Status and Applications of Diamond and Diamond-Like Materials: An Emerging Technology (1990)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

2 GROWTH PROCESSES This chapter addresses deposition processes of superhard materials as defined in Chapter 1. Although it is the discovery of diamond growth from a low-pressure vapor by enhanced CVD techniques that is the focus of this report, the wider range of materials and methods is also described. The high-temperature, high-pressure (HPHT) techniques for growth of diamond and cubic EN, which are now over 30 years old, can be considered a mature technology. However, there are still some key scientific questions relating even to basic nucleation and growth and potential technological developments (large-area single crystals with controlled purity and perfection) that must still be explored. If homoepitaxy of diamond proves to be important, an in-country capability, or at least expertise, in the preparation of HPHT substrate may be critical. It is the control of nonequilibrium gas phase chemistry that allows for the growth of diamond at low pressures. Over the past 10 years workers have moved from a generally held belief that no practical process for low-pressure diamond growth was possible to the current realization that nearly any method that produces a ready supply of atomic hydrogen and some carbon-containing species works. Furthermore, the crystal quality and related morphology are remarkably similar for all the various processes and their variants, even though they range from plasma-enhanced processes to heated-filament-assisted nonplasma processes. During the course of the writing of this report even an atmospheric oxyacetylene torch has been reported to produce polycrystalline diamond films. Such observations suggest a common mechanism of nucleation and growth. However, there are differences, such as the growth rates reported varying over 3 orders of magnitude (0.1 to 300 ~m/hr). Also, scratching the substrate with diamond (or other abrasive) is essential for high nucleation densities (>l/pm2) in some processes but not others (e.g., thermal plasmas and atmospheric torch). To compare the various deposition techniques, much more needs to be known about the generation and transport of critical vapor species to the growing film surface as well as surface processes during both nucleation and growth. Key technological concerns include the deposition rate, the area that can be uniformly covered, the throwing power of the process, the range of substrates on which diamond can nucleate, the substrate temperature, and the energy efficiency of the process. Eventually a comparative parameter such as the mass deposited per unit time per unit power input could be used to compare each of the processes on an economic basis, but results are currently too sketchy. Also, each application will weight the different components of such a parameter differently. 19

20 Although enhanced CVD processes for diamond growth have created a number of new application opportunities, there are a number of limitations. These include (a) the high substrate temperature (~1000.C) for optimum crystal perfection; (b) the large grain size, which leads to optical and IR scattering, high friction, etc.; (c) contamination from etching and re-deposition of the substrate and deposition system components in contact with the atomic hydrogen; (~) the higher defect densities reported in all CVD diamond; (e) diamond's low coefficient of thermal expansion compared to most substrate materials; (f) lack of adhesion; (g) high stress; ant] (h) lack of large area heteroepitaxial substrates. Certainly the hope is that, through further development of current techniques plus discovery of even other deposition processes, these problems can be overcome. Of course, in formulating new applications for CVD diamond it is well to keep in mind basic limitations that are inherent for diamond materials in general. For example, the upper use temperature for conventional diamonds in air is no more than approximately 600 to 7000C, and a similar limitation is expected to apply to CVD-grown material. For many applications the substrate temperature is the limiting factor, and often temperatures above 500.C are unacceptable. A number of physical vapor deposition (PVD) techniques, involving ion bombardment, have been developed over the past 18 years to produce these materials at close to room temperature. However, the DLC and DLHC have lower crystallinity, higher residual stress, and lower hardness. To their credit, DLC and DLHC films are smooth, dense, and reasonably hard and have a variable range of other properties such as refractive inciex, optical band gap, and electrical resistivity. A relatively unexplored area of deposition Space" is the use of both bombardment and chemical vapor deposition processes simultaneously at temperatures not exceeding 300 to 500.C. Such materials may have properties intermediate between the diamond and DLC and DEHC materials. nossibIv leading to new applications. , , _ _ , _ _ ~ _ _ One final research and development (R&D) opportunity is the possibility of polycrystalline cubic BN films with grain size similar to that of diamond films. Whether analogous approaches might be successful remains to be seen. Microcrystalline cubic BN has been produced by ion-bombardment-mediated PVD processes (similar to DLC and DLHC), and this provides some hope. However, no successful attempts have been reported for crystalline cubic BN. GENERIC AND FUNDAMENTAL ISSUES The two principal superhard materials, diamond and cubic BN, can be synthesized by high-pressure, high-temperature (HPHT) processes in the stable thermodynamic region by static or dynamic methods in the case of diamond, or by low-pressure techniques in regions of thermodynamic metastability. The former is a mature manufacturing process and will be described first. About 90 percent of all industrial diamond used is synthesized by the HPHT process because it provides a reproducible product in terms of size, shape, and toughness (or friability). All cubic BN must be synthesized, since it does not occur naturally. Research and development effort in certain areas of this field in the United States appears to be very limited compared to activities in some other countries. The vapor deposition of diamond now allows for preparation in many new forms--thin films over large areas, chemically bonded to a wide range of surfaces, on complex shaped surfaces, freestanding forms, powders, etc. It appears that cubic BN films with large crystal sizes (>1 '`m) are possible, which would open up even further new materials opportunities. Several years ago it seemed as if polycrystalline films of diamond at practical growth rates could not be prepared at all, and now it appears that just about every technique that can provide an intense source of atomic hydrogen during growth works. This creates the frustrating situation of having

21 too many choices. Ultimately it may turn out that there is no one best deposition method but rather that certain ones will be best for certain film forms or applications. This section deals mainly with diamond and diamond-like materials--i.e., amorphous and nanocrystalline carbon both with and without hydrogen and with high gram-atom number densities (fin > 0.2 g-atom/cm3) (Angus et al., 196--since they are the only superhard materials for which a significant body of literature exists. The limiting value of hardness of DLC and DEHC films is still not known. Cubic EN has been prepared as a microcrystalline coating, but to date no films with crystal sizes above 200 A have been reported, and the chemical and structural characterization of these materials is sketchy. The recent disclosure of a suboxide of boron of nominal composition of B22O (Badzian, 1988) as a material that can scratch diamond is also intriguing. This composition also can be prepared by reactive sputtering to have properties similar to that of the bulk material. However, such reports must be considered with caution, since even SiC has been known to scratch diamond. Superhard materials in the C-N system are yet to be explored. If cubic BN can be successfully vapor-deposited as large crystals and polycrystalline films, it is expected that research on this material will increase dramatically, just as when diamonds were successfully deposited from the vapor and the result recognized. Deposition approaches analogous to those for diamond are under way in the United States and elsewhere. In this report the deposition of superhard materials is categorized in a standard manner as chemical vapor deposition (CVD) and physical vapor deposition (PVD) processes. In general, CVD methods use gas-phase chemistry to control both the retention of the metastable phase when condensed on a surface and the details of the composition, impurities, defects, etc. In PVD it is usually ion bombardment processes that control the characteristics and thus the properties of the resulting metastable films. As mentioned earlier, it now seems that just about any process that efficiently produces atomic hydrogen and carbon-containing species can yield diamond films over some range of conditions. In some deposition processes both physical and chemical processes take place simultaneously, but such distinctions will not be made specifically. Very often the relative roles of the two different general processes are not well understood. HIGH- PRESSURE, HIGH-TEMPERATURE PROCESSES The first reproducible industrial process for making diamond was announced in 1955, and the synthesis of cubic BN was accomplished by 1957. Both phases were precipitated from solution at pressures >45 kb and temperatures >1100oC, where they are the thermodynamically stable phase. The unique chemistry for each of the two superhard materials is considered separately. A description is given of the critical mechanical equipment that makes the synthesis possible. Materials Growth Processes Direct Conversion It is possible to make both diamond and cubic BN by direct conversion of graphite and hexagonal BN, respectively. This can be done both by static and dynamic processes, but for diamond the pressure range (~120 kb) and temperature are prohibitively high and not economically viable because of the limited life of carbide dies used in static processes. However, under dynamic conditions (shock), the conversion is practical, and the DuPont process provides a marketable product. Because of the short reaction time, the crystals range in size from

22 submicrometer to several micrometers, and they are used primarily for polishing-grade material or for starting material for sintering. For cubic BN the pressure-temperature (P-T) conditions are less severe than for diamond, and a directly converter! product is available. The wurtzite form of BN, which is also hard, is made by shock synthesis and is used primarily for sintering polycrystalline aggregates. Static or Indirect Process The difficulties of direct conversion led to the development of processes to lower the pressure and temperature. This is accomplished by use of a solvent-catalyst system that permits growth to occur at conditions much nearer the diamond-graphite equilibrium line but at lower temperatures. The static synthesis process consists of first raising the pressure to the desired value and then raising the temperature of the cell by resistive heating. The P-T conditions for growth are held for the desired time, and the power to the heater is decreased to zero. The pressure is then released and the cell removed for separation of the product. The P-T conditions for both diamond and cubic BN are attained as described above, but the chemistry for the synthesis of the two materials is different. For diamond, the compositions are generally mixtures of Group VIII metals with graphite. Usually the metals iron, cobalt, or nickel are major components of the alloy with carbon, with manganese, aluminum, and boron the secondary components. The physical chemistry of these systems can be described in terms of a phase diagram of the selected metal alloy-carbon system (a T-X diagram) at pressures where diamond is a stable phase in equilibrium with a metal-rich liquid. In this system there exists a eutectic at lower P-T conditions than required for direct conversion. In these terms, the growth is from solution, and the driving force for nucleation is the difference in solubility between that of diamond and graphite under the HPHT conditions. Nucleation and growth begin as soon as the eutectic temperature of the system is exceeded. Some believe the process is better described as "catalytic," and the compromise term "solvent-catalyst" is frequently seen in the literature. There are still many things not understood about the nature of the solution and the transport in it under HPHT conditions. The first requirement is a mechanical system that will reproducibly attain 50-65 kb and 1200-17000C simultaneously. Since the material is produced in batches rather than continuously, it is an advantage to have as large a working volume as possible. There are several types of apparatus (e.g., uniaxial belts or girdles, tetrahedral, octahedral, and variations on the cube press, piston-cylinder) and cells (e.g., low and high aspect ratio). During the more than 30 years of experience in synthesizing diamonds, there has been a convergence in the use of uniaxial belts for manufacturing. The other methods have advantages for R&D purposes, but scale-up has generally favored the belt. The belt consists of a central die, generally of cemented tungsten carbide, surrounded by a series of "belted" steel rings to keep the carbide in compression. Two symmetrically opposed tapered anvils are pressed toward the die hydraulically and are separated from carbide-carbide contact by a set of gaskets made from natural or synthetic pyrophyllite or soapstone-like material (Figure 2-1~. Carbide and pyrophyllite are the key materials used in this and most other high-pressure apparatus. The gasket provides mechanical, thermal, and electrical insulation and contains the cell and its reactants. The anvil-die-anvil combination is adapted to commercially available presses. The cost of setting up a large apparatus is estimated to be in the range of at least

23 $250,000 to $1,000,000. This equipment must be capable of repeated excursions to and from HPHT conditions. Die breakage is related to fatigue and work-hardenir~g of the carbide dies and can be an important manufacturing cost factor. The supply and cost of carbide dies (especially for large volume) are probably key factors in limiting new entrants to the manufacturing of these materials. The attractiveness and relative simplicity of CVD as alternative processes will be obvious, as described later in this report. 1~j~ ~~, FIGURE 2-1 The Hall "belts apparatus. Shaded areas are cemented carbide, black is gasket material. There are many cell designs to cover the needed variations, but common to all is internal heating by passing a low-voltage high current through the cell. Direct heating usually means the current passes directly through the reactants (e.g., a mixture of iron and graphite for diamond synthesis). Indirect heating is used with nonconducting reactants (e.g., BN plus Li3N mixture for cubic BN) by packing them in a conducting cylindrical sleeve or around a conducting core (graphite or metal). Cell variations result from the need to separate reactants from containers and to control the heat distribution.

