3.1. Oxidation
The carbon mass flux was determined by dividing the mass loss by the exposure time and initial surface area. Results for atomic and molecular oxygen are plotted in Fig. 3 which show a strong dependence on surface temperature. Both trends increase with temperature up to a maximum beyond which the oxidation rate begins to decline. This behavior is consistent with the previous studies at lower pressures [4, 8, 18, 19] although the temperature associated with the peak shifts higher as the pressure increases. The peak reaction efficiency of CO in molecular beam experiments occurs at approximately 1100 K [24] and in these experiments the peak carbon mass flux occurs between 1500 and 1600 K. At all temperatures, the atomic oxygen is found to be more reactive than the molecular oxygen.
These results are compared to the work by Zhang et al. on graphite oxidized by atomic oxygen from a microwave discharge at 0.3 kPa [18]. Both sets show a temperature dependence with an oxidation rate that reaches a maximum and then declines as the temperature continues to increase. The peak oxidation rate occurs at a different temperature for the two sets of data, between 1500 K and 1600 K for the present study and between 1200 K and 1300 K for Zhang et al. The mass loss rates are an order of magnitude higher due to the higher pressure.
3.2. Nitridation
The atomic oxygen results require careful interpretation to properly account for the presence of other chemical species. The NO titration provided the information necessary to conduct experiments as close to the stoichiometric balance point of Eq. 1 as possible. However, leftover atomic nitrogen and NO are present in unknown concentrations at the sample surface as well as molecular oxygen that recombined along the length of the quartz tube. The degree to which these intruding species influence the results of the atomic oxygen tests is investigated here by measuring the reactivity of each species with graphite. The nitridation investigation consisted of four conditions: pure molecular nitrogen, nitrogen dissociated by the argon plasma, pure NO, and NO dissociated by the argon plasma. Argon remains in each test case as a carrier gas to maintain constant total pressure and reactant concentration. The mole fractions for each of these test conditions are presented in Table 1, along with the atomic and molecular oxygen conditions described in Section 2 which are listed as #1 and #2. For Condition #1, mole fractions are calculated assuming the case of an ideal stoichiometric balance of Eq. 1 such that there is no molecular oxygen, atomic nitrogen, or NO at the sample surface. The concentration of atomic nitrogen for Condition #4 was measured by NO titration at the sample surface to account for the recombination of nitrogen atoms along the quartz tube. Condition #6 is the NO dissociated by the argon plasma and the mole fractions are calculated assuming complete dissociation. This assumption is corroborated by the RGA measurements which detected oxygen and nitrogen during the test but no traceable amounts of NO. The procedure was identical to the oxygen tests: the chamber was adequately evacuated before filling with inert argon, then the graphite was heated to its steady state temperature before introducing the reactive agent. The resulting carbon mass flux data can be seen in Fig. 4.
Pure molecular nitrogen proved to be totally inert and resulted in no measurable mass loss for each temperature tested. The presence of molecular nitrogen in the atomic oxygen tests can therefore be safely neglected when evaluating the reactivity. NO in its non-dissociated form was also non-reactive and any NO molecule reaching the sample in the atomic oxygen tests would not contribute to the total mass loss. Atomic nitrogen was as nonreactive as molecular nitrogen at low temperatures and only at temperatures above 1873 K was it more reactive than NO. Zhang et al. [18] reports increasing reaction efficiency of graphite nitridation between 873 and 1373 K but provided no data at higher temperatures for comparison.
Condition #6 was the NO dissociated by the argon plasma. At low and intermediate temperatures these tests were highly comparable in reactivity to the atomic oxygen tests. At higher temperatures the two data sets diverge in trend. At 1673 K and 1873 K the NO plasma is more reactive than the atomic oxygen. At 1973 K, both conditions drastically decrease, but Condition #6 falls further. To understand this behavior consider the ideal case for Condition #1 in which the gas flow is in stoichiometric equilibrium at the instant it reaches the sample surface without any molecular oxygen, atomic nitrogen, or NO present. Consider the degree by which the real tests depart from this ideal scenario. Each molecule of NO that does not react according to Eq. 1 is dissociated by the plasma. Instead of producing one N2 molecule and one O-atom, it produces one N-atom and one O-atom. It also leaves one more N-atom that it did not react with. The result is N + N + O instead of N2 + O. In test Condition #6, in which the NO is dissociated in the argon plasma, the best it can do is produce N + O. At the lower and intermediate temperatures this extra N atom makes no difference and the mass loss for the two tests is the same. At the highest temperature, the atomic nitrogen is more reactive and the extra N atom in the first test case causes greater mass loss than in the case with just NO.
