The electronic and optical properties of diamond are reviewed, with particular emphasis on current research into the production of semiconducting devices. Although many of the properties of diamond seem, at first sight, to be ideally suited to the development of high power microwave and optoelectronic devices, there are also a number of obstacles to be overcome before diamond may be successfully exploited for device applications.

Diamond has an electric-field breakdown 20 times that of Si and GaAs, and a saturated velocity twice that of Si. This results in a predicted cut off frequency for high-power diamond transistors 40 times that of similar devices made of Si or GaAs. Boron is the only known impurity that can be used to lightly dope diamond. This p-type dopant has an activation energy of 0.3 to 0.4 eV, which results in high-resistivity material that is undesirable for devices. However, heavily boron doped diamond has a very small activation energy and a low resistivity and is of device quality. Transistors can be designed that use only undoped and heavily doped diamond. One of the steps in a device fabrication sequence is homoepitaxial diamond growth. Lightly and heavily doped homoepitaxial diamond films were characterized by scanning and transmission electron microscopy, x-ray diffraction, measurements of resistivity as a function of temperature, and secondary ion mass spectroscopy. It was found that under appropriate growth conditions these films are of device quality.

For the practical application of diamond semiconductor devices, the manufacture of planar-type devices is important. Homoepitaxial growth of diamonds would be a useful technique for making semiconducting diamond films. Homo-epitaxial growth of diamonds is achieved by microwave plasma CVD. The properties of epitaxial diamond films depend on the deposition condition and the substrate orientation. Boron doping was easily carried out by this method and boron doped films were determined to be ptype semiconductors. Properties of boron doped epitaxial films were found to be similar to those of natural IIb diamond. The Schottky junction on boron doped epitaxial films was obtained by the deposition of tungsten and aluminum. The Schottky properties of this combination were affected by the surface roughness of the epitaxial films. A field effect transistor and a light emitting device were successfully fabricated using boron doped epitaxial films and Schottky junctions.

We present self-consistent local density functional calculations of the electronic structure and bonding characteristics of an ideal diamond/Ni (001) interface and an unrelaxed diamond/BN (110) interface. At this stage the small lattice mismatches are not taken into account. Results include predictions of the band line-ups across the interfaces: the Schottky barrier height in the former case and the valence and conduction band discontinuities in the latter case.

Monocrystalline β-SiC, diamond and cubic boron nitride (c-BN) are the three ultimate semiconductors. These materials show great promise as active semiconductors for power electronics because of their wide bandgap, the existence of substitutional dopants, their high dielectric strength, their low dielectric constant, and their high thermal conductivity. To illustrate their superiority, a conventional Si.power FET is compared to a theoretical diamond FET. By assuming the diamond FET can operate hotter and at a higher voltage, it is calculated that the diamond FET should deliver up to 35 times more power-to-load than the Si FET.

Several highly disordered forms of carbon are termed “diamondlike” by virtue of their extreme hardness and significant optical gap. However, it is quite tempting to extend the analogy further and attribute these “diamondlike” properties to a high degree of local diamond order. Although an “amorphous diamond” character has been inferred from the optical properties in the 0-10 eV range, and the position of the π + σ plasmon peak in the electron energy loss spectrum, there is no direct evidence for such diamond order. However, there is a considerable body of work showing that DLC embodies a considerable degree of graphitic order which determines the optical gap and Raman spectrum. In order to determine to what extent a clearly non-diamond structure can exhibit specific “diamondlike” properties, we examine a simple structure derived from graphite. This “defected graphite” (DG) construct presumes an underlying graphitic structure which is insulating due to the strong localization of π -electrons by randomly distributed non-aromatic defects. Using specific defects suggested by experiment, the DG model reproduces a quite surprising array of observed properties of both hydrogenated and hydrogen-free “diamondlike” carbon.

Diamond can be grown as an equilibrium phase from a liquid metal solution containing carbon at high pressures and high temperatures. Diamond can also be grown as a metastable phase at subatmospheric pressures and moderate temperatures from hydrocarbon gases in the presence of atomic hydrogen. Atomic hydrogen serves several critical roles in CVD diamond growth, namely: 1) stabilization of the diamond surface, 2) reduction of the size of the critical nucleus, 3) “dissolution” of carbon in the gas, 4) production of carbon solubility minimum, 5) generation of condensable carbon radicals in the gas, 6) abstraction of hydrogen from hydrocarbons attached to surface, 7) production of vacant surface sites, 8) etching of graphite, 9) suppression of polycycic aromatic hydrocarbons. A search for substitutes for atomic hydrogen have been unsuccessful to date but several new processes that do not use atomic hydrogen are currently under study.