24 Cubic BN Chemistry The solvent-catalysts for cubic BN are based primarily on alkali or alkaline earth metals and their nitrides. The phase diagram for hexagonal BN with I~i3BN2 is analogous to that of graphite with a metal. Cubic BN can also be synthesized from Group VITI metals with a bit of aluminum added ant! from other systems, but the water and acid solubility of the alkali anti alkaline earth metals and nitrides favors easy removal of the product from these reactants. Control of Crystal Size The static process permits control of grain size simply by controlling the time the system remains at the desirer! P-T conditions. Carbon is transported from the graphite source through a thin metal film to the growing crystal. Most abrasive grain found in the marketplace is less than 1 mm in maximum dimension because the process becomes noneconomical if the growth time is too long. The practice has been to use sintered diamond when sizes greater than 1 mm are needed. The special case of large gem-quality crystals is considered below. Growth rates can be very high, but 1 mm per hour is commonly attained. The nucleation rate can be controlled by varying pressure conditions above the eutectic temperature for both diamond and cubic BN. Nucleation can also be controlled by seeding with small diamond or cubic BN crystals. Growth of Large Crystals In 1970 a thermal gradient process for growing large crystals (~1 mm) in a controlled manner was developed (Figure 2-2~. Using this technique, colorless, blue (B-doped), and yellow (N-doped) diamond! crystals of gem quality were grown to about 6 mm in maximum dimension and about 1.3 carats weight. The crystal size at that time was primarily limited by cell size. These crystals were grown one at a time for 100 hours or more. Since that time Sumitomo and DeBeers have both scaled up the process to grow larger crystals and many at a time. Yellow crystals up to 7 mm in size are available (from Sumitomo Electric) for tools, windows, scalpels, anvils, etc. DeBeers has reported a crystal greater than I] carats with a maximum dimension of 17 mm. The process requires large expensive equipment, which could be producing abrasive grain, so that some manufacturers have been reluctant to get involved. To improve the yield, many crystals must be grown at the same time by optimizing the use of the thermal gradient vertically in the cell of a belt apparatus. The gradient can be established in the cell with a sleeve heater, but then alloys with different eutectic temperatures must be used to allow growth in several layers in the cell. This is fairly simple for N-doped crystals, but there are fewer degrees of freedom for the aluminum alloys used to eliminate nitrogen and produce colorless and blue crystals. New alloy design is needed. Cubic BN crystals to about 3 mm maximum dimension have been grown by thermal gradient methods, but good-quality untwinned crystals are rare. Sumitomo has taken up this challenge, and this situation can be expected to change. NIRIM (Mishima, et al. 1987) has grown cubic BN pen junctions. The production of large crystals is of considerable current interest for electronic, optical, and dielectric applications and should be re-evaluated in the United States, especially in light of the obvious commitment by Japanese and South African diamond manufacturers. However, if the CVD techniques continue to show promise, the difficulty of growth at high pressures may be bypassed as epitaxial techniques (perhaps on SiC crystals) are developed.

25 METal BATH - - - HEATER _SMALl DIAMONDS SEED DIAMOND FIGURE 2-2 Cell design for growth of large crystals of diamond or cubic BN using a temperature gradient between carbon source and the seed crystal. For cubic BN, the transporting medium can also be an ionic melt. Control of Crystal Shape Diamond The octahedron is the predominant growth form for natural diamond; the (100) and (~10) faces are also common. It was found that the relative development of (111) and (100) could be controlled by the temperature of growth at a constant pressure. In a P-T plot of the diamond stability region (Figure 2-3), the cube (100) face predominates at the lower temperatures (near the metal-carbon eutectic) and the octahedron (~) at higher temperatures, near the diamond-graphite equilibrium curve. (This is opposite to the morphology changes in CVD diamond growth.) In between these extremes the cubo-octahedron predominates and, because this shape holds better in a binder, it is preferred. The morphology-temperature relationship is serendipitous and is essentially limited to yellow Type Ib diamond crystals, which predominate in synthesized abrasive grain. That the morphology is more than temperature-dependent is evident from results with the addition of aluminum to getter nitrogen, where only octahedra form is the fast growth mode that is characteristic of processes for abrasive grain. Cubic BN Similar morphological changes as a function of temperature are seen with respect to (111) and (100) faces, but there are some basic differences because of the noncentrosymmetric nature of cubic BN. The predominant face is (111) of either the (+) or (-) form, so the morphology is

26 ~ /l ~/~ a: - J on on 1,'', 111/~ 4~ METAL-CARBON I EUTECTIC GRAPHITE TEMPERATURE FIGURE 2-3 Morphology of diamond from metal solutions as a function of pressure and temperature. Cubes dominate at lower temperatures, octahedra at higher temperatures, and mixtures In between. usually a tetrahedron with truncated tips. However, it is possible with equal development of these forms for an octahedron to grow. Another common morphological form of cubic EN is that of hexagonal plates. Control of Chemistry. Doping, and Impurities Diamond Natural diamonds have been classified into four main types on the basis of their absorption spectra, which are dependent on the chemistry and structure of defects. Figure 2-4 shows a chemical classification of diamond based on a phase diagram. Type lA is shown as a two-phase C-N material (platelet diamond--90 percent of natural diamond and essentially unavailable from synthesis). Type IB is diamond with nitrogen in solid solution at lattice sites. Type IIA diamond is essentially pure carbon and exhibits the highest thermal conductivity. Type lIB is the p-doped semiconducting diamond with boron in lattice sites. Although many impurity elements are reported to be present in natural diamond, some are present in second-phase inclusions. The only elements that appear to be truly in substitutional sites are boron, nitrogen, and possibly beryllium. In synthesized materials the impurities are metal and graphite as inclusions and boron and nitrogen. Usually no effort is made to eliminate nitrogen when loading a cell, so all greenish-yellow crystals contain about 1000 or more ppm. If aluminum is added to remove nitrogen, a colorless crystal is obtained. If boron is added along with aluminum, the crystal is blue and a p-type semiconductor. A crystal with 10 ppm boron is very dark and opaque. Both nitrogen and boron can be nonuniformly dispersed in a crystal, usually along

27 11 A COLORLESS ~B — ~ / BLUE t HOPE ) \ S8\ YELLO'-\ ' — GREEN HAN" CAM, ABRASIVES GRAIN/ IA HOST NATURAL OtAVONDS COLOR LESS—YELLOW B— IN \ FIGURE 2-4 Schematic representation of the four principal types of diamond in terms of the systems B-C and C-N. Most natural diamonds (IA) are two-phase mixtures of platelets precipitated in the host diamond. certain directions, and differential wear and electrical properties can be associated with the distribution. N-type doping of diamond under equilibrium conditions has not been achieved. Cubic BN Cubic BN can be doped to be either n- or p-type under equilibrium conditions, and a pen Unction has been reported bv synthesis at HPHT (Mishima. et al.. 19871. Most unintentionally doped cubic EN Is amber, a color that is thought by some to be the result or excess boron. it behaves like an extrinsic semiconductor, with a decrease in resistivity on heating, as well as exhibiting a reversible thermochromism. Up to 10 ppm lithium (from the solvent-catalysts) has been measured in cubic BN, but its location in the lattice is not known. Cubic BN may also have inclusions of hexagonal BN and borides. Some products contain carbon as an inclusion, although solid solutions between diamond and cubic BN are reported. A ternary phase, BCN, is also said to exist. Pressure-S'ntered Material Both diamond anti cubic BN powder can be pressure-sintered into polycrystalline aggregates in their respective stability regions at HPHT conditions. Tools for cutting, drilling, sawing, and wire drawing have been common in the marketplace since about 1970. Dimensions of the diamond layer on a carbide substrate are at least as large as 50 mm diameter and ~ mm thick at this writing. Such "sandwich" materials can be used for wear plates but are more often cut into rectangular, square, or pie-shaped pieces for tools. Large wire die blanks, which consist of a ring of carbide around a cylinder of diamond at least as large as 14 x 14 mm, are routine trade items.

28 Although both diamond and cubic BN can be directly sintered, most of the compacts are made by a liquid-phase sintering process. In the simplest case a liquid metal (cobalt-rich) from the carbide infiltrates a mass of diamond grains in situ in the press. The liquid dissolves and reprecipitates diamond at grain contacts to make diamond-diamond bonds. The final product consists of two continuous phases: diamond and about 6 volume-percent metal. The latter can be removed by etching, leaving a diamond! compact of high strength and greater thermal stability than with the metal in place. However, it is still a brittle material, and thin layers of either diamond or cubic BN cannot be expected to withstand high stresses without the carbide support. Small pieces free of carbide can replace many applications of larger diamond if properly supported in a metal binder using the same technology that is well-tested for these materials. The random array of grains in a polycrystalline aggregate is a better material than a single crystal in two respects: elimination of both anisotropic wear and cleavage. Thus, wire-drawing dies of sintered diamond are far superior to a single-crystal die in terms of die life. There is an effort to use sintered diamond as anvils in high-pressure apparatus, with the hope of attaining higher pressures and longer life than with carbide alone. As will be cliscussed below, polycrystalline CVD diamond! films form a random, though under certain deposition conditions an ordered, array of grains that "sinter" due to the normal growth and undergrowth of individual polycrystalline grains. Cubic BN can be sintered by analogous techniques described for diamond with some chemical modifications that are in general proprietary. The cubic BN tools are especially useful for turning ferrous-based material. The direct sintering of cubic BN from hexagonal BN is within the capabilities of static apparatus, and such materials are available. If the grain size of this material can be kept small enough, precision turning tools may be developed that will eliminate the need for single crystals of this rather friable material. The directly sintered material also has potential for heat sinks. Polycrystalline CVD diamond is a potentially very attractive replacement as a coating or as grains because it is 100 percent diamond, with very clean grain boundaries and without porosity and second phases to disrupt thermal stability. The possibility of making unusual shapes and large sizes directly, if successful, will have considerable negative impact on the high- pressure process remaining economically competitive. One of the significant problems to be solved, however, is that of adhesion. There is room for improvement in the bonding of CVD diamond to any material. CHEMICAL VAPOR DEPOSITION PROCESSES It was the recognition of the role of atomic hydrogen in the vapor deposition approach that led Russian workers to the first successful growth of diamond films at commercially practical deposition rates (>1 '~m/hr). Since the early 1980s a wide variety of energetically assisted CVD methods have been employed. (Before that it seemed like nothing would work and now it seems like everything works--at least any process that efficiently produces atomic hydrogen.) These are subdivided into plasma-enhanced processes and nonplasma (primarily thermally assisted) processes. Distinctions between the processes become blurred in some cases, such as in thermal plasmas. Plasma-Enhanced Processes A plasma is an efficient way to dissociate gas molecules to produce nonequilibrium concentrations of gas-phase species, such as the high concentrations of atomic hydrogen needed

29 to make diamond. Plasmas can be generated by a number of energy sources (dc, rf, and microwave electric fields and flames), and the plasmas can be either cold or hot. The main techniques with their variations are given in Table 2-1, and schematics of typical experimental setups are shown in Figure 2-5. Also given are representative references. Although these preparation methods are not discussed in detail, a few general comments can be made. The most extensively studied process is microwave plasma-enhanced chemical vapor deposition (MPECVD). This trend has only increased over the past several years, especially in Japan (Figure 2-6~. The apparent reasons are (a) the stability and reproducibility of microwave cold plasmas; (b) their energy efficiency (high plasma density, low electron temperatures, low sheath potentials); (c) the increased availability of 1-2 kW microwave (wave-guided) power supplies and applicators; and (~) the potential to scale the process to larger substrates (4-5 in.~. However, the rf and dc plasma methods produce diamond films of comparable quality and, under certain conditions (e.g., substrate biasing), at much higher rates, and they are being used in commercial processes in the United States (Crystallume and Air Products). Although strides to scale the MPECVD process have been made, it seems that rf and dc plasmas are easier to scale. It is not as clear which of these processes would have an advantage in coating nonplanar substrates or which is amenable to large batch coating such as on vertically stacked wire mesh trays for cutting tool coatings. Also, dc plasma processes at present seem to have an advantage of potentially higher deposition rates (AX). The thermal plasma approach generates the necessary temperature to dissociate H2 directly in the plasma Although the plasma nozzle is nearby, nothing is known at present as to temperatures and carbide formation at the metal nozzle or, more generally, the stability of the process. To date this process has only been used successfully in Japan by NIRIM, Fujitsu and Asahi Diamond; U.S. companies and universities are pursuing this approach. It is known that the rates can be extremely high (200-500 um/hr) but the deposition area is small (<1 cam. The deposition "on-time" is thermally limited, and the uniformity and reproducibility are poor. However, if controlled, this approach could be very important. Of the remaining plasma-enhanced techniques listed in Table 2-1, the atmospheric torch is unique and potentially the most important. Although not strictly a plasma, an atmospheric flame is simple in design, cost, and use. Hydrogen, oxygen, and hydrocarbon gases (e.g., acetylene) are simply burned, and diamond forms in the reducing part of the flame. Hirose's results in atmospheric torch deposition (Hirose and Mitsuizumi, 1988) have now been repeated in several laboratories around the world (Naval Research Laboratory, Pennsylvania State University, and Sumitomo Electric), and it is expected that other groups will be pursuing this technique. The trade-offs among plasma stability, deposition rate, coating area, film quality (as defined by either scientific or technological criteria) for each of the plasma processes have yet to be determined. It is likely that there will be no single answer, solution, or technique; rather, each technique will fine! its niche (albeit, some small niches). Nonplasma Processes Besides the plasma-enhancement approach to vapor deposition of diamond, Derjaguin and Fedoseev (1977) recognized that high-temperature and catalytic processes could also produce a "superequilibrium" of atomic hydrogen, which appeared to be the necessary ingredient for a single-step deposition process.

