In order to illustrate the influence of quartz layer on the deposition effect of SiNx film, we deposited two samples using the same process parameters for 500s. One sample was introduced a quartz as shown in Fig. 1. Another one was a referential sample. Silicon substrate was directly placed on PECVD bottom electrode without any quartz layer. Figure 2a shows SiNx sample after introducing quartz layer, obviously displaying rectangular isochromatic stripes. Different colors in the film represent the corresponding film thickness (Seo et al. 2014). However, the sample without quartz layer demonstrates homogeneous color, indicating uniform thickness. Figure 2c shows the thickness comparison at the different positions along the radial direction. We can clearly notice that the sample with quartz layer presents a characteristic of thickness gradient, showing thin thickness in the center, and thick thickness around comparatively. Also, the thickness of SiNx film with thickness gradient are significantly smaller than that of without quartz layer at any corresponding position, and this difference is becoming smaller as closer to the edge. This work shows that SiNx thin film with thickness gradient can be obtained after introducing dielectric layer into PECVD deposition system.
3.1 Effect of deposition time on thickness gradient of SiNx thin films
In order to investigate the influence of different deposition time on various characteristic data of SiNx thin film with thickness gradient, we singly changed the deposition time to be 400s, 500s, and 600s, respectively, while kept the other process parameters constant. Except the above mentioned, RF power was 80W and dielectric thickness was 1mm.
Figure 3a displays the thickness of the different position for the three samples. It can be seen that the thickness shows symmetrical distribution along the center of the samples, and gradually increases from the center to the periphery. This indicates that the deposition rate also gradually increases from the center to the edge, and the deposition rates of thin films in the central area are about 0.29 nm/s, 0.31 nm/s and 0.32 nm/s for the samples 400s, 500s, and 600s, respectively. It was surprisingly found that the film thickness can be approximately fitted as a power function along the radial direction, which may be explained by the radial diffusion of reactants. The partial voltage of the quartz dielectric layer leads to reduce the peak electric field in the central area, and resultantly reducing the ionization rate of the process gas. With the decrease of gas ionization rate in area of the quartz dielectric layer, the concentration of reaction precursors is the minimum in the center, and gradually increases to the outside, thus results in thickness gradient of the deposited SiNx film, as shown in Fig. 3a. To clearly compare the influence of different deposition time on the thickness distribution of SiNx thin films, we normalized the film thickness based on the minimum value. The result of the normalized thickness distribution is shown in Fig. 3b. It can be clearly seen that the three normalized thickness curves in the radial direction are almost repetitive. This may be due to the inhomogeneous growth of SiNx film with time for the certain position, and the concentration difference of reactants in the reaction chamber is time independent (Smietana et al. 2011).
3.2 Influence of RF Power on Thickness Gradient of SiNx Film
Figure 4a shows the thickness variation for the SiNX films deposited with the different RF power from 80W to 100W. The film thickness of these three samples also shows gradual increase from the center to the edge, showing the characteristics of thin center area and thick edge area.
From Fig. 4a, it can be seen that the film thickness increases gradually with the increasing of deposition power. The deposition rates are calculated to be 0.29 nm/s, 0.32 nm/s and 0.37 nm/s, respectively, and the deposition rate at the edge of 120W deposited sample reaches 0.97 nm/s. The increase in deposition rate is related to the increasing of plasma concentration with change of RF power(Zhang et al. 1996). But this value is still lower than that of traditional uniform thin film with power of 80W, as shown in Fig. 2b, who’s the deposition rate is 0.99 nm/s. We ascribe this to the significant decreasing of RF bias because of the dielectric layer. As we can see, the right values are larger than the left ones, this may be caused by that the measurement points are not strictly symmetrical.
In order to study the effect of RF power on the thickness gradient, we normalized the radial thickness data based on the minimum value, and the radial distribution of film thickness is shown in Fig. 4b. We notice that the gradient for the 80W deposited SiNX is the smallest, while the gradient of the other two samples is almost same. That is to say, when the RF power increases to be a certain value, with the increasing of RF power, the deposition rate will increase, but it has no obvious effect on the radial film thickness gradient. This phenomenon should be related to the radial concentration difference of precursors in the chamber but needs to be further explored.
3.3 The Influence of different dielectric layer thickness on thickness gradient of SiNX film
In the process of capacitively coupled gas discharge, the partial voltage of the dielectric layer will reduce the longitudinal electric field, which will affect the ionization rate of gas molecules and the deposition rate of thin films. Therefore, in the actual production process, the gas discharge process can be influenced by selecting the suitable dielectric layer, so that it is easier to control various characteristic parameters of the thin film(Kawamura et al. 2017).
Figure 5a shows the thickness changing for the SiNX films deposited using the different dielectric layer with the thickness 0.5mm, 1mm, and 2mm, respectively. We can clearly see the deposition rate dramatically increases with decreasing dielectric layer thickness. The deposition rates of central area are 0.19nm/s, 0.29nm/s, and 0.46nm/s for three samples. This variation is because partial bias falls on the dielectric layer. It is more obvious in the thicker dielectric layer, which reduces the space electric field, leading to a decrease in the ionization rate of process gas and the concentration of precursors required for film formation, thus reducing the deposition rate of SiNx films with the increasing of the dielectric layer.
Figure 5b displays the normalized curves for the three samples. The edge thickness of the film is 2.4 times more than the thickness of the central area when the dielectric layer is 2mm, and 1.5 times when the dielectric layer is 0.5 mm in thickness. Therefore, the thickness gradient of the SiNx film can increase with the increase of the dielectric layer thickness, and shows the tendency of sacrificing the deposition speed for a larger film thickness gradient. Xu, etal reported the introduction of dielectric layer leads to uneven distribution of reaction precursors, which will lead to radial diffusion [13]. Therefore, it can be speculated that using a thicker dielectric layer will lead to a greater radial concentration difference of aminoalkane precursors and enhance radial diffusion, thus leading to a greater thickness gradient of SiNx film.
3.4 Refractive index radial variation of SiNx thin films with gradient thickness
Figure 6 depicts the refractive index variation along the radial direction for a typical SiNx film. The results show a tendency to increase gradually from the center to the edge. Generally speaking, the refractive index of SiNx films is mainly linearly related to the silicon/nitrogen ratio and atomic density(Gardeniers et al. 1996). The refractive index of SiNx film is almost stable at 1.86 in the range of 1.5cm in radius. In this range, the refractive index difference of SiNx films has little change due to the small change in thickness. When the radius reaches from 1.5 cm to 3.5 cm, the refractive index increases from 1.85 to 1.98. This should be caused by increasing silicon/nitrogen ratio along with radial direction.