The presence of nitrogen in the atomic oxygen tests could block available surface sites from the O-atoms. Molecular beam experiments conducted with a mixture of N and O-atoms at surface temperatures between 1100 K and 1700 K showed no noticeable reduction in O-atom reactivity [13]. The experiments indicated that the presence of N atoms in small amounts, mole fractions between 0.02–0.08, actually caused a slight increase in the O-atom reactivity. This low pressure result is in contrast to the high pressure carbon mass flux data reported in Fig. 4 suggesting the assumption made by Zhang et al. [18] that the two species can operate independently does not hold as pressure increases.
In addition to the carbon mass flux results it is also helpful in the study of chemical kinetics and the development of computational models to compare reaction probabilities. One model for calculating the probability of reaction is the Hertz-Knudsen equation which can be used to relate the carbon atom flux away from the surface to the oxygen atom flux to the surface:
$$ϵ=\frac{\dot{Z}}{\frac{1}{4}\frac{{p}_{i}}{{k}_{b}T}\sqrt{\frac{8{k}_{b}T}{\pi {m}_{i}}}}$$
2
\(\dot{Z}\) is the carbon atom flux, pi is the partial pressure of the species, kb is Boltzmann’s constant, T is the gas temperature, and mi is the atomic mass of the species. Figure 5 shows the reaction probability of molecular oxygen against the reciprocal of temperature. A non-linear trend is evidence of non-simple Arrhenius behavior. The results of the present study are compared to the existing body of work in this field as well as the 1976 model developed by Park [3]. Results are in good agreement with the results of Rosner and Allendorf [19]. In the case of atomic oxygen, calculations for reaction probability are an order of magnitude lower than those reported by Rosner and Allendorf. Most likely the flow reactor is operating in a diffusion limited regime in which the true reaction kinetics of atomic oxygen cannot be extracted without properly accounting for mass transport effects [4, 25]. This was not an issue with molecular oxygen because its lower reactivity compared to atomic oxygen shifted the experiments into a kinetically limited regime. Damkohler numbers calculated for atomic and molecular oxygen support this conclusion and a thorough investigation of diffusion and reaction limited experiments in this flow reactor will be provided in a future publication. The development of a mass transport model as well as a more precise measurement of the atomic oxygen concentration by absorption spectroscopy will allow for the determination of more accurate reaction probabilities.
3.3. Mass Spectroscopy Results
The dominant chemical products in the oxidation of solid carbon are CO2 for temperatures below 973 K and CO at higher temperatures [15]. The RGA was used to measure both species. A disadvantage of the experimental system as it was designed was that CO and N2 share the same molecular mass of 28 g/mol making them indistinguishable to the RGA. However, since the flow rate of N2 does not change throughout the experiment the m/z = 28 signal could still be used to extract CO data from the total measurement. Subtracting the RGA signal after oxidation (N2 only) from the signal during oxidation (N2 and CO) resulted in the relative intensity of CO alone. The left plot of Fig. 6 shows the RGA results for the tests with atomic oxygen, a1-a8 in Table 2. The right plot of Fig. 6 is the result of a single oxidation test by O2 during which the temperature was incrementally increased from 973 to 2073 K. Both sets of data show the CO signal increasing with temperature and the CO2 signal decreasing as expected. The precise concentration of products cannot be determined without proper calibration of the RGA. However, based on the ratio of the two signals and the measured mass loss of the graphite (∆), the mass of CO and CO2 can be inferred and are shown in Fig. 7.
3.4. Electron Microscope Results
SEM characterization of the graphite samples after oxidation showed evidence of roughening and in some cases the formation of pits, as shown in Fig. 8.
The virgin (untreated) material is shown for comparison.
The observed pitting appears in two distinct sizes. The samples that reacted with nitrogen and with oxygen show pits at the level of 1–10 microns that grew out of the nucleation of a single intermolecular reaction. The initial reaction site left more exposed surface area for new reactions to occur and the area became preferentially eroded. This behavior has been reported before for graphite [18] and FiberForm [15, 17]. The samples tested with molecular oxygen show a different sort of pitting that is on a much larger scale than anything reported before. The craters are on the order of 10–100 microns. It is possible that these are the result of impurities in the graphite that cause highly localized heat conduction which effectively burn holes into the surrounding carbon material. This particular graphite grade contains over 100 ppm of silicon, calcium, and iron. If it is caused by a material property, the presence of the craters would be observed with other test gases as well. Samples tested with the NO plasma exhibit craters but not as frequently as the molecular oxygen. This may be due to the presence of nitrogen inhibiting one or more forms of the oxidation process. Experiments with higher quality graphite grades and fewer impurities are required to further investigate this behavior. We also note that the craters form more readily at higher temperatures which is consistent with the overall trend of reactivity and increased mass flux.