30 TABLE 2-1 Plasma-Enhanced Chemical Vapor Deposition (PECVD) Processes for Diamond Film Growth Technique Deposition Rate Selected (,~m/hr) References Microwave-PECVD 0. 1-5 1,2,3 - magnetic field at high pressure (>1 torr) - ECR (<0.1 torr) rf-PECVD ? 4,5 - inductively coupled - capacitively coupled dc-PECVD - capacitively coupled . · . - hollow cathode discharge Thermal plasmas - dc thermal plasma torch rf thermal plasma torch atmospheric flame torch - vacuum arc 0.01-100 6,7,8 10-300 9,10,1 1 KeY to References: 1. Kamo, M. Y. Sato, S. Matsumoto, and N. Setaka. 1983. Diamon synthesis from gas phase in microwave plasma. J. Cryst. Growth, Vol. 62, p. 642. 2. Saito, S., S. Matsuda, and S. Nogita. 1986. I. Mat. Sci. Lett., Vol. 5,p. 565. 3. Kawarada, H., K. S. Mar, and A. Hiraki. 1987. Larbe area chemical vapour depositor of diamond particles and films using magneto-microwave plasma. Jpn. ]. Appl. Phys., Vol. 26, p. L1032. 4. Matsumoto, Se 1985. Chemical vapour deposition of diamond in rf glow discharge. I. Mat. Sci. Lett., Vol. 4, p. 600. 5. Marria, R., L. Stobierski, and R. Pampuch. 1981. Diamond syhthesis in cool plasma. Cryst. Res. Tech., Vol. 16, p. 785. 6. Suzuki, K., H. Yasuda, and T. Inuzuka. 1987. Growth of diamond thin films by dc plasma chemical vapor deposition. Appl. Phys. Lett., Vol. 50, no. 12, p. 728. 7. Pinneo, J. M. 1987. Diamond Technology Initiative Workshop, paper 4. Sponsored by the Office of Naval Research at M.I.T. Lincoln Laboratories, Boston, Massachusetts (February 2~. 8. Singh, B., O. R. Mesker, A. W. Levine, and Y. Arie. 1988. Diamond synthesis by hollow cathode plasma assisted chemical vapor deposition. Proc. SPIE, Vol. 877, p. 70. 9. Koshino, N., K. Kurihara, M. Kawarad~a, and K. Sasaki. 1988. 1988 MRS Spring Meeting, Symposium D Extended Abstracts, p. 95. M. Geis, G. H. lohnson, and A. R. Badzian, eds. 10. Matsumoto, S., M. Hino, and T. Kobayashi. 1987. Synthesis of diamond films in a FR induction thermal plasma. Appl. Phys. Lett., Vol. 51, p. 737. 11. Hirose, Y., and Kondo. Spring 1988. APL Meeting, Abstracts, p. 434.

31 H3 + CH3 1 - . last" .t .b~ Wave guide _ 474 Magnetron bum, mu Substrate IT Plunger Vacuum pump (a) Substrate Working Coil To Manometer Feed gas ~ ~ _~ Am' ', - To Pump Water RF generator |13.56 MHz| Hi Quartz tube ~ 1 4' (b) Water _- ~ Substrate ~ cl ~ ~ (c) Gas Inlet Carrier + Reactant Gas Plasma gas (P)~ 1~1 Sheath gas (S)_ ~1 1 —' -'it 11 ~ Substrate - , ~ _ - /r- Substrate holder I (d) 1 . 1 O RF O Workcoil O -_ Water Gases ~- T:~D 11 1; Plasma jet ~.''1'. ~ . ~ ~ : : C 1` r Water (e) FIGURE 2-5 Schematic diagrams of various plasma-enhanced CVD techniques for diamond growth: (a) microwave PECVD (Kamo et al., 1983~; (b) rf PECVD (Matsumoto, 1985~; (c) do PECVD (Suzuki et al., 1987~; (~) rf thermal plasma CVD (Matsumoto, 1985~; (e) dc thermal plasma CVD (Koshino et al., 1988~.

32 20 In c) En m 10 lo o LL m ~7 o In .. 1186 2/86 Methods HF/EACVD _ ~3 RF _ ~ MW Lo DC UV/Laser Other Theory 1/87 2/87 FIGURE 2-6 Statistics of abstracts published at the meetings of the Japanese Applied Physics Society on different preparation methods for diamond thin films (Bachmann et al., 1988~. \ Other than a patent by Vickery (1973) and the recent paper by Badzian et al. (1988)' the idea and use of catalytic processes in diamond vapor growth has not been explored. These approaches may prove to be important from an energy (economy) viewpoint, and thus more work is justified. However, chemistry control is always difficult to unravel, and adding catalytic mechanisms to an already complex and poorly understood process involving atomic hydrogen and hydrocarbon species will require considerably more time and effort. Thermally enhanced processes would appear to be the easiest to understand experimentally (Figure 2-7) and conceptually. The optimum conditions for diamond growth involve, among other factors, a substrate temperature of about 1000oC and a high percentage of atomic hydrogen. Since this latter factor cannot be achieved at the 1000.C substrate surface, a secondary filament in close proximity to the substrate is an option. In fact, this approach is the simplest and most effective to date. The nonplasma methods are listed in Table 2-2. The heated filament approaches can have some severe limitations due to hydrogen incorporation and carbide formation of the refractory metal filament (usually tungsten or tantalum). The filament may form the carbide before diamond growth starts (delay time up to hours) and then is brittle and sags or distorts during deposition. This leads to deposition conditions that are nonstable and difficult to reproduce. The small-bore-tube approach seems to minimize these effects, and the gas flow both over the heated filament and at the substrate can be better controlled. The filament temperature is generally 2000 to 23000C, and thermochemical calculations show that between 5 and 10 percent of H2 is dissociated to atomic hydrogen at those temperatures. HYbrid Processes The combination of plasma formation and particle bombardment (e.g., due to self-biases and imposed biases at the growing film) will need to be considered. Although it is considered separately in the next section of this report, it may very well be the blending of plasma chemistry with ion bombardment that produces the most practical coatings--superhard, tough, and adherent to metal and ceramic substrates.

33 Furnace ~ 6 :: m tee_ i Feed ~ FEW filament Substrate Silica holder ~ Thermocouple To pump (a) 1 mm tantalum I. . wire col substrate holder 1/8 inch O.D. tantalum tubing Gas Feed Pipe Graphite substrate heating _ elements Alumina insulator PtlPt-Rh thermocouple IBM FIGURE 2-7 Schematic diagrams of two different nonplasma approaches for diamond growth: (a) heated-filament-assisted CVD (Matsumoto et al., 1982 (a, b); (b) heated-tube-assisted CVD (Yarbrough and Roy, lamb. TABLE 2-2 Nonplasma Deposition Processes for Diamond Film Growth Deposition Rate Techniques (,~m/hr) References Heated-Filament CVD 0.1-20 1,2 - wound filament - small-bore tube - substrate biasing Laser-enhanced CVD 3 Kerr to References: 1. Matsumoto, S., Y. Sato, M. Kamo, and N. Setaka. 1982a. Vapor deposition of diamond particles from methane. Jpn. J. Appl. Phys., Vol. 21, p. L183.1. Matsumoto, S., Y. Sato, M. Kamo, J. Tanaka, and N. Setaka. 1982b. Chemical vapor deposition of diamond from methane--hydrogen gas. Proc. 7th Int. Conf. Vac. Metallurgy, Tokyo, p. 386. 2. Yarbrough, W. A., and R. Roy. 1988. Spring MRS Meeting Extended Abstracts, Symposium D, p. 33. M. Gels, G. H. Johnson, and A. R. Badzian, eds. 3. Kitahama, K., K. Hirata, H. Nakamatsu, S. Kawai, N. Fujimori, T. Imai, H. Yoshino, and A. Doi. 1986. Synthesis of diamond by laser-induced chemical vapor deposition. Appl. Phys. Lett., Vol. 49, p. 634.

34 VaDor-Liguid-Solid (VLS) Growth The growth of diamond whiskers has been reported (Derjaguin et al., 1968), but no significant further work appears to have been done in the intervening 20 years. It was claimed that the diamond whisker growth proceeded by the VLS process, but it is not known how the conditions differed from those used to grow graphite whiskers. In the latter case, iron beads approximately 100 to 200 A in diameter provide transport of carbon from the vapor to the growing tip of the whisker under conditions not too far removed from CVD diamoncl growth. Careful experiments are needed to determine what factors encourage diamond whisker growth and the possible role of hydrogen in the process. Critical CVD Growth Parameters Diamond crystals and films with the highest perfection have been grown by controlling the gas-phase chemistry. Although many of the details of the fundamental nucleation and growth processes are still unknown (see later section), the critical deposition parameters of nucleation and growth have been explored extensively over the past several years. Interestingly, with few exceptions, the critical parameters are similar for all of the various techniques described covering a wide range of gas activation. Thus, unless noted, the general parameters apply to all of these diamond deposition approaches. Substrates and Their Preparation Diamond has been nucleated and grown on a wide range of substrates, including metals, semiconductors, graphite, insulators, and even fused silica glass. Although the present evidence is inconclusive, stable carbides at the diamond film-nondiamond substrate interface appear to be important. The origin of the carbide may be due to direct chemical reaction between the activated gaseous species intentionally introduced into the system and the metal element in the substrate or to gas species originating from atomic hydrogen etching of the deposition reactor wall or the substrate. There is usually a delay time between the start of the deposition to the point where individual crystallites are seen and the time when they grow together to form a continuous film. The most effective methods for increasing the nucleation density and decreasing the delay time for nucleation and growth have been to either scratch or seed the surface with diamond powder or add metal impurities (Si, Ti, B) in the gas phase. In the first case, the diamond crystals littered on the substrate surface are, of course, excellent nucleation sites for immediate diamond growth. Epitaxial growth of diamond on diamond has been demonstrated by Spitsyn and coworkers (1981), and Gels (1988~. Other nondiamond powders (e.g., cubic BN, SiC, TaC) have been used to "seed" the surface (Bachmann et al., 1988~. Substrate Temperature For the diamond films with the highest level of crystal perfection, the substrate temperature is usually in the range of 900 to 1000.C. This result seems to be independent of deposition technique (either PECVD or TECVD), and in all cases the film morphology consists of approximately 1- to 5-m-size crystallites in films about 5 am thick. The temperature obviously limits substrates to rather refractory materials.

35 Diamond films with crystallite sizes of approximately 2000 A have been reported at substrate temperatures around 6500C while 100- to 200-A-size crystallites occur for substrate temperatures less than 600 C but greater than 4500C. The optimum temperature for crystal perfection is similar to that for the maximum in film growth rate. Gases and Gas Composition A large number of hydrocarbon gases have been used to deposit diamond. Although there are no reported significant differences in the resulting films, methane has been the gas of choice. In nearly all reported cases, large hydrogen gas dilution (e.g., about 1 percent CH4 in H2) is necessary with a trade-off between high rates (at about 5 percent CH4) and crystal size and perfection (at about 0.2 to 0.5 percent CHIP. Recently diamond films of apparent comparable quality have been grown from CO. For CO, dilution with hydrogen appears radically different. Good-quality diamond films are obtained at 50 percent CO concentrations and even greater. Gas Flow Rate and Pressure The largest diamond crystals are obtained at pressures from 20 to 100 torr in both the plasma-assisted and thermally-assisted processes. In the case of PECVD, at low gas pressures, SiC formation increases while the plasma starts to become unstable at pressures above 100 torn In the optimum range (especially at 50 to 100 torr) the plasma volume is small enough so it is not in contact with the reactor tube wall (typically 38- to 50-mm-<iiameter tubes for MPECVD). The film quality and deposition rate are relatively insensitive to the flow rate, which is typically 50 to 1000 sccm. Additions to Gas Very little systematic work has been reported on either inert or reactive gas additions to the Hz-hydrocarbon gas mixtures. More work is clearly needed. Position of Substrate In all reports on PECVD diamond the substrates are immersed in the plasma or separated only by the dark space sheath in do and rf plasmas. Remote plasma deposition in MPECVD has not been reported as yet, but the possibility cannot be discounted. For thermally assisted processes the substrate generally is placed about 1 cm from the heated filament (or tube or thermal plasma nozzle). Remote deposition at distances greater than several centimeters has not been reported. Whether this is related to the lifetime of active species (atomic hydrogen and hydrogen-containing radicals, ions, or molecules) is yet to be explored. Parameters Specific to Individual Processes In the case of heated-filament CVD the filament temperature is typically 2000 to 2300 C. Although the gas flow over and near the filament should be critical, this has not been controlled to date. For thermal plasmas the results are sketchy, but gas temperatures of 3000 to 10,000 K are reported. Although the high deposition rates and excellent quality are encouraging, the apparent lack of film reproducibility (N. Koshino, personal communication) may simply reflect the lack of control of the process at present.

