Different stages during CVD deposition on porous substrates

doi:10.1016/S0042-207X(01)00348-7

Vacuum

Volume 64, Issues 3–4, January 2002, Pages 381-386

https://doi.org/10.1016/S0042-207X(01)00348-7 Get rights and content

Abstract

Silicon oxynitride films have been deposited on polished silicon and porous silicon nitride (pore size=0.22 μm) substrates, by plasma assisted chemical vapour deposition (PACVD) from silane (2% in N₂), oxygen and ammonia gas mixtures (SiH₄/O₂/NH₃,2 : 5 : 55). The morphology and thickness of the deposited layers have been analysed by scanning electron microscopy. For identical experimental conditions, the deposition mechanism strongly changes depending on the surface structure (polished or porous). As usual in chemical vapour (CVD) deposition processes, the thickness of a layer deposited on the polished substrate, varies linearly with the deposition time. However, when the porous surface is being coated, a more complex deposition process takes place. Initially, the formation reaction seems to be favoured on the surface protrusions, so that the deposit reproduces the bare porous surface. Thereafter, the covered areas act as nucleation sites and the grains start to grow, giving rise to larger oxynitride particles. When the particle size becomes larger than the substrate pore, a columnar structure is formed giving as a result a uniform layer on the top of the substrate. The deposition mechanism for the CVD coating of a porous surface will be discussed.

Introduction

Chemical vapor (CVD) deposition has been exploited to produce dense as well as microporous films on porous supports by several groups of investigators [1], [2], [3]. The formation of a layer covering a substrate during CVD deposition processes takes place through a heterogeneous reaction on the growing surface, thus the surface plays a key role in the growth mechanism as well as in the properties of the deposited film [4]. As it is known, porous substrates are mainly used for membrane applications and thus the chemical reaction is usually favoured inside the pores for modification purposes. In this cases, infiltration techniques, named chemical vapor infiltration (CVI), are required and the deposit grows directly on the internal wall of the pores. The mechanisms for narrowing the pores have been widely studied, however the formation process of a layer deposited directly on a porous surface is poorly known.

On the other hand, the films deposited by CVD can grow in several ways depending on the conditions, so usually it is not precisely known which is the dominant growth mechanism. According to Pierson [5], the structure of a CVD material can be divided into three major types: columnar grains, faceted columnar growths or fine equiaxed grains. The microstructure varies depending on the material being deposited. Generally, CVD ceramic deposits as silicon oxide or nitride, alumina and most dielectric materials tend to be amorphous or, at least, have a very small grain microstructure. All these aspects are more complex when a porous surface instead of a polished one is used as substrate. In this case, it would be expected the flaws (irregularities) on the growing surface would strongly influence the structure of the deposited coating.

In this work, we show and compare the results obtained during the deposition of CVD silicon oxynitride on polished silicon and porous silicon nitride substrates for increasing deposition times.

Section snippets

Experimental

Silicon oxynitride films were deposited by plasma assisted chemical vapour deposition (PACVD) from silane (2% in N₂), oxygen and ammonia gas mixtures (SiH₄/O₂/NH₃, 2 : 5 : 55). The deposits were obtained at 100 Pa, 300°C and a radiofrequency power of 200 W in a conventional plasma-CVD apparatus radial types (STS, mod 310pc) operating at 13.56 MHz, the deposition time varying from 5 up to 180 min. The samples were deposited onto porous silicon nitride substrates (pore size 0.22 μm, after the supplier

Results

As expected, the deposition rate of the PACVD silicon oxynitride films, grown onto polished silicon substrates, remains almost constant along the process (see Fig. 3) [6]. Since the same conditions are used for all the experiments, the samples present exactly the same composition, as detected by infrared spectroscopy (IR spectra not shown). In addition, the SEM images of these deposits, obtained on polished silicon, show always a homogeneous silicon oxynitride film, likely with a very small

Discussion

Fig. 3 displays the time variation of the mean particle size (evaluated from the SEM top images) as well as the layer thickness of the samples deposited on a porous substrate. Also as comparison, the variation of the thickness of the smooth layers deposited on polished silicon wafers is included in the figure. As it can be observed, the particle size clearly increases with the deposition time, although its variation differs depending on the time range. Thus, for low deposition periods (⩽30 min),

Conclusions

Silicon oxynitride films have been deposited by CVD from SiH₄/O₂/NH₃ gas mixtures, on polished silicon and porous silicon nitride substrates. On polished silicon substrates, smooth oxynitride layers were deposited at a high constant deposition rate (12.1 nm/min). However, when porous substrates are used, grainy appearance films were deposited at lower deposition rates (5.75 nm/min). The particle size of the grainy films increases with the deposition time. Therefore, during the silicon oxynitride

References (8)

C.L. Lin et al.
J Membr Sci
(1994)
L. Huang et al.
Thin Solid Films,
(1999)
M. Tsapatsis et al.
AICHE J
(1992)
Burggraaf AJ, Bouwmeester HJM, Boukamp BA, Uhlhorn RJR, Zarpalis V. In: Nowotny, editor. Science of ceramic interfaces....

