|The Growth Chemistry of CVD Diamond Films
Robert W. Shaw1, C. S. Feigerle2, and James Chenault2
1. Analytical Spectroscopy and Microinstrumentation Group, ORNL
Growth of diamond films from the vapor phase is a relatively uncomplicated synthesis. However, an understanding of the chemistry - both gas phase and surface - has proven harder to clarify. The recipe for chemical vapor deposition (CVD) diamond is as follows: decompose a mixture of 0.5% methane in hydrogen at 50-100 Torr total pressure at a hot filament (about 2000 C) in the presence of a substrate heated to approximately 1000 C. Polycrystalline diamond will grow on the substrate at about 1 mm/hr. While these parameters are easy to achieve, it is interesting that the acceptable operational window (i.e., methane concentration, filament temperature, substrate temperature) over which quality diamond films may be grown is somewhat narrow. We are attempting to understand the mechanism of this chemical process by developing spectroscopic tools to monitor the concentrations of intermediate species, e.g., atomic hydrogen and methyl radical. With such an understanding in place, it may be possible to grow diamond films under less hostile conditions (opening the way for a wider variety of substrates) or under similar conditions, but as higher growth rates.
For gas phase studies, we have selected resonantly enhanced multi-photon ionization (REMPI)/time-of-flight mass spectrometry as our detection approach. We extract reacting gases from the near surface region above the substrate via a 200 mm diameter orifice in the substrate. This gas expands as a free jet into a vacuum chamber that is differentially pumped to 1 mTorr pressure. There, because the pressure is sufficiently low, there are no bimolecular collisions, so reactive species survive. Gas from the flow is then sampled using a conical skimmer and admitted into the ionization region of the mass spectrometer. There the neutrals can be ionized by crossing them either with a beam of electrons (for survey mass spectra) or with a focused UV laser beam (for molecule specific detection). This method exhibits double selectivity; because we utilize both the optical spectroscopic signature of the neutral (by scanning the laser wavelength) and the product ion mass (from the observed ion time of flight), we have complete certainty concerning the identity of the species monitored. We have followed the concentration of hydrogen atom and methyl radical by this means as we varied operational parameters such as filament and substrate temperature. An interesting observation we have made is a decline in the atomic hydrogen concentration in the near surface gases as we increase the substrate temperature; this is most likely due to abstraction of surface-bound hydrogen atoms to form gaseous molecular hydrogen - an activated process that yields surface growth sites. We have recently begun a study of diamond growth using chlorinated hydrocarbons as precursors, instead of methane. The C-Cl bond is more easily broken than C-H, and new chemical growth mechanisms may be observed at lower operating temperatures.
We are also investigating the use of surface non-linear optics as an in situ probe of surface chemistry. Using a technique termed surface sum frequency generation, we will attempt to record the vibrational spectra of species at the film growing surface. Before we can conduct those experiments, we must first develop a tunable, picosecond (10-12 second) pulse, mid-infrared laser to use as a source. That development is now underway. In a similar fashion to the gas phase techniques discussed above, we will then be able to vary a growth parameter while monitoring an induced chemical change, in this case, surface species alteration at the growing surface.
Additionally, we have measured the metal impurity levels in our diamond films, due to metal evaporation from the hot filament. Neutron activation analysis (NAA) was employed for these determinations. We have investigated tungsten, tantalum, and rhenium filaments. The former yields the lowest metal impurity contamination (few parts per million-mass), while the latter, the highest level.