System: packed gold reactor as used in Series 7 and 8. Temperature SS8°C., total pressure lS psia.,. argon diluent, the trace oxygen level was below 2 ppm. For the next shoot, data were obtained after a further 10 h treatment with 8% oxygen gas. This tube was not exposed to hot oxygen but probably contained carbonaceous deposits from Series 6. butane/mole product) in the effluent.
After batch 15, the reactor was purged with oxygen for 9 hours, then with hydrogen for 7 minutes, then purged with argon for 4 days before batch 16 was performed.
Run 44a was performed under different reactor conditions to investigate the start-up behavior once an apparent steady state was reached. An elementary calculation showed that the conversions in runs 18a and 19 could not be satisfactorily accounted for by the differences in the average temperatures of the runs. The instabilities of Runs 93 to 95b were taken as an indication of the formation of carbonaceous deposits in the reactor.
Hydrogen formation in an acid-treated reactor compared to the yield calculated from the hydrogen mass balance expression. The calculation used the methane yields as determined using both chromatographs to link the hydrogen determinations to the hydrocarbon yields.
Series 13
APPENDIX 2
The purpose of the following discussion is to present the steady-state solution for the above system. The validity of the steady-state treatment applied to the pyrolysis of n-butane has been demonstrated by Blakemore and Corcoran (41). They used the mechanism proposed by Wang (40) and numerically solved the differential equations for the formation of the products.
Perhaps the estimates of the individual rate constants are sufficiently in error to accommodate the contributions of the half-order and 1.5-order terms. Because reaction 7 is usually not in its first-order region (1), the contribution of the half-order term is almost certainly overestimated. An elementary calculation using steady-state concentrations of free radicals shows that the above possible termination reactions would be fully two orders of magnitude slower than termination by two ethyl radicals.
The difference between the predictions of the steady-state model and the experimental results can be easily resolved if it is assumed that the estimates of k. There were only two estimates of the rate of reaction 4; one was from a complex photolysis experiment (21), the other was calculated from a study of the pyrolysis of n-butane (1) and therefore does not have a satisfactory cross-check on the. The large contribution of the half-order term may provide a partial explanation of the "breakdown".
A brief overview of the corrections applied to the plug-flow model is given. In any system undergoing chemical changes, the chemical kinetic behavior cannot be strictly separated from the heat and mass transfer characteristics of the system. One experimental study used a modification of the plug-flow model that takes into account longitudinal diffusion (5).
The correction factor was a function of the magnitude of the response and of the order of the response. These solutions were summarized as a series of graphs giving the correction factor for non-plug flow behavior in terms of dimensionless groups characterizing the heat of reaction, the rate of reaction and also the extent of reaction. Input and output effects associated with the injection of the inert gas at the ends of the reactor will be neglected.
APPENDIX 4
The data for the various hydrocarbons tested show an approximately linear relationship as a function of polarity. The packaging was then dried for several hours at a suitable temperature before a final weighing which checked the accuracy of the preparation. An aspirator line was connected to the plugged end of the column and the packing was then slowly poured into the column.
When the packing was complete, the open end of the column was plugged and Swagelock fittings were added. The relative molar response for paraffins was directly related to the carbon number of the sample. Air was used as the purge gas because its use was thought to extend the range of linearity of the detector (13).
To determine the relative response factors, multicomponent gases of known composition were prepared. These mixtures could be prepared with an accuracy of + 1.5%, with the main source of error being due to the accuracy of the pressure measurements (+ 0.02 cm Hg). The above findings indicate that the detector noise level was a function of signal strength.
These data are the arithmetic means of the relative molar response factors as determined in three independent tests. Detection with thermal conductivity cells was necessary because of the negligible response of hydrogen in conventional ionization detectors. The response of the thermal conductivity detector to hydrogen in a helium carrier stream was very small.
The relative reactions of hydrogen and methane were sensitive to changes in retention times. As a result of the gradual contamination of the column, the hydrogen to methane ratio in the sample would therefore be underestimated by up to 50%. Using the peak areas instead of peak heights as the basis for the calibrations would improve long-term stability.
APPENDIX 5
A small scrubber was installed upstream of the cell where the liquid in the cell was dilute sodium hydroxide solution. A major error in the calibration of the cell was due to current leakage in the electrolysis unit. A major limitation of the system was the high degree of pulsation in the compressed gas.
