ICMCTF1999 Session D2: Synthesis and Characterization of Diamond and Related Materials

It was shown lately that the addition of very small amounts of nitrogen in the feed gas could drastically influence the diamond film morphology. Nitrogen is always present in more or less concentration in most commercially available gases. In addition, in most deposition installations, vacuum leaks can be responsible for a non-negligible contribution of nitrogen in the plasma, since N₂ is the major constituent of air. A strict and permanent control over the process parameters is therefore crucial to achieve some reproducibility in diamond deposition experiments. Today, the mechanisms of nitrogen incorporation in a diamond film are still an unsolved riddle. This is mainly due to the complexity of the processes involved, as they do not only depend on empirical process parameters.

To increase our knowledge in the part taken by nitrogen in the plasma chemistry and in the surface chemical reactions, we present an exhaustive study of the influence of minute nitrogen addition in which the plasma chemistry is correlated to the diamond film properties during microwave plasma assisted Chemical Vapour Deposition of diamond. In addition, we present a series of plasma and surface chemical reactions that can account for the growth of diamond films under these particular synthesis conditions.

The issue of the existence of a suprahard phase beta-C₃N₄, with atomic structure analogous to beta-Si₃N₄, is at a stalemate. A continuous flux of papers, during the last decade, seemingly has not brought a resolution to this embarrassing situation, when the statement has been made about possessing powerful computational tools allowing us to predict "properties of substances even before we have created them." Among such properties is hardness and it was assumed that the covalent form of C₃N₄ predisposes it to be harder than diamond. This assumption contradicts what chemists have done since 1816, experimenting on carbon nitrides. Never a single hint has been given to the existence of a covalent, single bond C-N network nor was it documented in the last decade. To resolve this confusion we report on a real hard material silicon carbonitride with atomic structure of pseudo alpha-Si₃N₄. This phase possesses hardness comparable to cubic boron nitride and a band gap of 3.8 eV. The Si-N-C phase can be described tentatively as a solid solution of alpha-C₃N₄ and alpha-Si₃N₄. The present experiments indicate that only 6 atomic% of C can be incorporated in alpha-Si₃N₄. We suggest that first principle calculations should be undertaken to explain the limited solubility of carbon in a alpha-Si₃N₄ phase.

Having the confusion about superhard beta-C₃N₄ resolved, we then focus our study on the evaluation of physical properties of silicon carbonitride and its prospective applications. Among these properties are: hardness, oxidation resistance and wide band gap characteristics. Tests on cutting tool coatings and anticorrosive resistance coatings will be reported.

In recent years, surface pre-treatment methods have attracted considerable interest in order to produce high quality coatings for a wide range of industrial applications. Pre-treating the substrate prior to deposition can be closely related to the nucleation density as well as on the growth rate. Diamond coatings are being investigated for a wide range of uses in applications such as optics, microelectronics, bio-medical and cutting tools. Hot filament chemical vapour deposition (HFCVD) process has been used to deposit diamond on silicon, copper and AlN substrates.

In this paper we look at what effects surface pre-treatment methods such as DC-biasing and surface abrasion have on the nucleation density, growth rate and the morphology of the diamond films. Substrates were DC-biased at various negative voltages over a range of bias times. Furthermore, abrading the substrate surface with Al₂O₃ powder followed by diamond powder produces high quality crystalline films. The residues from the polishing process play an important role in the nucleation stage. However, the results suggest that silicon substrates allow better quality diamond to be deposited whereas diamond on copper and AlN is of a lower quality. Biasing has proven to be an effective method of producing optically smooth diamond films for potential optical applications.

Tungsten incorporated carbon films(W-C:H) were deposited using a new technique with two W screen grids incorporated inside an ECR-CVD chamber. This screen grid technique was previously developed for depositing diamond-like carbon(DLC) films under direct dc bias in an ECR-CVD system¹. In that study, different biases were applied to two stainless steel grids situated above the substrates to control the energy of the ions. It was found that at large bias, and in the presence of heavy ions(e.g. Ar+), sputtering of the grids can be significant resulting in the incorporation of metals in the growing films. Therefore, with appropriate metal grids, this technique can be used to incorporate different metals into DLC films. The biases at the grids and the distance of the grids from the substrates can be varied to control the amount of metal incorporated. The energy of the hydrocarbon ions can be separately controlled by biasing the substrate holder, resulting in a great variety of Me-C:H films with very different structural properties that can be deposited. This technique can be considered as plasma reactive sputtering.

