transmittance deviation of the single-layer SiNx film is greater than 1% at the minimum point, which has obviously exceeded the accuracy of the spectrophotometer (±0.3%).

In summary, SiNx films were deposited by the PECVD technique on Si, glass, and FS substrates. The analyses of the structures of the SiNx films showed that all the deposited films were amorphous. Si-centered tetrahedral Si–SivN4−v bonds were formed in the SiNx films, and the Si–N4 bonds forming in SiNx1 films deposited on the crystalline Si substrates, while the Si–Si3N, Si–Si2N2, and Si–SiN3 bonds formed in SiNx films deposited on the amorphous substrates (glass and FS). The optical properties of the SiNx films showed that the SiNx1 films have the maximum refractive indices (i.e., n=1.83 at 550 nm) in the range of 400–1100 nm, and the SiNx3 films have the lower refractive indices (i.e., n=1.78 at 550 nm). All of the SiNx films displayed the properties of low absorption (the extinction coefficients are lower than 10−4). These substrates have less effect on the deposition rates of the SiNx films.

The PECVD technique is suitable for fabricating amorphous SiNx films that show extremely low absorption in the Vis to near-infrared region. In this study, the types and amounts of chemical bond configurations in the SiNx films were affected by the various substrates, which leads to the differences in the refractive indices. Thus, before designing and fabricating optical multilayer filters and devices, the effects of various substrate materials on structural and optical properties of amorphous SiNx films should be paid attention.

Further work is to study the effects of various deposition parameters on the structural and optical properties of SiNx films, such as reactive gas flow rates and temperatures.