36 Size of Substrates To make diamond deposition commercially feasible, high deposition efficiencies are needed. One approach is to deposit over large areas, and for many applications (e.g., optical windows, semiconductor wafers, large numbers of cutting tools) this is essential. For substrate diameter greater than 5 in., uniform substrate heating at approximately 1000oC for optimum diamond growth will become a major problem. Although microwave plasmas have the advantage of stability and efficiency of production of plasma species, the geometrical constraints of coupling microwave energy to a plasma volume are significant. However, work in this direction is ongoing in both the United States and Japan. The ability to scale up dc and rf deposition processes is easier than with microwave; they seem less restrictive and have been used successfully in the a-Si:H solar cell field, for example. Scaling up of the heated-filament CAD method is a matter of constructing an array of heated wires and tubes, with the main drawback being the uniformity of deposition thickness and properties. Electric and Magnetic Field Effects Biasing of the substrates during dc plasma deposition results in diamond formation when positively biased and in carbide formation when negatively biased. The mechanism is not known not even whether it involves only electrons or both electrons and ions (positive and/or negative). If high-resistivity diamond films are eventually made, then dc plasmas will not work due to space charge buildup. If a positive bias on the substrate is essential, then this could be a limitation. For microwave plasmas the electron cyclotron resonance (ECR) condition (875 gauss) for energy coupling has been used. At high pressures (~1 torr) ECR does not occur due to gas phase collisions, but positive results have been claimed regarding the ability to increase the extent of the plasma so as to scale to larger substrates. At the lower pressures (<0.1 torr) where the ECR condition is valid, diamond growth has been reported at low temperatures (6500C). This area of research of high plasma density deserves further investigation. To the extent that electric and magnetic fields allow for the acceleration, deceleration, and deflection of active plasma species, these process controls will allow merging between chemistry-controlled and bombardment-controlled growth of diamond. It is expected that this area will prove to be scientifically and technologically important. Fundamental Issues in CVD Processes There have been few reported studies that directly address the fundamental issues of nucleation and growth of vapor-deposited diamond films and the resulting film's structural and chemical defects. This section, therefore, deals more with speculations and attempts to identify critical areas of research. As is to be expected for an emerging material technology, knowledge of how to deposit films far exceeds the understanding of fundamental mechanisms and kinetics of nucleation and growth. Nucleation Since diamond is a metastable solid, the issue of nucleation involves not only that of diamond but also graphite and the hexagonal polytype of diamond, lonsdaleite. Furthermore, planar defects associated with the {111) planes in diamond--stacking faults, twinning, and introduction of planar graphite planes between the (111) planes of diamond--have all been

37 observed (Badzian et al., 1988; Davis et al., 1987~. The free-energy differences of these carbon phases are relatively small, so that kinetic factors, size effects' surface reconstructions, etc., during nucleation could easily be more important than thermodynamic factors. The widely varying morphological forms reported (Matsumoto and Matsui, 1983) for vapor-deposited diamond- -cube, cuboctahedra, twinned cuboctaheda, pseudo- icosaheclra, decahedra!, Wulff polyhedra--are not all fount! in natural or HPHT synthetic diamond crystals. There appear to be, however, some similarities between HPHT and vapor-deposited diamond. In both instances, stable carbides may provide favorable centers for cliamond nucleation and subsequent growth. Whether this is a requirement for heterogeneous nucleation on nondiamond substrates remains to be seen. Badzian and DeVries (1988), and others have reviewed the chemical kinetics model for diamond vapor deposition developed by Derjaguin and Fedoseev (1977) and Fedoseev and coworkers (1984~. They start from thermodynamic calculations that show that the equilibrium pressure of carbon vapor over ~ diamond is about 2 times higher than over graphite. In the absence of other factors, graphite nucleation is more probable than diamond. The Russian scientists considered ways to overcome this inclination. In this approach, which considers chemical kinetics in the context of classical nucleation theory, there are several key points. First the presence of a diamond! surface (e.g., autoepitaxy or diamond-seeded surface) increases the probability of diamond nucleation. Second, there should be a small range of conditions over which the nucleation rate of cliamonci is greater than that of graphite, although this is not well defined (Figure 2-~. Third, H2 dilution decreases the growth rate of graphite more than diamond. And fourth, atomic hydrogen etches graphite faster than diamond. Thus, the Russian literature concluder! that a s~ngle-step process for making diamond was possible, as shown in Figure 2-9. In Figure 2-9 note that (a) in the case of CH4 pyrolysis at certain supersaturation conditions, the specific growth rate of diamond (ad) is higher than that of graphite (v ), but both increase together; (b) in the presence of H. etching of diamond is insignificant, but the etching of graphite is dominant, and (c) for mixtures of CH4 and H there are conditions of pressure and temperature where growth of diamond is significant anti graphite is etched. In this competition graphite nucleation and growth will essentially cease or will be covered by the diamond phase (Badzian and DeVries, 1988~. Data on diamond deposition rate as a function of temperature (Derjaguin et al., 1968) show a maximum at about lOOO.C (under a limited range of conditions), which is consistent with this model. O x FIGURE 2-8 Ratio of diamond to graphite nucleation velocity as a function of supersaturation x (arbitrary units) (Badzian and DeVries, 1988~.

38 A. Only CH4 Growth only B. Only H° Etching only 1 - ~ . _ . _ I' C. CH4 + H° Combined growth and etching _ 9 d FIGURE 2-9 Relative growth and etch rates of diamond and graphite in the presence of a hydrocarbon (CH4) and atomic hydrogen (Badzian and DeVries, 1988~. BadzIan and Devries (1988) extend this approach by considering surface reconstructions and relaxations on the (111) surface of diamond, which are known to occur also in the 900 to 1000.C range (Pate, 1986~. Other phenomena and properties that critically depend on temperature in this same range include adsorption and Resorption of hydrogen, etch pit orientation, coefficient of static friction with metals, and oxidation rate. Based on this set of circumstantial evidence as a whole, a nucleation mode! that requires a critical temperature was suggested. Clearly, else nucleation stage in diamond growth poses the most important questions that will have to be answered before there is significant progress in solving practical problems of low- defect-density polycrystalline films, high purity, homoepitaxy, heteroepitaxy, and controlled morphology. Yet the current models of nucleation are primarily speculative, with a minimal amount of conclusive results reported. There is still no real understanding of critical nucleus size and crystallographic orientation; effect of impurities; catalytic effects (if any); intermediate layers berg., distinct or graded, stable carbides, hydrogenated); defect generation, propagation, and elimination; and gas-phase nucleation. At an even more basic level, we still do not know enough about the detailed surface chemistry and crystallography in this carbon-hydrogen system to know whether diamond growth is the thermodynamically unexpected or expected structure. It has been suggested recently (Machlin, 1988) that there is a thermodynamically stable region for diamond in the C-H system and that the diamond could be the expected phase to nucleate and grow under certain low- pressure conditions. Growth Mechanisms Before understanding diamond nucleation and growth from a vapor, a mechanistic and chemical kinetics mode} must be developed. The key to any mode} will, of course, be a knowledge of the species at the growing film surface. Derjaguin and Fedoseev (1977) first

39 suggested that CH4 H complexes dominate in diamond growth (and CH3 radicals dominate for graphite growth) in conventional (nonenhanced) CVD processes from hydrocarbon gases. Recently, two alternative mechanistic models have been proposed that attempt to take into account both thermal and plasma enhancement~of the hydrocarbon CVD process. Tsuda et al. (1986) and Marria et al. (1981) have proposed methyl radicals as the dominant growth species, while Freuklach and Spear (1988) suggest acetylene molecules. In both models, quantum chemical calculations for the transition states have been carried out (Tsuda et al., 1987; Huang et al., 1988~. Although the acetylene-based mechanism is more energetically favored, other mechanisms cannot be discounted. - Optical emission, mass spectrometric (Harris et al., 1988), and infrared absorption spectroscopy (Cell) et al., 1988) measurements of gas-phase species during enhanced CVD diamond growth have shown that both CH3 and C2H2 species are present in sufficient quantities to account for diamond growth. Harris et al. (1988) showed that, other than those two species, Lli4 and U2~4 are the only other species present in large enough quantities to account for diamond deposition. Their mass spectroscopy study of species present in the heated-filament- enhanced CVD as a function of filament distance from the substrate is shown in Figure 2-10. More work is required before a definitive judgment can be made regarding the dominant growth species; however, progress is being made. O^^' an ° °° ° z O O 0.001 ~ O LU 0.0001 o _ ~ _ A to g O at. S. . ·e · . , . 1 Q 20 30 FILAMENT TO SUBSTRATE D ISTANCE (MM) 40 FIGURE 2-10 CH4 (open squares) and C2H2 (solid circles) mole fractions at the growing diamond surface as a function of distance between substrate and filament. The CH4 and C2H2 signals include contributions from CH3 (small) and C2H (negligible), respectively. The clasl~ed line shows the input CH4 mole fraction (Harris et al., 1988~.

are: Impurities 40 Some of the important aspects of diamond film growth that remain essentially unanswered . Species on the growing film surface. Role of atomic hydrogen on the different crystallographic faces of diamond during growth. · Surface structure during film growth. · Catalytic processes, if any. . Surface kinetics and surface diffusion. . Ion, electron, and radical bombardment during film growth. · Synergistic effects of surface chemistry and bombardment. · Growth rates--e."., rate-limiting process~es); effect of impurities. The current vapor-deposited materials are extremely impure, especially by semiconductor standards. Atomic hydrogen, which is a necessary ingredient in diamond film deposition, is a highly reactive element. Since silicon substrates and fused silica reactor tubes and bell jars are often used, silicon is a common impurity. Levels of 0.2 atomic percent silicon have been reported (Badzian et al., 1988) for films with close to optimum characteristics. It is expected that deposition techniques that provide sufficient H° at the growing film surface and minimize H° interactions with the substrate and containment vessel will be developed for those applications requiring low or controlled levels of impurities. At that point the purity of the starting gases will become important. The main nonmetallic impurity is hydrogen, with the lowest levels reported (Lanford, 1987) being 435 + 79 ppma. The only systematic measurements reported (Hartnett, 1988) show that there is an inverse relation between [H] in film and the H2 concentration employed during deposition, with the lowest level (see above) prepared in a 0.5 percent H2-CH4 gas mixture. In some instances (Lanford, 1987), the [H] concentration was over 1 atomic percent, and yet the films still showed the diamond diffraction pattern. Silicon impurities have been related to an SO-fold increase in nucleation rate and a 2 to 3 times increase in growth rate (Badzian et al., 1988~. However, the Raman spectra indicate an associated increase in the 1332 cm~4 and a strong, narrow band luminescence peak near 6000 cam. The origin of this luminescence band is still not understood, although an increase in the silicon content has been suggested to be related to the presence of silicon in octahedral holes of the diamond lattice (Hartnett, 1988). The elimination and then controlled addition of impurity atoms is expected to be an important aspect in both the scientific understanding and the technological development. Little information has been reported to date.

41 Defects Essentially nothing is known about point defects in CVD diamond films (vacancies, interstitials, voids, or substitutional elements), either extrinsic or impurity-related. Theoretical and high-resolution TEM (Glass et al., 1988) studies are expected to help understanding. Planar defects such as microtwins, stacking faults, and graphite inclusions have been viewed by medium- and high-resolution TEM (Glass etal., 1988~. However, their relation to preparation conditions, impurities, etc., is unknown. Other defects such as dislocations and grain boundaries have not been fully investigated. Since defects are important to essentially all film properties, considerable emphasis will have to be placed here. Nucleation and renucleation are expected to be related to defects and impurities. At present candy a little is known about the most obvious defects associated with stacking faults. Morphology Although a wicie range of techniques have been used to deposit diamond films, the film morphologies reported are similar. Also, in all cases the most structurally perfect crystals occur over a narrow range of process parameters, as described earlier. The diamond crystals grow as cubes, octahedra, cuboctahedra, and single and multiple (111) twins and intergrow as polycrystalline layers with morphologies ranging from well-defined facets to poorly defined rounded shapes, highly irregular forms, etc. The crystal facets in many of the diamond films - show layer-like growth as expected from surface diffusion to steps and re-entrants (Badzian et al., 1988) and appear similar to crystals grown rather fast under high supersaturation conditions. One of the problems facing diamond thin films is the fact that, at the conditions for growth of the most structurally perfect crystals, the faceted top surface of a typical 5-,~m-thick film shows a distribution of facet sizes usually centered around 1 to 2 am. Such films are highly scattering, thus limiting optical applications as well as optical analysis techniques such as spectroscopic ellipsometry and IR and optical transmission. Rough films are also unsuitable for many tribological applications as well. There are two approaches to obtaining smooth films: essentially zero nucleation density (single-crystal films) and extremely high nucleation density (submicrometer grain-size films). For applications in which mechanical properties and deposition on a nondiamond substrate are important, the latter approach is better.- However, associated with fine-grain-size diamond films appears to be an increasing percentage of nondiamond material, as indicated by their Raman spectra (see Chapter 3~. Certainly, as the morphological units get smaller, the internal surface area (internal morphology) increases, and the volume of material affected by these surfaces can be a significant fraction of the total volume, especially when the crystallite size approaches 200 to 100 A. To what extent this phase purity problem is important remains to be seen. Certainly the answer for each application is expected to be different, and trade-offs will have to be made. Other properties expected to be related to morphology (both internal and external) are adhesion, internal stress, friction, chemical stability, interdiffusion, electrical conductivity, and thermal conductivity. An evolutionary growth model for diamond films is expected to be important as attempts are made to design and engineer films for specific applications. A complete description and understanding will entail a number of the fundamental issues already discussed--nucleation intermediate layers, renucleation, intergrowth of crystals, evolutionary selection, and effects of