In this work, W doped films were deposited by varying the flow ratio of CH4 to Ar from 0.15 to 0.5, at a constant pressure of 6mTorr. A bias of -330V was applied to the upper and lower grids with the substrate floating. The films were characterized in terms of their conductivity, IR absorption, atomic concentration(XPS and RBS), XRD and hardness. The optical gap, which is seldom reported for such Me-C:H films, was also investigated. The atomic concentration of W was found to range from 2% to 8%. The resistivity and the optical gap decreased by six orders of magnitude and 1.5eV respectively with increasing W incorporation. W is mainly bonded in the form of WO2 and WO3 as deduced from XPS. These results help to probe the structure of Me-C:H films, which have been shown to have good adhesion and excellent tribological properties that could be better than DLC².

¹S. F. Yoon, Rusli, J. Ahn, Q. Zhang, Y. S. Wu and H. Yang, Diamond and Related Mater., accepted for publication.
²C. P. Klages and R. Memming, Material Science Forum, 52&53 (1989) pg. 609

SiC is noted for its wide bandgap, high thermal conductivity, chemical inertness, and outstanding mechanical properties, making it an excellent material for high temperature MEMS. For surface micromachining usage, structural layers must be deposited on sacrificial layers, such as SiO₂, and high temperature insulators, such as Si₃N₄. Little is known about nucleation and growth of SiC on these substrates. This paper presents the results of a study of SiC films grown on SiO₂ and Si₃N₄ by APCVD.

An APCVD system was used to deposit polycrystalline SiC (poly-SiC) films using silane and propane as source gases, and hydrogen as a carrier gas. Phosphine was used as a doping gas. SiO₂- and Si₃N₄-coated Si wafers were used as substrates. The growth temperature was fixed at 1050⁰C and growth times of 30 sec and 30 min were used. The as-deposited films were characterized by stylus profilometry, SEM, XRD, TEM, and four-point probe.

SEM micrographs from poly-SiC films deposited for 30 sec on Si₃N₄ show that the nucleation density was about 1.7x10¹⁰ cm^-2, and the average grain size was about 50 nm. The nucleation density was only about 7x10⁸ cm^-2 on SiO₂, but the grain size ranged from 50 nm to 250 nm. For the 30-min. depositions, the average surface roughness of undoped poly-SiC films deposited on on Si₃N₄ and SiO₂ was 41Å and 175Å, respectively. XRD and TEM indicated that the films on both substrate materials were randomly oriented, with the grain size of SiC films grown on SiO₂ being larger.

For films deposited by APCVD, the observed differences may be due to a higher activation energy of SiC on SiO₂, which reduces the nucleation density of SiC. A similar phenomenon was found for the deposition of polycrystalline Si on SiO₂ by APCVD¹.

The extended paper will present details concerning SiC film growth, characterization, an explanation of the substrate effect, and implications of this effect to device fabrication.

¹ T. I. Kamins, and T. R. Case, Thin Solid films 16, pp.147-165, May, 1973.

Silicon Carbonitrides as analogues to Si₃N₄ and to the hypothetical C₃N₄ are attractive compounds for improving surface properties for many applications. For the synthesis of such ternary systems RF magnetron sputtering and ion implantation are the most promising techniques providing tunable stoichiometries. Therefore, Si-C-N thin films were reactively sputtered using ¹⁵N enriched N₂/Ar sputtering gas and co-sputter targets with different Si/C areas resulting in defined and reproducible Si/C ratios at constant nitrogen concentrations. In order to achieve crystalline layers, substrate temperatures up to 1000⁰C were applied during sputter deposition. Moreover, the variation of the substrate temperature was found to be the only way to influence the N content of the Si-C-N compounds. In contrast, surface modification by sequential high fluence implantation of C and N ions into silicon allows to define the concentration of all layer constituents. Both preparation techniques are discussed with respect to the attainability of the predefined stoichiometry range.

The chemical composition of the ternary systems were characterized by means of in-situ X-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy (AES). Therefore, a sputtering set-up has been directly coupled to an ESCALAB 5 electron spectrometer. In case of the buried implanted layers chemical binding states, however, only can be determined from XPS after sputter etching. Nearly undisturbed surface regions can be obtained using 400 eV Ar ions due to the negligible projected range with respect to the information depth of XPS. For quantification XPS and AES data were calibrated with absolute concentration values from n-RBS (non-Rutherford backscattering spectrometry). Furthermore, both preparation techniques have the advantage that ¹⁵N and ¹³C isotopes can be introduced into the layers enabling non-destructive nuclear reaction analysis (NRA) for depth profiling.