There are more references available in the full text version of this article.

Cited by (4)

Characteristics of chemical vapour deposition in micro pore structure in char layer of polymer composites
2020, Polymer Degradation and Stability
Citation Excerpt :
In this stage, the reaction gas and the surface of the char layer can't fully contact and the reaction area is relatively small, resulting in a low deposition rate. In addition, in the pore structure of the char layer, the observed linear growth trend of the deposition reaction rate with temperature differs from the slow growth trend of the deposition rate in the temperature range under the reaction gas flow around the solid surface [29,30] and the exponential growth trend obtained by Hu et al. based on the porous structure with a millimetre pore size [38–40]. Comparing the deposition experimental environment to that in previous studies [29,30,38–40], the reaction gas was found to be in the uniform temperature field protected by argon, and no other factors affected the cracking process of the reaction gas.
The deposition of hydrocarbon pyrolysis gas in the pore structure of the char layer of polymer composites is a special chemical vapour deposition process. While investigation of hydrocarbon pyrolysis gas deposition characteristics is highly significant for obtaining an in-depth understanding of the ablation mechanism of polymer composites and the establishment of accurate ablation models, no studies on this topic have been reported to date. In this study, the deposition characteristics in the pore structure of the char layer were examined using an experimental device designed to simulate pyrolysis gas deposition in the char layer during ablation. The experimental results show that the deposition rate is affected by the change in the pore structure during the deposition process. The deposition process can be divided into rapid deposition in the pore structure, transition deposition and slow deposition on the surface of the char layer. In the temperature range of 800–1000 $°C$ , the deposition rate in the pore structure of the char layer is significant affected by the temperature, showing an obvious linear growth relationship.
Formation of inorganic nanocomposites by filling TiO<inf>2</inf> nanopores with indium and antimony sulfide precursor aerosols
2014, Thin Solid Films
Citation Excerpt :
For pores with diameters between 3 and 10 Ǻ, surface diffusion dominates, while activated diffusion, also called molecular sieving dominates when the pore diameter is less than 3 Ǻ [3,5,6]. Several methods have been used to deposit semiconductor materials onto porous substrates including chemical vapor deposition (CVD) [9,10], electrochemical vapor deposition (EVD) [11,12], atomic layer epitaxy (ALE) [13], spray pyrolysis [14], successive ionic layer adsorption and reaction (SILAR) [1], atomic layer deposition (ALD) [2], plasma sputtering [15] and ion layer gas reaction (ILGAR) [16,17]. ILGAR is a sequential and cyclic deposition technique suitable for the deposition of conformal thin films on different types of substrates [17,18].
Nanocomposites of nanoporous-TiO₂/In₂S₃ and np-TiO₂/Sb₂S₃ were formed by deposition of In₂S₃ or Sb₂S₃ using spray ion layer gas reaction technique from their precursor solutions onto nanoporous TiO₂ substrates at temperatures of 150, 175 and 200 °C. The least penetration of the precursor into np-TiO₂ was achieved for np-TiO₂/In₂S₃ nanocomposites from indium acetylacetonate salt. The deepest penetration was obtained for both np-TiO₂/In₂S₃(Cl) and np-TiO₂/Sb₂S₃ nanocomposites with effective diffusion coefficients of 3.3 × 10^− 3 cm²/s and 3.2 × 10^− 3 cm²/s, respectively. The transport of the precursors in np-TiO₂ and the formation of different nanocomposites were described the regime of the Knudsen diffusion model.
Granular nanocrystalline zirconia electrolyte layers deposited on porous SOFC cathode substrates
2009, Materials Science and Engineering: B
Thin granular yttria-stabilized zirconia (YSZ) electrolyte layers were prepared by chemical vapor synthesis and deposition (CVD/CVS) on a porous substoichiometric lanthanum–strontium–manganite (ULSM) solid oxide fuel cell cathode substrate. The substrate porosity was optimized with a screen printed fine porous buffer layer. Structural analysis by scanning electron microscopy showed a homogeneous, granular nanocrystalline layer with a microstructure that was controlled via reactor settings. The CVD/CVS gas-phase process enabled the deposition of crack-free granular YSZ films on porous ULSM substrates. The electrolyte layers characterized with impedance spectroscopy exhibited enhanced grain boundary conductivity.
Comparison of surface interactions for NH and NH <inf>2</inf> on polymer and metal substrates during NH <inf>3</inf> plasma processing
2002, Journal of Applied Physics

View full text