42 impurities and defects. A knowledge of how to control nucleation rate and density, and perhaps to vary the growth rate in a deliberate manner during the early growth stage or throughout the process (e.g., continuously graded or periodically graded to make multilayers), will be a desirable goal in future studies. Diamond film morphology (often referred to as microstructure) has at least five facets, namely, (a) growth cones, (b) grain size, (c) grain shape, (~) grain orientation, and (e) defects. Growth cones are a very common feature of many different types of deposition processes and, in fact, are a common occurrence in deposition of some of the materials discussed in this report, e.g., SiC. Growth cones commonly affect a variety of properties, usually in a negative fashion. Being collections of grains of very limited misorientation, they can commonly lead to anisotropy of properties, i.e., different properties through the thickness than in the place of the coating. They typically exacerbate surface roughness and surface finishing and limit fracture toughness and, especially, strength due to their acting as pseudo-large grains, and their boundaries or frequent locations for accumulation of impurities, as well as possible porosity entrapment. Both of the latter factors, of course, can have other negative affects, e.g., on thermal conductivity. Grain size has a variety of effects that can be either good or bad. Hardness typically decreases with increasing grain size, even in quite hard materials, e.g., SiC and B4C. Wear resistance follows a similar trend, again as shown in such hard materials as SiC and B4C. It is not clear whether fracture toughness has any significant dependence on grain size in cubic materials, but tensile and compressive strengths clearly do, again with larger grains giving lower strengths. Clearly, for some of the electronic semiconductor applications, grain sizes must be large enough to give a single crystal region of sufficient size for the semiconductor junction to have its proper effects. Finally, grain size also plays a role in resultant surface finish. Larger grains tend to lead to rougher as-deposited surfaces and also often limit the quality of surface finishing that can result from machining. Grain shape, in particular the issue of whether or not columnar grains are formed, can also be important,- since columnar grains may lead to lower transverse fracture toughness, i.e., easier transverse cracking. On the other hand, columnar grains may lead to enhanced transverse thermal conductivity. Defects ranging from point, line, and planar lattice defects to voids and impurity particles can all play a role in a variety of properties ranging from electronic to thermal to mechanical. Generally, defects are detrimental to most of these properties; they can in some cases, however, be either neutral or beneficial (e.g., lattice defects can be beneficial to hardness. Film morphology has also been related to residual stresses in PVD fibers, although not in a quantitative manner. Generally dense, fine-grain-size fibers are in a high state of compressive stress, whereas coarser grain fibers often are closer to a neutral stress state or even slightly tensile. Such stresses are, again, a very common problem in deposited coating. Such stresses are frequently observed to be serious problems in some of the materials discussed in this report, e.g., SiC. Another possible important source of such residual stresses may be stoichiometry variations, which might thus be less severe in diamond coatings, but certainly could continue to be a similar problem in BN coatings. Such stresses become of increasing concern and seriousness as both the thickness as well as the lateral dimensions of the coating increase. Hence, these can become an important issue in scaling-up for many applications.

43 PHYSICAL VAPOR DEPOSITION PROCESSES During the past approximately 35 years, many methods involving ion-assisted deposition for the production of a variety of superhard materials have been studied. These processes have all attempted to take advantage of the production of metastable materials by the condensation of energetic particles. The energetic particles have come from plasma sources, ion sources, and sputtering sources. Very hard materials that are amorphous, by x-ray diffraction, can be made by these processes. Small crystallites of single-crystal material have been reported from these ion-assistec! deposition processes. However, success in producing macro-polycrystalline materials has been limited. In these processes the energy provided is by exchange with energetic particles rather than from thermal sources. Thus these materials can be grown on relatively low-temperature substrates. In the case of diamond-like carbon films, the coatings have been reported to possess high electrical resistivity, high hardness, low coefficient of friction, and a relatively high transparency in the infrared. Optically transparent films have been reported (Lettington et al., 1987; Wort and Lewis, 1987), and consequently many applications for these coating have been directed toward optical systems. Basic Processes Many processes have been evaluated for the production and deposition of energetic or ionized particles. Angus and coworkers (1986) classifier! the different preparation techniques as to whether hydrocarbon gases or solid carbon sources are used. These are shown schematically in Figures 2-11 and 2-12, respectively. Table 2-3 (as modified from Weissmantel, 1985) is a compilation of many references related to this work. Examination of this body of work reveals that basically three processes have been exploited for formation of these materials: (a) extraction of ions from a plasma essentially in the same region ant! pressure regime as the sample; (b) extraction or repulsion of ions from a plasma created in a region physically or electrically separated from the sample location; and (c) use of ions to sputter material from a target to the sample. Figure 2-13 shows a typical ion plating configuration for the generation of ions and their ~ e . . . , ~ me, ~ , ~ ~ ~ ~ ~ ~ propulsion to the substrate and growing Elm. In tills case, electrons emitted trom a not cathode create a plasma discharge region or sheath in which molecules are introduced either through a gas inlet or through evaporation by a separate e-beam evaporator. These volatile molecules are ionized in the plasma discharge region and extracted from the region by the potential between the substrate and the plasma potential. Reactive and inert gas ions bombard the surface at a potential related to the bias potential applied between the substrate and plasma potential zone. The ion energy spread is wide clue to collisions at the gas pressures used for generation of the plasma. An oscillating potential can be applied to the anode grid in order to increase the electron flight path and hence increase ionization efficiency. Depending on the bias voltage, the hydrocarbon molecules, and the chamber pressure, film growth rates have been measured in the 50 to 100 A/sec range. Commercial equipment is available that employs this technology for production of diamond-like coatings, and related systems have been employed to coat substrates up to 1 m2 in size.

44 ~ ~ I: a. RF parallel plate b. RF inductive discharge C. DC glow discharge ~- 1 ~ T ~ d. DC glow discharge with biased screen · rnlt't' : ::::! ~~N i 1 '~< at_ ~ —Act CATHODE Cm ~ BIAS e. Triode I. Hot filament discharger ~ ~ SU BSTRATE 1 ~i T _ n 9. Hot filament discharge h. Pulsed discharge rail gun with auxiliary ion beam FIGURE 2-11 Processes for growing carbon films from hydrocarbon gases (Angus et al., 1986~. SU BSTRATE And CARBON TARGET a. Dual beam sputtering SU BSTRATE LASER ~ Ar `~'~ CARBON TARGET c. Laser evaporation with auxiliary ion beam CARBON ELECTRODES i~ ~ SUBSTRATE T ~ r1 < :~ ~STRATE \ /C CARBON TARGET b. Single beam sputtering o o] Ar I ' ~ Ar CARBON "C ~ CATHODE BIAS d. DC glow discharge win accelerating grids ANODE CARBON Rl NG SU BSTRATE CATS BIAS e. Pulsed discharge rail gun f. Vacuum arc~ischarge runs off cathode material FIGURE 2-12 Processes for growing carbon films using solid carbon (Angus et al., 1986).

45 TABLE 2-3 Decomposition Data for DLC, BE, and Related Composites Techninue ~PAr~t.~nt.~:; Approximate Deposition Ion Energy (eV) Pressure or Sample __ (Torr ) Bias Comments References DC plasm C282 10 3-10 4 - Glossy fiLns 1,2 Oepositi~ -2 C2H4, (Ar) 1.2x10 2000-5000 V Black hard films 3-5 C2H2 0 9 300- 400 V Yellow or brown inert films 6, 7 RE. Plasma C4H1o,(CH4,C3H~) 0.2 lO0-1900 V Hard insulating films, 8-14 Depositian depend' ng on parameters C2H2 0 5 Yellow-brownish insulating 15 f ilms -2 C2H6, (C2H2,C2H4,C3H`3) 7x10 500-1000 V Hard insulating films, p = 1.9- 16-18 _3 2.0 gtcm C2H4 7x10 10-1400 V Very low friction coefficients 19,20 c3~8 .15 2 400- 700 V Optical coatings 21 C6H6 2. 2x10 3 400-1800 V Optical and protective coatings 22, 23 CH4,C2H6 7x10 500-2000 V Coatings with diamond-like 24, 25 -2 properties CH4 3x10 540 V Investigations on kinetics and 26, 27 structure -2 CH4 ,C2H2, C2H4 7x10 - Particle conglomerates or films 28 containing microcrystals C3H~ (Ar ) ~ ~ GeC interlayer to improve 28 aches ion 7. 5x10 6 10-1400 V Study of tribology properties 30 - 4 Sputter C, (Ar,Kr) 1. 5x10 - Ion-beam sputtering 31-36 Deposition C,H2, (Ar) .4 - Ion-bean sputtering with H 37,38 participation _ f C,CH4, (Ar) 1.5x10 3 200-1000 eV Dual-ion-beam sputtering 33,38-43 C, (Ar) 4.5x10 3 0- 300 V RF bias sputtering, 44 C'C2H2, (Ar) lx10 - Hybrid process: do magnetron 45-50 sputtering and plasma 2 decomposition C ~C4Hl0 3. 7x10 0- 100 V Hybrid process, low-stress 51 f iLrns - Ar 3 x10 2 150 V Ion bearT~ sputter with assist 52 beam Ar 7. 5x10 3 - Magnetron sputter at various 53 power levels Ar 7 . 5x10 3 0- 150 V Unbalanced magnetron with bias 54 - - - Fibrous carbon source 55 -4 Ian C6H6 (Tetraline, 2.2x10 20-5000 V Deposition rates of 30-100 8, 38, 56-58 Plate antrhacene ) nrn/min at low pressures Tech quotes 2 C6H6,Al,Si,Ti,Cr 2.2x10 200-3000 V Preparation of MeJi-corT~posites 38, 59-54 C,CH4,(Ar) 3000-5000 V Columnar film morphology Ion Ben C,(Ar) 3.7x10 5 20- 100 eV Very hard, inert, transparent 9,65-70 Deposition f i lens C ~ cH4 -2 900 eV Epitaxial layers on diamond CH4 7. 5x10 20- 800 V Hard films containing micro- 71 9 crystals C, (Ar) 5 x 10 30- 100 eV Mass filtered ion beam, small 72 crystallites Cl (Ar) 7 x 10 / 10-1000 eV Mass filtered ion beard high 73 res, stivity coatings C:) 5 x 10 10 1- 300 eV Mass filtered ion beam 74 CH4 (Ar) 2 x 10 5 100 eV Dual ion beam deposition 75 CH4 (Ar) 5 x 10 90- 250 eV Single beam direct deposit, on 76

46 TABLE 2-3 Decomposition Data for DLC, BN, and Redacted Composites (Continued) Approx' mate Deposition Ion Energy (eV) Pressure or Sandpile Technique Reactants ( Torr ~ Bias OtCher Techni quotes C, (Ar) ~ O- 980 en cH4 C, (Ar ) C(Ar) Deposition BM, N2, (Ar ~ of i-" BN,N2, it) B,N2,NH3, (Art BH3, Bin, N2, NH3 B3,N3,H6, (Ar) B ,~2 B,N2 B3N3~6 BN,Ar Review Articles 0.37 750 1 x 10 6 7. 5xlO 3 4 . SxlO 3 -3 1. SxlD 0- 100 V 0-3500 V 2.2xlO 4 200-1000 ~ 7. 5xlD ~ 40 ken 5 x 10 7 _,- 5 x 10 ~ Continents References . . Laser evaporation with ion bombardment Pulsed plasma process Ion generation from randomly moving arc spot, high depos- ition rates Coaxial carbon plasma gun RF sputtering, stoichiometric films RF bias sputtering, inert, dense films Ion plating combined with electron beam evaporation Plasma decomposition Plasma decomposition Electron beam evaporation of B with nitrogen ion bombardment 2- 25 kV Boron F-beam evaporated, nitrogen ions Stoichiometric EN FAB dual beam sputter deposition 34,7? 80 81, 82 83 21, 32, 38, 5S, ~0 84-86 32,56,87 88, 89 ~0 91 92 Review of: UK RSRE applications 93 A review of recent work on hard 94 DLC films DLC for thin film media for 95 magnetic recording Kev to References: 1 Schmellenmeir, H. 1953. Exp. Tech. Phys., Vol. 1, p. 49. 2. Schmellenmeir, H. 1955-56. Z. Phys. Chem., Vol. 205, p. 349. 3. Whitmell, D. S., and R. Williamson. 1976. Thin Solid Films, Vol. 35, p. 255. 4. Pickering, M., N. R. S. Tait, and D. W. I`. Tolfree. 1980. Deposition and characteristics of ion plated carbon coatings. Philos. Mat., Vol. A42, p. 257. 5. Tait, N. R. S., and D. W. L. Tolfree. 1981. Phye. Status Solid, Vol. A69, p. 329. 6. Meyerson, B., and F. W. Smith. 1980a. Electrical and optical properties of hydrogenated amorphous carbon films. J. Non-Crystal. Solids, Vol. 35736, p. 435. 7. Meyerson, B., and F. W. Smith. 1980b. Chemical modification of the electrical properties of hydrogenated amorphous carbon films. Solid State Comm., Vol. 34, p. 532. 8. Holland, L., and S. M. Ojha. 1979. The growth of carbon film with random atomic structure from ion impact damage in a hydrocarbon plasma. Thin Solid Films, Vol. 58, p. 107. 9. Holland, L., and S. M. Ojha. 1976. Deposition of hard and insulating carbonaceous films on an RIP target in a butane plasma. Thin Solid Films, Vol. 38, p. L17. 10. Holland, L., and S. M. Ojha. 1978. Infrared transparent and amorphous carbon grown under ion impact in a butane plasma. Thin Solid Films, Vol. 48, p. L21. 11. Holland, L. 1980. A review of plasma process studies. Surface. Technol., Vol. 11, p. 145. 12. Ojha, S. M., and L. Holland. 1977. Proc. Int. Vac. Congr., 7th, Vienna, 1977, p. 1667. 13. O3ha, S. M., H. Norstrom, and D. McCulluch. 1979. The growth kinetics and properties of hard and insulating carbonaceous films grown in an RF discharge. Thin Solid Films Vol. 60, p. 213.

