Abstract
The crystalline volume of nanocrystalline silicon (Si) films could be successfully controlled simply by changing the substrate scan speed at the high working pressure of 300 Torr. The Si crystalline volume fraction was increased from 30% to 57% by increasing the scan speed from 8 to 30 mm/s. When the Si film was prepared at a low scan speed (8 mm/s), Si crystals of size 5 nm grew homogeneously through the whole film. The higher scan speed was found to accelerate crystallization, and crystals of size up to 25 nm were deposited in the Si film deposited when the scan speed was 30 mm/s.
1. Introduction
Nanocrystalline silicon (Si) films are mixed-phase materials containing a crystalline Si phase and an amorphous Si phase; they have attracted a lot of interests because of their superior properties, such as high stability against light soaking, high doping efficiency, and improved carrier mobility, compared to those of amorphous Si [1, 2]. Control of the crystalline volume is crucial because the Si crystallinity has a significant influence on the optical gap and conductivity of the materials [3]. For example, in a tandem Si thin film solar cell, the best performance is achieved with a material of low crystalline volume [4]. In contrast, for the gate of a thin film transistor (TFT), nanocrystalline Si with a high crystalline volume and uniform grain size is required for high mobility and good stability [5, 6]. The commonest method of obtaining nanocrystalline Si is plasma-enhanced chemical vapor deposition (PECVD) using a mixture of silane (SiH4) and hydrogen (H2) gases; hydrogenated nanocrystalline Si (nc-Si:H) is formed, and its crystallinity depends on the H2 flow rate or the plasma power [3, 7–9]. Hydrogen radicals are known to preferentially etch amorphous Si, and the exothermic heat generated during etching allows local recrystallization by chemical annealing [10, 11].
Here, we tried to develop a new method of fabricating homogeneous nc-Si:H films. We used a high working pressure plasma-enhanced chemical vapor deposition (HWP-PECVD) system with a cylindrical rotary electrode; this system is superior to conventional PECVD because it has the following features: a high deposition rate as a result of the high partial pressure of the reactive gas and a high plasma density by the very high frequency of 150 MHz; the ability to control the film uniformity because of the homogeneous distribution of reactants by the rotary electrode system; low bombardment damage because of the lower kinetic energy [12–16]. We were able to synthesize nc-Si:H films using HWP-PECVD and even control the crystallinity using a simple parameter, that is, the substrate scan speed. This is in contrast to most research focusing on conventional PECVD, where the crystallinity of nc-Si:H films was controlled by the SiH4 : H2 ratio, as mentioned above.
2. Experimental
The HWP-PECVD system used to produce nc-Si:H was based on previously reported designs (see Figure 1) [12–16]. The diameter of the cylindrical rotary electrode was 200 mm, and the width was 100 mm. The HWP-PECVD of nc-Si:H was performed on 0.7 mm thick soda-lime glass (50 50 mm) substrates at 250°C, with a deposition pressure of 300 Torr. Before deposition, the base pressure of the chamber was reduced to 3 × 10−5 Torr, using dry and turbo molecular pumps. After closing the main valve connected to the turbo molecular pump, helium (He), H2, and SiH4 gases were simultaneously injected into the chamber until a pressure of 300 Torr was attained, and not supplied into chamber during Si film deposition. The purity of the gases was 99.999%. The SiH4 concentration was 0.05%, and the ratio of the H2/SiH4 flow-rates was 70. The electrode rotation speed was 1000 rpm, and the deposition gap between the electrode and the substrate was 0.5 mm. The substrate scan distance was 50 mm, and the scan speed was varied from 8 to 50 mm/s. All of the Si films were 500 nm thick by regulating the number of substrate scan depending on the scan speed. An impedance matching unit supplied 150 MHz VHF power of 200 W to the electrode.
The Si film thickness on the glass was measured by α-step (TENCOR P-11). The deposited films were identified by X-ray diffraction (XRD: X-PERT PRO, PANalytical, Almelo, The Netherlands) using Cu Kα radiation with an angular domain between 20° and 50° (2θ). For confirmation of the Si crystalline volume fraction, Raman spectra were measured using a Jobin Yvon LabRam HR800 (Horibo, Ltd., Kyoto, Japan) UV/micro-Raman spectrometer at room temperature. The measurements were carried out at 514.5 nm using an Ar+ laser, below 50 mW to avoid thermally induced crystallization. The microstructures of the films were observed by transmission electron microscopy (TEM: JEM-2100F, JEOL, Tokyo, Japan) at 200 kV. The TEM specimens for cross-sectional observations were prepared by using Ar ion source of 3.2 keV with precision ion polishing system (PIPS).