Bulk diamond is unstable relative to bulk graphite except at high pressure and temperature. In spite of this, well crystallized diamond has been grown using numerous CVD methods, most of which have in common the production of atomic hydrogen and hydrocarbon radicals in regimes where solid carbon is expected to be a stable product. Two fundamentally different points of view have emerged in the effort to explain why well crystallized diamond, and not g”reatcphheitde” obry v aittroemouics hcyadrbroogne, nis aot bas erarvtee dh iignh tehre sthea enx dpiearmimoenndt sa. nTd hhee fnicrset dairagmueosn dth iast gkrinapethiictael liys stable with respect to graphite. If diamond formation is kinetically controlled the deposition mechanism is critical and much debate has centered on the mechanism and species involved. Alternatively it is argued that at the growth interface, diamond surfaces are stabilized by termination with hydrogen. If this is correct, and bulk reorganization ignored, then it is shown that a global understanding of the parameters important to the growth of diamond can be obtained without detailed kinetic analyses. Thus it is argued that single crystal diamond films of arbitrarily high purity and perfection are theoretically possible by CVD in spite of the bulk instability of diamond. It is also suggested that general principles exist which might be applied to the growth of other well crystallized metastable phases - notably cubic boron nitride.

The incorporation of methane into low-pressure CVD diamond thin films has been compared to that of acetylene. 13CH4 and 12C2 H2 were used as the hydrocarbon sources in a heated-filament CVD diamond growth process at a total concentration of 0.5% hydrocarbon in 99.5% hydrogen. Results indicated that methane and/or methyl radical is the dominant carbon source for diamond growth in a hot filament reactor under steady state conditions and that acetylene is rapidly hydrogenated to methane. Results also indicated that diamond surface reactions play an important role in determining the relative methane to acetylene ratios.

In order to gain insight into diamond growth, we have calculated equilibrium geometries for several adsorbates on the hydrogenated diamond <111> surface. While the adsorption height of a single methane radical onto a dangling bond is found to be in excellent agreement with the bulk-diamond bond length, the back bonded hydrogens of adjacent adsorbed methyl radicals repel one another. In contrast, adjacent acetlyinic radicals do not repel one another but lead to the introduction of double carbon bonds, misplaced carbon atoms above the active layer and a bond length which is too short in comparison to that of bulk diamond. Our calculations on the acetylene molecule near a dangling bond indicate that the resulting adsorbate bond length is substantially too large and that the carbon atom is unlikely to be stable directly above the surface carbon atom. Of the adsorbates studied, geometrical arguments suggest that the methyl radical is likely to be the most ideal adsorbate.

The deposition of diamond from a reactive vapor phase has been dominated by the use of very dilute gas mixtures of hydrocarbon molecules, usually methane, in hydrogen. Recently there have been reports using nitrogen and oxygen containing simple organics such as amines and alcohols which seem promising. The role of atoms other than hydrogen is still not very clear. There is need for more experimental data to further our understanding in this regard. Here we give a preliminary report on our investigation of deposition from methanol-rare gas mixtures using a small microwave reactor. An important result is that good diamond can be deposited from methanol without addition of hydrogen gas and the quality of the crystallites can be influenced by Ar in the plasma.

In situ mass spectral measurements of the gas composition at the substrate surface were made during filament-assisted diamond growth. The input gases were various mixtures of CH4, O2, and H2 chosen in order to discern the effects of oxygen addition on diamond formation and growth. The gas phase chemistry was modeled as a 1-dimensional flow reactor, and the measured and calculated species mole fractions were in good agreement. The model was then used to estimate mole fractions of several atomic and radical species which could not be measured. We find that addition of O2 has only a small effect on the radical mole fractions. However, O2 can reduce the effective initial hydrocarbon mole fraction, which is important because higher quality diamond is grown at lower initial hydrocarbon mole fraction. Most importantly, perhaps, O2 addition leads to the formation of sufficient gas phase OH to remove non-diamond (pyrolytic) carbon from the film. Thus, O2 addition allows diamond films to be grown under composition and temperature conditions which otherwise would produce largely non-diamond carbon.

Diamond thin films were deposited on different substrates at low temperatures (lowest temperature∼ 300°C, estimated) in a microwave plasma enhanced chemical vapor deposition (MPCVD) system. The deposited films were amorphous carbon or diamond films depending on the different gas mixtures used. The growth rate of diamond thin films was decreased by adding oxygen to the gas mixture. The addition of oxygen to the gas mixtures was found to be important for diamond growth at low temperatures. Different concentrations of oxygen have been added into the gas mixture. Without oxygen, the deposited films were white soots and easily scratched off. Increasing the oxygen input improved the quality of the Raman peaks and increased the film transpancy. The diamond films were also characterized by scanning electron microscopy (SEM).

It is well known that diamond films synthesized from the gas phase have well defined crystal habits which are affected strongly by synthesis conditions. Though there have been many studies of the morphologies of diamond films synthesized by microwave plasma CVD [1,2,3], there have been relatively few reports on the morphologies of these films grown using new high growth rate techniques such as DC plasma jet CVD [4]. Morphology control is very important to keep flat surface, when producing thick diamond films by high growth rate techniques. In this paper we report our investigation of the morphology and growth of diamond films synthesized by DC plasma jet CVD.