47 14. Vora, H., and R. J. Mora~rec. 1981. Structural investigation of thin films of diamondlike carbon. J. Appl. Phys., Vol. 52, p. 6151. 15. Anderson, D. A. 1977. The electrical and optical properties of amorphous carbon preapred by the Glow discharge technique. Philos. Mag., Vol. 35, p. 17. 16. Andersson, L. P. 1981. A review of recent work on hard I-C fimns. Thin Solid Films, Vol. 86, p. 183. 17. Andersson, L. P., and S. Berg. 1978. Initial etching in an of butane plasma. Vacuum, Vol. 28, p. 449. 18. Andersson, L. P., S. Berg, H. Norstrom, and S. Towta. 1979. Properties and coating rates of diamond-like carbon films produced by FR glow discharge of hydrocarbon gases. Thin Solid Films, Vol. 63, p. 155. 19. Enke, K. 1981. Thin Solid Films, Vol. 80, p. 227. 20. Gambino, R. J., and J. A. Thompson. 1980. Spin resonance spectroscopy of amorphous carbon films. Solid State Commun., Vol. 34, p. 15. 21. Beale, H. A. 1979. Cubic boron nitride. Ind. Res./Dev. (June), p. 143. 22. Bubenser, A., B. Dischler, and A. Nysish. 1982. Optical properties of hydrogenated hard carbon thin films. Thin Solid Films, Vol. 91, p. 81. 23. Dischler, B., A. Bubenzer, and P. Koidl. 1983. Hard carbon coatings with low optical absorption. Appl. Phys. Lett., Vol. 42, p. 636. 24. Fujimori, N., A. Doi, and T. Yoshioka. 1983. Proc. Conf. Japan Society of Applied Physics, 30th, Tokyo, p. 214. 25. Doi, A., N. Fujimori, T. Yoshiolta, and Y. Doi. 1983. Plasma deposition of crystalline carbon and I-Carbon film. Proc. Int. Ion Eng. Cong., Kyoto, 1983 IEEJ, Tokyo, p. 1137. 26. Njalesh and Nowak, 1983. 27. Hashimoto, K., T. Sakamoto, N. Matuda, S. Baba, and A. Kinbara. 1983. Physical proeprties of carbon films produced by plasma CVD. Proc. Int. Ion Eng. Cong., Kyoto, TEEJ, Tokyo, p. 1125. 28. Has, Z., S. Mitura, and B. Wendler. 1983. rf-plasma deposition of thin diamond-lie carbon film. Proc. Int. Ion Eng. Cong., Kyoto, 198S IEEJ, Tokyo, p. 1143. 29. Lettington, A. H. 1987. Les Editions de Physique Paris, Vol. XVII. 30. Gautherin, G., and C. Weisemantel. 1978. Some trends in preparing film structures by ion beam methods. Thin Solid Films, Vol. 60, p. 135. 31. Weissmantel, C. 1981. Ion beam deposition of special film structures. J. Vac. Sci. Technol., Vol. 18, p. 179. 32. Preusse, J. 1984. Thesis, Technische Hochschule, Karl-Marx-Stadt, German Democratic Republic. 33. Fujimori, S., T. Kasai, and T. Inamura. 1982. Carbon film formation by laser evaporation and ion beam sputtering. Thin Solid Films, Vol. 92, p. 71. 34. Mathine, D., R. O. Dillon, A. A. Khan, G. Bu-Abbud, J. A. Woollam, D. C. Liu, B. Banks, and S. Domitz. 1984. Optical, Raman, and photoemission results on amorphous ~diamondlike" carbon films. J. Vac. Sci. Technol., Vol. A2, p. 365. 35. Strelniski, V. E., V. G. Padalka, and S. I. Vakula. 1978. Properties of the diamond-like carbon film produced by the condensation of a plasma stream with an rf potential. Zh. Teck. Fiz., Vol. 48, p. 377. 36. Mackowski, J. M., M. Berton, P. Ganau, and Y. Touze. 1982. Vide, Vol. 212, p. 99. 37. Weissmantel, C., K. Bewilogua, K. Breauer, D. Dietr~ch, U. Ebersbach, H. J. Erler, B. Rau, and G. Reisse. 1982. Preparation and properties of hard I-C and I-BN coatings. Thin Solid Films, Vol. 96, p. 31. 38. Weisemantel, C. 1977. Trends in thin film depositor methods. Proc. Int. Vac. Cong., 7th, Vienna, 1977, p. 1533. 39. Weissmantel, C. 1976. Ion beam utilization for etching and film deposition. Vide, Vol. 183, p. 107. 40. Weissmantel, C., G. Reisse, H. J. Erler, F. Henny, K. Bewilogua, U. Ebersbach, and C. Schurer. 1979a. Preparation of hard coatings by ion beam methods. Thin Solid Films, Vol. 3, p. 315. 41. Nikiforo~ra, N. N., M. B. GuseYa, and V. G. Babaev. 1982. Proc. USSR Conf. Interactions Atomic Particles Solids, 6th, Minsk, 1982, p. 112. 42. Banks, B. A., and S. K. Rutledge. 1982. Ion beam sputter-deposited diamondlil~e films. J. Vac. Sci. Technol., Vol. 21, p. 807. 43. Sathyamoorthy, A., and W. Weisweiler. 1982. Studies of the sputter deposition of carbon, silicon and SiC films. Thin Solid Films, Vol. 87, p. 33. 44. Harding, G. L., B. Window, D. R. McKensie, A. R. Collins, and C. M. Horwitz. 1979. Cylindrical magnetron sputtering system for coating solar selective surfaces onto batches of tubes. J. Vac. Sci. Technol., Vol. 16, p. 2105. 45. Harding, G. L. 1980. Absorptance and emittance of metal carbide selective surfaces sputter deposited onto glass tubes. Sol. Energy Mater., Vol. 7, p. 101. 46. Harding, G. L., S. Craig, and B. Window. 1982. Production and properties of sputtered thin films for solar selective absorbing surfaces. Appl. Surf. Sci., Vol. 11/12, p. 315. 47. Craig, S., and G. L. Harding. 1982. Structure, optical properties and decomposition kinetics of sputtered hydrogenated carbon. Thin Solid Films, Vol. 97, p. 345. 48. McKenzie, D. R., and L. M. Brigge. 1981. Properties of hydrogenated carbon films produced by reactive magnetron sputtering. Sol. Energy Mater., Vol. 6, p. 79. 49. McKenzie, R. C. McPhedran, N. Sawides, and L. C. Botton. 1983. Properties and structure of amorphous hydrogenated carbon films. Philos. Mag., Vol. B48, p. 341. 50. Zelez, J. 1983. Low-stress diamondlike carbon films. J. Vac. Sci. Technol., Vol. At, p. 305. 51. Pellicori, S. F., C. M. Peterson, and T. P. Henson. 1986. Transparent carbon fimns: comparison of properties between ion- and plasma-deposited processes. J. Vac. Sci. TechnoI., Vol. Ad, No. 5, p. 2350. 52. Sawides, N., and B. Window. 1985. Diamondlike amorphous carbon films prepared by magnetron sputtering of graphite. J. Vac. Sci. Technol., Vol. A3, No. 6. 53. Sav~rides, N., and B. Window. 1986. Unbalanced magnetron ion-assisted deposition and property modification of thin films. J. Vac. Sci. Technol., Vol. Ad, No. 3. 54. Pulker, H. K., and R. Staffler. 1979. U.S. Patent 4,173,S33.

~8 55. Weissmantel, C., K. Bewilogua, D. Dietrich, H. J. Erler, H. J. Hinneberg, S. Klose, W. Nowick, and G. Reisse. 1980a. Structure and properties of quasi-amorphous films prepared by ion beam techniques. Thin Solid Films, Vol. 72, p. 19. 56. Weisemantel, C., H. J. Erler, and G. Reisse. 1979b. Ion beam techniques for thin and thick film deposition. Surf. Sci., Vol. 86, p. 207. 57. Weisemantel, K. Bewilogua, H. J. Erler, H. J. Hinneberg, S. Klose, W. Nowick, and G. Reisse. 1980b. p. 188. In Low Energy Ion Beams, I. H. Wilson and K. G. Stephens, ede. Institute of Physics Conf. Series 54, Institute of Physics, Bristol and London. 58. Weissmantel, C., K. Breuer, and B. Winde. 1985. Hard films of unusual microstructure. Thin Solid Films, Vol. 100, p. 383. 59. Weisemantel, C. 1983a. Proc. Int. Vac. Congr., 9th, Madrid, 1983, p. 299. 60. Bewilogua, K., E. Bugiel, B. Rau, C. Schurer, and C. Weiesmantel. 1980. Krist. und Tech., Vol. 15, p. 1205. 61. Weisemantel, C. 1983b. lon-based preparation of semiconductor and hard materials films. Proc. Int. Ion Eng. Cong., Kyoto, IEEJ, Tokyo, p. 1257. 62. Teer, D. G., and M. Salama. 1977. The ion plating of carbon. Thin Solid Films, Vol. 45, p. 553. 63. Salama M. S. 1983. Deposition and characteristics of ion plated carbon coatings. Proc. Int. Ion Eng. Cong., Kyoto, IEEJ, Tokyo, p. 1 64. Aisenberg, S., and R. W. Chabot. 1971. Ion-beam deposition of thin film of diamondlike carbon. J. Appl. Phys., Vol. 42, p. 2953. 6S. Spencer, E. G., P. H. Sachmidt, D. 3. Joy, and F. J. Sanesalone. 1976. Ion-beam-deposited polycrystalline diamondlike fi}rne. Appl. Phys. Lett., Vol. 29, p. 118. 66. Aisenberg, S. 1984. Properties and applications of diamondlike carbon films. J. Vac. Sci. Technol., Vol. A2, p. 369. 67. Furuse, T., T. Susuki, S. Matsumoto, K. Nishida, and Y. Nannichi. 1978. Insulating carbon coating on (ALGA) as DH laser facets. Appl. Phye. Lett., Vol. 33, p. 317. 68. Mora~rec, T. J., and T. W. Orent. 1981. Electron spectroscopy of ion beam and hydrocarbon plasma gnerated diamondlike carbon films. J. Vac. Sci. Technol. Vol. 18, p. 226. 69. Freeman, J. H., W. Temple, D. Beanland, and G. A. Gard. 1976. Ion beam studies. T. Teh retardation of ion beams to very low energies in an implantation accelerator. Nucl. Instrum. Methods, Vol. 1SS, p. I. 70. Mori, T., and Y. Namba. 1983. Hard diamondlike carbon films deposited by ionized deposition of methane gas. J. Vac. Sci. Technol., Vol. At, p. 23. 71. Chaiko~rskii, E. F.l V. M. Pusiko~r, and A. V. Semenou. 1981. Sov. Phys. Crystallogr., Vol. 26, No. 1. 72. Ishikawa, J., Y. Takeiri, K. Ogawa, and T. Takagi. 1987. Transparent carbon film prepared by mass-separated negative-carbon- ion-beam deposition. Jpr. J. Appl. Phys., Vol. 61, No. 7, pp. 2S09-2515. 73. Kasi, S., H. Kang, and J. W. Rabalais. 1987. Cherrucally bonded diamondlike carbon films from ion-beam depositor. Phys. Rear. Letters, Vol. 59, No. 1. 74. Mirtich, M. J. 1981. Ion beam deposited protective films. NASA Technical Memorandum 81722. NASA Lewis Research Center. 7S. Nir, D., and M. Mirtich. 1986. Thin film growth rate ejects for primary ion beam deposited diamondlike carbon films. J. Vac. Sci. Technol., Vol. Ad, No. 3. 76. Fujimori, N., and K. Nagai. 1981. The effect of ion bombardment on carbon films prepared by laser evaporation. Jpn. J. Appl. Phye., Vol. 20, p. L194. 77. Sokolowski, M., and A. Sokolowska. 1982. Electric charge influence on the metastable phase nucleation. J. Crystal Growth, Vol. 57, p. 185. 78. Dorodno~r, A.M. 1978. Zh. Tekh. Fis., Vol. 9, p. 48. 79. Sater et. al., 1984. 80. Aita, C. R., and N. C. Tran. 1982. J. Appl. Phys., Vol. 54, p. 6052. 81. Wiggins, M. D., C. R. Aita, and F. S. Hickernell. 1983. 3. Vac. Sci. Technol., Vol. A2, p. 322. 82. Roth, D., H. 3. Erler, B. Rather, L. Richter, K. Peter, and B. Winde. 1984. Proc. Conf. Hochvakuum, Grenzflachen, Dunne Schichten, 9th Dresden, 1984. Vol. 1, p. 265, Phye. Sac. of GDR, Berlin. 83. Gafri, D., A. Grill, D. Itshak, A. Inspel~tor, and R. Audi. 1980. Boron nitride coatings on steel and graphite produced with a low pressure rf plasma. Thin Solid Films, Vol. 73, p. 523. 84. Miyamoto, H., M. Hirose, and Y. Osaka. 1983. Structural and electronic characterization of discharge-produced boron nitride. Jpn. J. Appl. Phye., Vol. 22, p. L216. 85. Sokolowaki, M., A. Sokolowska, A. Michaelski, Z. Romanowski, A. Rusek-Masurek, and M. Wronikowski. 1981. Thin Solid Films, Vol. 80, p. 249. 86. Shanfield, S., and R. Wolfram. 1983. Ion beam synthesis of curbi boron nitride. J. Vac. Sci. Technol., Vol. At, p. 323. 87. Satou, M., K. Matuda, and F. Fujimoto. 1983. Proc. Symp. Ion Sources and Ton-Assisted Technology, 6th, Tokyo, 1982, p. 425. 88. Satou, M., and F. Fujimoto. 1982. Formation of cubic boron nitride films by boron evaporation and nitrogen ion bean, bombardment. Jpn. J. Appl. Phys., Vol. 22, p. L171. 89. Andoh, K. O., Y. Suzuki, E. Kamijo, M. Satou, and F. E'ujimoto. 1987. Nuclear Instruments. 90. Hal~rerson, W., and D. T. Quinto. 1985. Effects of charge neutralization on ion-beam-deposited boron nitride films. J. Vac. Sci. Technol., Vol. As, No. 6. 91. Shimokawa, F., H. Kuwano, and K. Nagai. 1985. Proc. 9th Symp. on ISlAT, Tokyo. 92. Lettington, A. H., J. C. Lewis, C. J. H. Wart, B. C. Monachan, and A. J. N. Hope. 1987. Developments in GeC as a Durable Coating Strasbourg, June 1987. 93. Tsai, H., and D. B. Bogy. 1987. Characterization of diamondlike carbon films and their application as overcoats on thin-film media for magnetic recording. J. Vac. Sci. Technol., Vol. As, No. 6. 94. Angus, J. C., and C. Hayman. 1988. Low-pressure, metastable growth of diamond and "diamondlike" phases. Science, Vol. 214, no. 4868, pp. 913-921. 95. Angus, J. C., P. Koidl, and S. Domits. 1986. Carbon thin films. Chapter 4 in Plasma Deposition of Thin Films. J. Mort and F. Jansen, ede. Boca Raton, Florida: CRC Press.