3. Results and Discussion
The XRD results show that pure cubic Si was successfully deposited, regardless of the scan speed. The most significant difference among the Si film samples obtained at various scan speeds was that the Si (111) or (220) peaks of films produced at lower scan speeds appeared to be broader. Since the crystallinity reduces the path differences of the X-rays, the degree of crystallinity can be inferred from the full-width half-maximum (FWHM) of the XRD peaks. The differences in the FWHMs imply that the scan speed affected the crystalline volume fraction as well as the crystalline size; the crystallinity improved at faster scan speeds. The crystallinity of the Si films was confirmed by Raman spectroscopy, as shown in Figures 2(b)–2(e). The Si transverse optical (TO) peaks were deconvoluted into their integrated crystalline Gaussian peak (, ~520 cm−1), amorphous Gaussian peak (, ~480 cm−1), and intermediate Gaussian peak (, ~510 cm−1) [17, 18], denoted by pink, blue, and green lines, respectively. In the case of the low substrate scan speed of 8 mm/s, shown in Figure 2(b), the integrated amorphous Gaussian peak comprised the largest proportion of the total area of the three peaks. The crystallinity in the deposited Si film was enhanced by the increase in the substrate scan speed, and thus the prevalent peak was found to be the crystalline Gaussian peak when the Si film was deposited at a scan speed of 30 mm/s. The crystalline volume fraction () was calculated from the simple equation, = ( + )/( + + ), and was found to increase gradually from 30% to 57% as the substrate scan speed increased from 8 to 30 mm/s. We could therefore successfully control the crystallinity of pure Si thin films by the scan speed in the HWP-PECVD method.
(a)
(b)
(c)
(d)
(e)
We performed TEM observations to investigate the detailed microstructures of the Si films. Figures 3(b) and 4(b) show the bright-field TEM (BF-TEM) images of the Si films deposited at scan speeds of 8 mm/s and 30 mm/s, respectively. Both images indicated that the thicknesses of the Si films were all around 500 nm and there appeared to be no defective structures such as voids and cracks. Dark-field TEM (DF-TEM) images were also recorded, as shown in Figures 3(a) and 4(a); DF-TEM images are able to directly reveal the crystallinity of the Si films since the delicate contrast differences, dominant in samples, such as those in this study, comprising mixtures of embedded crystalline and amorphous phases, can be detected by reducing the beam intensity [19]. In the DF-TEM image in Figure 3(a), the crystalline phase exhibits a brighter contrast and a few nanometer-sized Si crystals seem to be homogeneously distributed through the whole Si film. To confirm the results obtained from the DF-TEM images, high-resolution TEM (HR-TEM) images, shown in Figures 3(c), 3(d), and 3(e), were taken from the top, middle, and bottom areas of the Si film. It must first be noted that the HR-TEM images were converted into colored images in which the amorphous phase is green and the crystalline phase is red/yellow. The crystalline phase areas denoted by red/yellow exhibit clear lattice images, but the green areas denoting the amorphous phases do not indicate any atomically ordered structures. Thus, as the DF-TEM image in Figure 3(a) implied, the size of the Si crystals surrounded by the Si amorphous phase was maintained, regardless of the observed area. The crystalline size was determined to be around 5 nm from the HR-TEM images.
In contrast to the Si film obtained at 8 mm/s, crystallization was faster and thus the Si crystals became larger through the whole of the Si film obtained at 30 mm/s. Figure 4 shows that Si crystals of size around 25 nm were observed in the bottom, middle, and top layers. Our results therefore suggest that Si nanocrystals embedded in a Si amorphous phase can be fabricated by the HWP-PECVD method, and the nanocrystalline size can be controlled by simply using the substrate scan speed. This controllability of Si nanocrystalline size by substrate scan speed is found to be related to the effective deposition thickness, as shown in Figure 5. As the substrate scan speed was raised from 8 to 50 mm/s, the effective deposition thickness during each scan was found to decrease from 8 to 1.5 nm. The crystalline size was increased from 5 to 25 nm when the substrate scan speed was increased from 8 to 30 mm/s, and further the size was slightly increased to 30 nm at the speed of 50 mm/s. The enhanced crystal nucleation and growth in the thinner Si film can be explained by classical crystal-growth and nucleation theory, described by where is the critical size of the crystal nucleus, is the concentration of the solute, is the equilibrium solubility, is the atomic volume, and is the interfacial energy per unit area. Since the interfacial energy and the deposition temperature are regarded as being invariant in our HWP-PECVD, regardless of the scan speed, the only variable causing significant changes in the crystal-growth behavior is assumed to be the reactant supersaturation, expressed by /. The thinner Si film is exposed to more of the active radicals and ionized atoms in the plasma, which are involved in augmented reactant supersaturation. Smaller Si crystal nuclei initially form according to (1), and the Si crystal growth is then enhanced because the crystal growth can be improved by smaller crystal nuclei as well as higher reactant supersaturation [20, 21]. Besides, as the effective deposition thickness decreased, the surface diffusion involved with crystal nucleation is thought to be more effective. The substrate scan speed is therefore regarded as being a key factor in the case of HWP-PECVD, unlike the case of conventional low-pressure PECVD techniques without a moving substrate.
4. Conclusions
Nanocrystalline Si films were deposited by HWP-PECVD, using a cylindrical rotary electrode. The crystalline volume fraction of Si could be controlled by changing the substrate scan speed. At a low scan speed of 8 mm/s, the was 30% and Si crystals of size around 5 nm were homogeneously distributed through the whole Si film. As the substrate scan speed was increased to 30 mm/s, the gradually increased to 57%, crystal growth was accelerated, and Si crystals around 25 nm in size were found to form in the deposited Si film.
Acknowledgment
The present research was supported by the research fund of the Korea Institute of Materials Science, a subsidiary branch of the Korea Institute of Machinery and Materials.