Detailed chemical kinetics models of complex reactive systems are large-scale mathematical models that combine lengthy lists of elementary chemical reactions (the reaction mechanism) with transport for an accurate description of process chemistry and rates of transformation. The models are suitable for many purposes; for example, they are used to describe reactive flows such as those encountered in the chemical process industry, or to better understand combustion or environmental chemistry.

Figure 1: Chloromethoxy radical

Supercomputing Institute Associate Fellow Robert W. Carr, who is Professor of Chemical Engineering and Materials Science, and his group of researchers are focusing on the development of models for chemical vapor deposition and for atmospheric chemistry. The use of supercomputers, the development of efficient numerical algorithms, and advances in the kinetics of elementary reactions and in computational chemistry, all contribute to progress in the development of detailed chemical kinetics models.

Chemical vapor deposition (CVD) is a process for depositing thin films on solid surfaces by flowing a molecular source of the elements to be deposited over a heated substrate in a cold-walled reactor. The source is commonly a vapor and the substrate temperature is frequently about 1000 K. In the hot region, thermal decomposition of the source species occurs both on the surface and in the adjacent gas phase. Reactive intermediates adsorb on the surface where reactions leading to film growth take place. Accurate description of this chemistry requires tens to hundreds of elementary chemical reactions. The development of mathematical models for CVD reactors has advanced to the point where it is possible, with supercomputers, to compute the flow and temperature fields for almost any reactor configuration of interest. However, computation of film growth rates and material properties is still in an embryonic state because of gaps in knowledge of the kinetics and mechanisms of the gas-phase and surface chemistry occurring during deposition.

Figure 2: Transition state of Cl elimination

Mechanism development consists of a mix of experimentation, literature searching, and application of theory. When literature data cannot be found, key reactions must be studied experimentally to obtain accurate values of rate coefficients. Other reactions can be investigated by the methods of computational chemistry, and some relatively unimportant ones can be treated by empirical means. One of the reactions in the GaN deposition system the Carr group recently investigated is the formation of the Lewis acid-base pair between trimethylgallium and ammonia. This reaction is thought to inhibit GaN deposition, but there was a scarcity of reliable data. Measuring the equilibrium constant as a function of temperature, the group found the enthalpy of reaction to be -15.2 kcal mol-1.

Figure 3: Transition state of HCl elimination

Ab initio quantum chemistry calculations on the reactant and the product yielded an enthalpy of the reaction of -15.9 kcal mol-1, in good agreement with experiment. With the energetics of the reaction firmly established, the researchers have been using the theory of unimolecular reactions to get an estimate of the rate coefficient for unimolecular dissociation of the adduct. Application of microscopic reversibility then gives an estimate of the rate coefficient for the association of trimethylgallium and ammonia. These ab initio calculations revealed the appropriate levels of theory for reliable estimation of the energetics of organogallium compounds. This experience has allowed the group to calculate good estimates of the rate coefficients for successive unimolecular elimination of each of the three methyl radicals from trimethylgallium. The decomposition of trimethylgallium may be the rate limiting step in GaN deposition, in which case it will be important to know the decomposition rate completely.

Another area of study is the atmospheric chemistry of halogenated species, including replacements for the ozone-depleting chlorofluorocarbons. These compounds are present in the atmosphere in trace amounts, and can be removed through a multistep oxidative degradation process in which free radical intermediates play a role. Among these are halogenated alkoxy radicals, species whose reactivity is not well understood, and which are frequently a stumbling block in attempting to understand the oxidative degradation chemistry. The group has been using time-resolved mass spectrometry to investigate the kinetics of some halogenated alkoxy radicals. The experimental method is limited to studies at pressures of a few torr to 40 or 50 torr, and temperatures that typically are higher than the approximately 200 K minimum that is found at the troposphere-stratosphere boundary. The radicals being studied are unstable, undergoing unimolecular decomposition with pressure and temperature dependent rate coefficients. To determine rate coefficients over the full range of atmospheric pressure and temperature, ab initio calculations of the decomposition energetics are being performed, the results of which can be used in conjunction with the theory of unimolecular reactions to calculate the rate coefficients. Matching the calculations with experimental data ensures reliable values of the rate coefficients over the entire temperature and pressure range. Figures 1Ð3 show transition states for Cl-atom and HCl elimination from the chloromethoxy radical. The Carr group recently computed these structures using G2 and G2(MP2) theories on the IBM SP.