49 ~ ,+r4: Ub Ua Uh · Electrons O ions FIGURE 2-13 Ion plating configuration. i,.,,,/t ,,,, I,;;,... _I)_t i' i ~ . ~ wi~ I , I, Electron beam I ~ evaporator Substrate Growing film , Plasma sheath , Screen , Anode Hot cathode /Gas inlet An alternative approach to the ion plating technique is the use of an rf-plasma-assisted CVD system also using a voltage bias on the sample holder. A typical configuration is shown in Figure 2-14. Radio frequency energy of 13.56 MHz is applied to the lower electrode. This creates a plasma between the lower electrode and upper electrode that is at rf ground potential. Superimposed on the rf potential is a negative bias, typically 300 to 1000 V. This results in bombardment of the lower electrode (holding the samples) by energetic species extracted from the plasma. An alternative configuration of this' same technology allows the sample electrode to self-bias due to the very different mobilities of electrons and ions. This will also result in energetic bombardment of the samples. This configuration has seen significant development for the production of optical coatings. One application has been for the deposition of hard a-C:H coatings on germanium substrates. Equipment has been constructed for coating objects up to 1 m in diameter. Figure 2-15 shows the original sacrificial cathode ion source developed for direct ion beam deposition of diamond-like carbon coatings. In this process, a high-voltage discharge is initiated between the carbon cathode and carbon anode in the ion source. Argon is supplied to this discharge, and the resulting charged argon ions sputter the negative electrode, introducing sputtered carbon into the plasma discharge. Ions of argon and carbon are extracted through an orifice by appropriate electronic extraction optics and are impinged on the sample held in a second vacuum chamber at substantially lower pressure. This technique results in amorphous carbon coatings with little or no hydrogen. In addition, the low vacuum pressure of the deposition chamber results in a more uniform ion energy profile. Small crystallites (up to 5 ~m) of diamond have also been produced by this technique.

so Hydrocarbon Gas Inlet ~r~ ~ ~~ 10-3 Torr .- 600 v . r | Upper Elect strode (Cu) l C+, H+, e- Plasma 1 1 s 1 ~ S I I S I | Lower Electrode (Cu) | 1 ~ ~ '_ J _—13.56 MHz Vacuum l Pump ~ ~ FIGURE 2-14 rf plasma configuration. Argon ~1 1111111 4~ glass to Carbon ' __~\\~1 Carbon Water channel VV"I~I ~ ~) Teflon sleeve ~ ~ . ' Lit ' 0OOOWO; Vac. port L 6- Pyrex glass Deposition chamber Teflon sleeve Slide or Silicon chip ~// q`~\\ " External ~ magnetic //~//A_~ W) field Teflon tube FIGURE 2-15 Direct ion beam deposition from solid precursor configuration.

51 Another manifestation of the single ion beam direct-deposition approach is shown in Figure 2-16. In this technique, a standard Kaufman-type ion source is employed for the generation of energetic carbon and argon ions. Because a hydrocarbon feed is used, substantial quantities (up to 30 atomic percent) of hydrogen are incorporated into the resulting amorphous coating. This inclusion of hydrogen can be beneficial in tailoring the refractive index of the resulting material. In addition to diamond-like carbon, this technique has also been employed for the production of cubic BN coatings. MAGNET COIL ANODE DISCHARGE CHAMBER B3N3M6 VAPOR osOo 06~°< CATHODE ~ 0 ° ~ FILAMEN ~ o) O it J ~ . ~ 1 o C Ido lolc Iclo talc Va ." CHARGE NEUTRALIZER FILAM ENT 1' ~ co ~ o ·0 ~ O Leo ° · ° Oo o ° 60 o o O ION o o.o · ~ EXTRACTION GRID SUBSTRATE HOLDER SUBSTRATE /. ELECTRONS lo IONS / FIGURE 2-16 Direct ion beam deposition from gaseous precursor configuration (Holland and Ojba, 1978; and Berg and Anderson, 1979~. Figure 2-17 shows a modification of the single direct ion beam deposition configuration. In this technique, originally developed by workers at NASA,s Lewis Research Center (Mirtich, 1981) and other laboratories, a second ion source is typically operated at higher ion energy than the depositing source. The resulting films have been shown to be clearer than diamond-like films produced from a single beam ion source using volatile precursor molecules. Another modification of the dual ion beam system is shown in Figure 2-18. In this system, carbon is sputtered from a graphite target by the first ion source. The sputtered carbon is allowed to deposit on the substrate with or without secondary ion bombardment. A second ion source, operated at a lower ion energy, can be employed to further bombard the sample surface. This results again in films with improved mechanical and optical properties. This technique has also been employed for the production of cubic BN films.

52 , 8 cm lon Source C ~ | 3 cm Ion Source ~: //~/\~: 1 ~ \ l I. I . ._. I ~ Cathode Neutralizer Diamond- - Like Film l l Beam Mixture of Methane and Ardors FIGURE 2-17 Dual ion beam deposition configuration (Mirtich, 1981~. ~'~'~o "A `\:source Substrate ~ Irma W%,~1 ~ ~~'_~ ~'`N fig ~ Electron gun \ FIGURE 2-~8 Ion beam sputtering with ion beam assist configuration.

53 Critical Growth Parameters Amorphous diamond-like carbon and cubic BN thin films have been produced by ion-assisted processes primarily through close control of ion energy. Many of the fundamental processes involving the nucleation of these films, the critical role of the impinging ion energy, and the relative roles of deposition ions (i.e., Chin+) versus argon ions are still not well understood despite extensive study over many years. However, it is abundantly clear that a new class of materials with very interesting physical, electronic, and optical properties can be produced by this technology. The following sections point out some of the considerations necessary in producing these materials. Substrates and Film Nucleation ~ ~ . . . . . . ~ ~ Since it is possible to operate ion-assisted processes at temperatures below lOOoC, a wide variety of substrates have been studied in the deposition of these amorphous thin films. substrates tested Include a wide variety of optical materials (CR 39 optical plastic, polycarbonate, glass, quartz, sapphire); infrarecI-transmitting optical materials, including germanium, zinc sulfide, and zinc selenide; a variety of hard substrates including normal steels. tool steels. and cemented carbide substrates; and a variety of electronic-grade materials. Although thin films can be produced on this wide variety of substrates, the adhesion properties of the coatings and their resistance to abrasion vary widely. Generally, sputtering of the substrate prior to deposition is employed to improve adhesion properties. In addition, thin interlayers such as germanium on ZnS have also been used to improve adhesion (Lettington et al., lo. The processes involved in the initial nucleation of the coating vary, depending on the deposition process. Generally, coatings produced by ion plating techniques and those produced by ion beam sputtering nucleate quite easily, and films grow rapidly thereafter. However, films producer! by direct ion beam deposition show a significant influence of substrate type on initial coating nucleation. Seeding techniques such as those discussed in the earlier section entitled "Substrates and Their Preparations are not generally employed for these amorphous coatings. Deposition Pressure As can be seen from Table 2-3, at least two distinct regions of process pressure are used in depositing these coatings. Coatings produced by ion plating and related techniques, wherein the gas discharge is in the same pressure regime as the sample, employ chamber pressures in the millitorr region. Direct ion beam deposition techniques, on the other hand, utilize pressures in the submillitorr region. Because of this fundamental difference, coatings produced by ion plating typically grow at faster rates but have poorer optical and electronic properties than those produced by ion beam processes. Ion Energy Considerations Perhaps the single most widely studied process parameter employed in producing ion-assisted coatings is control of the energy of the incident energetic particles. These studies have clearly shown that many of the resulting film properties can be directly related to the control of particle energy. Ion plating and related techniques typically use sample bias voltages of 500 to 1500 V. Because of the relatively high gas pressures utilized in these processes, the actual energy of particles impacting the surface is considerably less. In addition, a wide spread of particle energies can be expected in these techniques.

54 Direct ion beam deposition processes have focused on ion energies ranging from 50 to 150 eV. Careful control of ion energy in high-vacuum systems employing mass filtering of carbon ions has been shown to produce crystalline diamond materials up to several micrometers in size (Chaikovskii et al., 1981~. Size of Substrates Process equipment for coating substrates up to 1 m in diameter has been constructed for ion plating and related deposition systems. Until recently, the area coated by direct ion beam processes has been restricted by the unavailability of large-area ion beam source technology. Most of the experimental work has been conducted utilizing sources capable of covering areas ~15 cm in diameter or smaller. Recently, however, commercial ion beam sources have become available up to 38 cm in diameter. In addition, ion sources up to 1.5 m in diameter have been constructed at the NASA Lewis Research Center. The availability of large-coverage-area ion sources will accelerate development of direct deposition of these materials. Fundamental Issues in PVD Processes Ion-assisted PVD processes take advantage of the production of metastable species by the condensation of energetic particles. The energy contained in the condensing particle is employed in preferentially rearranging the bonding structure on the material surface. Much of the fundamental chemistry and physics occurring on the surface is still not well understood. However, some issues are clearly important, ant] a variety of theories have been advanced to explain the nature of these resulting unique materials. Growth Mechanisms As particles of varying energy impact on a solid surface, a number of basic processes can occur, including recoil of the particle from the surface, ejection of material from the surface caused by the impact (sputtering), sticking of the particle to the surface, also with resulting energy transfer from the particle to the solid surface, and implantation of the particle into the surface with resulting energy transfer. A variety and combination of these processes may well be occurring in many of the deposition approaches used for making amorphous diamond-like carbon and other hard materials. In the case of the production of diamond-like carbon it is important to understand why carbon atoms are deposited in the tetrahedrally bonded (sp3j structures versus the non-sp3 bonded carbon structure. One theory proposed to explain this observed result is that of preferential sputtering (Spencer et al., 1976~. Under the conditions used in the deposition of these coatings, an arriving energetic particle is most likely to either recoil from the surface, condense on the surface, or eject material from the surface. Studies done by Angus and coworkers (1986) and by Miyazawa and coworkers (1984) indicated that impinging energetic particles may have sufficient energy to preferentially sputter non-sp3 bonded carbon materials. Consequently, over a given time period the material remaining on the surface will contain a preferential enrichment of Sp3 structures. The second proposed concept for the stabilization of Sp3 bonded structures involves the high transient temperatures and pressures that appear in the immediate vicinity of an impacting energetic particle. Weissmantel (1981) proposed that transformations into Sp3 structures can occur during these short-lived pressure and temperature events. Although the correctness of this theory