The growth of diamond films on Si(001), polycrystalline Ni, Mo, Ta, and W substrates by biased controlled chemical vapor deposition is discussed. Biasing effects were examined using the Si(001) substrates. The film quality as judged by Raman spectroscopy and scanning electron microscopy depended strongly on the biasing conditions. Under low current reverse bias conditions, highly faceted cubooctahedral polycrystalline diamond growth exhibiting a single sharp Raman line at 1332 cm-1 was obtained. Transmission electron microscopy indicated that these films contained relatively low defect densities and no significant interfacial layers. Biasing into high current conditions which created a plasma resulted in multiply twinned, microcrystalline growth incorporating sp2 bonded carbon into the diamond film. Such films were found to contain very high defect densities and a relatively thick interfacial layer. An investigation of the effects of substrate material was also conducted. Films grown on Si, Ni and W exhibited the best quality. The relationship between this quality and substrate properties such as surface energy and lattice parameter is discussed.

Diamond films have been grown on single crystal Si and polycrystalline mullite substrates by hot filament chemical vapor deposition. Both substrates offer a good thermal expansion match with diamond. Gas mixture ratios of CH4 :H2 ranged from 0.1–1.0%. The remaining deposition parameters were: substrate temperature, 750°C; filament temperature, 1800°C; gas pressure, 5×103 Pa; gas flow rate, 52 seem except for the 0.1% CH4 :H2 ratio deposition in which the flow rate was 120 seem. Film thicknesses were determined from cross sectional SEM micrographs. The average growth rates on the mullite increased nearly linearly with increasing CH4 gas fraction, ranging from 0.05 μm/hr to 0.21 μm/hr. Growth rates on the Si substrates were slightly lower for identical growth conditions. Surface SEM micrographs revealed that the roughness of the films decreased with increasing CH4 gas fraction during deposition. X-ray diffraction patterns showed that crystalline grain size in the films decreased with increasing CH4 gas fraction. No preferred crystallographic orientation was seen in the diffraction patterns. Isolated particles were also grown on Si substrates under the same deposition conditions as the film growth. As the CH4 gas fraction increased, the particle growth displayed increasing rates of secondary particle nucleation resulting in a less well faceted morphology.

The technology of low pressure chemical vapor deposition (CVD) of polycrystalline diamond films has advanced substantially in recent years [1–3]. However, fundamental understanding of the chemistry and physics occurring in this CVD process is still lagging. Although the key role that H atoms play in diamond CVD has long been recognized [4–6], the identity of the gaseous diamond precursors and the mechanism by which diamond is formed are still unclear. Only recently has interest in these critical issues grown. For example, theoretical predictions and quantum mechanical calculations of gas-solid reaction paths involving CH3 and CH3 + [7] or C2H2 [8] have been reported, and the thermodynamic analyses of diamond CVD processes have been examined [9,10]. In addition, experimental results and chemical models [11–16] have been presented in attempts to support specific species as the essential precursors of diamond growth. Nevertheless, no consensus has been reached concerning the growth species and mechanism in CVD diamond processes.

The behavior of tungsten filaments heated up to 2500°C in mixtures of CH4 or C2 H2 and H2 at P = 9 and 25 Torr is analyzed. It has been found that the filament resistance R, spectral emissivity ?, power consumption, and CH4 and C2H2 partial pressures depend critically onl the filament temperature as well as the reactant CH4/ H2 or C2H2/ H2 fraction. We interpret these results on the basis of a thermodynamic model which predicts that transitions between deposition and etching of solid carbon occur on the filament surface at high temperatures. The implications of these results for diamond CVD are discussed. Evidence for gas phase reactions is also presen ted.

Diamond films have been deposited on silicon wafers by scanning an oxyacetylene flame against water-cooled silicon substrates. High quality diamond films about 20 um thick can be achieved by scanning the flame at a speed of about 0.2 mm per minute. The as-deposited films exhibit high electrical resistivity exceeding 1014Ωcm. Current voltage measurement shows the trap-dominated current conduction characteristics similar to that for dehydrogenated diamond films deposited by methane/hydrogen plasmas.

Thin carbon films were prepared by RF diode sputtering of a graphite target in a mixed nitrogen/argon plasma. A series of carbon films were deposited as a function of nitrogen partial pressure. We have observed a systematic variation of the properties of the carbon films with an increase of the nitrogen partial pressure. AES, XPS and energy gap studies showed that nitrogen will enhance the diamond sp3 bonding. Consistent with this the optical energy gap of our C:N films shows an increase from 1.1 eV to 1.4 eV using, respectively, 25 to 100 % nitrogen plasma. The mechanical properties also are shown to be enhanced for certain applications.