55 has been debated, it appears clear that the growth surface is undergoing energetic bombardment of sufficient energy for a variety of transformations to occur. It does appear that, whatever the detailed mechanism of formation, the final structure of diamond-like hydrocarbons is close to that predicted for completely constrained, random covalent networks. Microstructure Diamond-like carbon films have been shown to contain a variety of hydrogen contents and range of atomic structures. The majority of the dense films have been found by x-ray examination to be amorphous. Sometimes small crystallites have been identified, particularly in films produced by ion beam processes. These crystallites have been identified as having both graphite and diamond structures. OTHER DEPOSITION PROCESSES It is interesting to speculate that if part of the diamond-making process involves the stabilization of the surface, whether it be in the equilibrium or metastable regions, then there may be other ways of doing this. There are two claims in the category of synthesis in the presence of a liquid phase under metastable conditions: (a) the report of overgrowths on diamond in a molten salt in the presence of nickel (Patel and Cherian, 1981~; and (b) the breakdown of Al6,C3 in the presence of a carbon halide in a molten salt (French and French, 1981, 1982~. The latter definitely produced fine-grained hard powders. Although there has been adequate time for the use of such processes if they were indeed successful, there appears to have been no consummation of what are obviously exciting results. These should be reexamined. Another regime for study is the C-H-O system in the pressure range up to about 20 kb. There is an apparent convergence in the literature on the reality of carbon deposition from this system in nature in liquids essentially immiscible with silicates. Yet another approach to making diamond (and other high-pressure phases) at atmospheric pressure is the solid (graphite)-solid (diamond) transformation process mediated by intense laser irradiation of a loose, free-falling powder. Fedoseev and coworkers (1983) were the first to report successful experiments by this process using a CO2 laser, and recently Alam and coworkers (1988) have confirmed these results using both CO2 and Nd-YAG lasers. The current drawbacks to this approach are the low conversion efficiency of the transformation process and the fact that powder and not continuous films are produced. OTHER SUPERHARD MATERIALS Silicon carbide is considered a superhard material in the context of this report. It is about one-third as hard as diamond and comparable to cubic BN. Cubic EN is about half the hardness of diamond and is the second hardest material known. Corundum, a-SiC, is the hexagonal (6H) phase that normally forms; it has been produced commercially for many years by the Acheson process. Generally these crystals form dark platelets of low purity. High-purity transparent SiC is more difficult to obtain because of difficulty in purity and stoichiometry control, but crystals of 1 x 5 x 5 mm have been produced.

56 There are a large number of hexagonal and rhombohedral polytypes and only one cubic (3C) polytype, 13-SiC. The ,B-phase has been produced by vapor deposition processes, and various procedures have been developed for heteroepitaxial growth on large-area silicone substrates based on various buffer layers to satisfy the mismatch in lattice parameter and thermal expansion coefficient. Thick layers have been successfully prepared by a number of workers (e.g., Nishino et al., 1980; Liaw and Davis, 1985~. Although high-purity heteroepitaxial films could be achieved, they still contain a high defect density of microtwins, intrinsic stacking faults, and antiphase boundaries. Recent results from the North Carolina State University group (see extensive review by Davis et al., 198S, which also discusses diamond films deposition) of the deposition on Acheson-type a-SiC [both Si (0001) and C (0001) faces] show that these defects could be essentially eliminated, with the only remaining defect being of the double-positioning-boundary (DPB) type. And even these DPB defects could be eliminated by 3° off-axis tilting of the substrate. However, in this latter case the resulting film is c`-SiC and not ,B-SiC. Silicon carbide films can be doped both n- and p-type, and this combined with its high thermal conductivity and electrical field breakdown strength, makes it an excellent material for electronic devices. The potential electronic and nonelectronic uses of this new high-pur~ty and low-defect density material are just beginning to be explored. REFERENCES - Alam, M., T. DebRoy, R.Roy, and E. Brevel. 1988. High-pressure phases of SiO/sub 2/ made in air by Fedoseev-Derjaguin laser process. Appl. Phys. Lett., Vol. 52, no./S, p. 16870. Bachmann, P., W. Drawe, D. Knight, R. Weimer, and R. Messier. 1988. Diamond and Diamond- like Materials Synthesis (Extended Abstracts), p. 99. M. Gels, G. H. Johnson, and A. R. Badzian, eds., Materials Research Society. Badzian, A. R. 1988. Superhard material comparable in hardness to diamonds. Vol. 53, No. 25, P. 2495. Appl. Phys. Lett. Badzian, A. R., and R. C. DeVries. 1988. Crystallization of diamond from the gas phase: I. Mat. Res. Bull., Vol. 23, no. 3, p. 385. Badzian, A. R., T. Badzian, R. Roy, R. Messier, and K. E. Spear. 1988. Crystallization of diamond crystas and filsm by microwave assisted CVD: Part II. Mat. Res. Bull., Vol. 23, p. 531-548. Berg, S., and L. P. Anderson. 1979. Thin Films, Vol. 5S, pp. 117-120. Celli, F. G., P. E. Pehrsson, H. T. Wang, and I. E. Butler. 1988. Infrared detection of gaseous species during the filament-assisted growth of diamond. Appl. Phys. Lett., Vol. 52, p. 2043. Davis, R. F., I. T. Glass, G. Lucovsky, and K. J. Bachman. 1987. Growth, Characterization, and Device Development in Monocrystalline Diamond Films. Annual Report to Office of Naval Research (Contract N000 1 1 86-K-0666, June 1987~.

57 Davis, R. F., Z. Sitar, B. E. Williams, H. S. Hong, H. I. Kim, I. W. Palmour, J. A. Edmond, I. Ryu, I. T. Glass, and C. H. Carter. 1988. Critical evaluation of the status of the areas for future research regarding the wide band gap semiconductors diamond, gallium nitride and silicon carbide. Mat. Sci. Eng., Vol. B1, p. 77. Derjaguin, B. V., and D. V. Fedoseev. 1977. Growth of Diamond and Graphite From the Gas Phase (in Russian). Moscow: Nauka. Derjaguin, B. V., L. L. Bouilov, and B. V. Spitsyn. 1968. Archiwum Nauki o Materiatach, Vol.7,p. 111. Fedoseev, D. V., V. L. Bukhovets, I. G. Varshavskaya, A. V. Laurent ev, anti B. V. Derjaguin. 1983. Transition of graphite into diamond in a solid phase under atmospheric pressure. Carbon, Vol. 21, p. 237. Fedoseev, D. V., B. V. Derjaguin, Y. G. Varshavskaja, and A. C. Semienova Tjan-Shanskaja. 1984. Crystallization of Diamond (in Russian). Moscow: Nauka. French, F. A., and D. A. French. 1981. Production of Ultrahard Particles. U.S. Patent No. 4,275,050 (filed ~ 1/~/79, issued 6/23/81~. French, F. A., and D. A. French. 1982. Ultrahard Particles of Carbon Produced by Reacting Metal Carbide With Nonmetal Halide in Hot Melt System. U.S. Patent No. 4,352,787 (filed 11/~/79, issued 10/5/82~. Freuklach, M., and K. E. Spear. 1988. Growth mechanism of vapor-<ieposite~i diamond. I. Mat. Res., Vol. 3, p. 133. Gels, M. W., G. H. Johnson, and A. R. Badzian, eds. MRS Spring Meeting Extencled Abstracts, Symposium D, p. 1 15. Glass, I. T., B. E. Williams, R. F. Davis, and K. Kobashi. 1988. Proceedings of the 1st Int. Conf. On the New Diamond Science and Technology, Tokyo, October 24-26, Paper 3-10, p. 165. Harris, S. J., A. M. Weiner, and T. A. Perry. 1988. Measurement of stable species present during filament-assiste~i diamond growth. Appl. Phys. Lett., Vol. 53, p. 1605. Hartnett, T. 1988. Characterization of Diamond Deposition in a Microwave Plasma. M.S. Thesis, Pennsylvania State University. Hirose, Y., and M. Mitsuizumi. 1988. New Diamond (in Japanese), Vol. 4, p.3~. Holland, L., and Ojba, S. M. 1978. Thin Solid Films, Vol. 4S, pp. 21-21. Huang, D., M. Freuklach, and M. Maroncelli. 1988. I. Phys. Chem., Vol. 92, p. 6379. Kamo, M. Y. Sato, S. Matsumoto, and N. Setaka. 1983. Diamond synthesis from gas phase in microwave plasma. J. Cryst. Growth, Vol. 62, p. 642. Koshino, N., K. Kurihara, M. Kawarada, and K. Sasaki. 1988. Diamond and Diamondlike Materials (Extended Abstracts), p. 95. M. Gels, G. H. Johnson, and A. R. Badzian, eds., Materials Research Society.

58 Lanford, W. 1987. 2nd Annual Diamond Technology Initiative Meeting. Sponsored by the Office of Naval Research and SDIO/IST, Durham, North Carolina, Paper 8-P-O. Liaw, P. H., and R. F. Davis. 1985. Epitaxial growth and characterization of beta -SIC thin films. I. Electrochem. Soc., Vol. 132, p. 642. Littleton, A. H., J. C. Lewis, C. ]. H. Wort, B. C. Monachan, and A. ]. N. Hope. 1987. Developments in GeC as ~ durable coating. Amorphous Hydrogenated Carbon Films, In E- MRS Meeting, Strasbourg, June. Machlin, E. S. 1988. On the stabilization of diamond relative to graphite by surfaces. ]. Mat. Res., Vol. 3, p. 958. Matsumoto, S. 1985. Chemical vapour deposition of diamond in rf glow discharge. ]. Mater. Sci. Lett. Vol. 4, no. 5, pp. 600-602. Matsumoto, S., and Y. Matsui. 1983. Electron microscopic observation of diamond particles grown from the vapour phase. I. Mat. Sci., Vol. IS, p. 1785. V. K.: Chapman & Hall. Mirtich, M. ]. 1981. Ion beam deposited protective films. NASA Technical Memorandum 81722. NASA Lewis Research Center. Mishima, O., I. Tanaka, S. Yamaoka, and O. Fukanaga. 1987. High-temperature cubic boron nitride pen junction diode made at high pressure. Science. Vol. 23S, p. 181. Miyazawa, T., S. Misawa, S. Yoshida, and S. Gonda. 1984. Preparation and structure of carbon film deposited by a mass-separated CSUP(+) ion beam. Materials Division, Electrotechnical Laboratory. Japan. ]. Appl. Phys. 55~1), pp. 188-193. Nishino, S., Y. Hazuki, H. Matsunami, and T. Tankaka. 1980. Chemical vapor deposition of single crystalline beta-SiC films on silicon substrate with sputtered SiC intermediate layer. ]. Electro chem. Soc., Vol. 127, p. 2674. Pate, B. B. 1986. The diamond surface atomic and electronic structure. Surface Sci., Vol. 165, -p. 83. Patel, A. R., and K. A. Cherian. 1981. Crystallization of diamond at atmospheric pressure. Indian J. of Pure and Applied Phys., Vol. 19, pp. 803-820. Sater, D. M., D. A. Gulino, and S. K. Rutledge. 1984. NASA Tech. Memo 83600. Spitsyn, B. V., L. L. Bouilov, and B. V. Derjaguin. 1981. Vapor growth of diamond on diamond and other surfaces. I. Cryst. Growth, Vol. 52, p. 219. Suzuki, K., H. Yasuda, and T. Inuzuka. 1987. Growth of diamond thin films by dc plasma chemical vapor deposition. Appl. Phys. Lett., Vol. 50, no. 12, p. 728. Tsuda, M., M. Nakajima, and" S. Oikawa. 1987. The importance of the positively charged surface for the epitaxial growth of diamonds at low pressure. Jpn. J. Appl. Phys., Vol. 26, p. L527.

s9 Tsuda, M., M. Nakajima, and S. Oikawa. 1986. Epitaxial growth mechanism of diamond crystal in CH/sub 4/-Hsub 2/plasma. I. Am. Chem. Soc., Vol. IDS, p. 5780. Vickery, E. 1973. Process for Epitaxial Growth of Diamonds. U.S. Patent 3, 714, 334. Weissmantel, C. 1985. Preparation, Structure and Properties of Hard Coatings, in Thin films from Free Atoms and Particles. New York: Academic Press. Wort, C. I. H., and J. C. Lewis. 1987. Erosion Resistant Coatings for Infra-red Windows in the S-14 am Band. In Electro-Optical Systems and Image Analysis for Airborne Applications. AGARD Avionics Panel Symposium, Athens, Greece, October 1987. Yarbrough, W. A., and R. Roy. 1988. Diamond and Diamondlike Materials (Extended Abstracts), p. 33. M. Gels, G. H. Johnson, and A. R. Badzian, eds., Materials Research Society.