ON-LINE SAMPLING AND GAS CHROMATOGRAPHIC METHOD FOR A MICRO-SCALEGAS-LIQUID OXIDATION REACTOR

摘要

The development and application of a novel 5-cc micro reactor system forrngas-liquid hydrocarbon oxidation kinetics studies under commercially-relevantrnconditions is described. A unique system feature is the use of a novel automatedrnmethod for direct sampling and analysis of the liquid phase. This is accomplished byrnmeans of a custom-designed sampling valve that is integrated into the micro reactorrnbody and connected to a GC system for direct injection of the vaporized liquidrnsample via heated capillary transfer tubing. The advantages of this approach overrnconventional manual sampling include: (1) the effect of sample size on therninstantaneous reactor volume is negligible; (2) post-reaction of the sample once it exitsrnthe reactor is minimized; (3) the sample statistics are improved over conventionalrnsampling methods; (4) light-boiling components and dissolved gases are directlyrncaptured for subsequent analysis; (5) reactor operation is more safe thanrncommercially-available laboratory reactors owing to the small volume (5 cc or less);rnand (6) unattended operation with precise intermittent sampling of the liquid phase isrnpossible.rnThe sampling concept was first demonstrated using a custom-designed testrnvessel whose geometry mimicked a 5 cc stirred gas-liquid microreactor. Two modes ofrnliquid sampling were evaluated: (1) flashing of the liquid sample into a calibrated looprnof a VICI? Valco six-port gas sampling valve; (2) flashing of the liquid sample directlyrninto the injection port of a HP 5890 Series II gas chromatograph. A variation of thernsecond mode involved the addition of a ten-foot heated section of capillary tubingrnbetween the exit of the vessel and the GC injector. The test samples consisted of anrnaqueous mixture of ca. 1 % each of C1 to C4 alcohols, and a crude reaction productrnobtained from a commercial-scale cyclohexane oxidation reactor.rnIt was found that the ability to fill the Valco valve sampling loop wasrndependent upon the pressure developed by the flashing liquid. For a fixed vesselrnpressure, the amount of vapor in the loop reaches a maximum and then decreases asrnthe pulse valve duration increases. This occurs because the pressure created by thernflashing liquid greatly reduces the amount of sample that can exit the valve. In therncase of the direct injection methods, this effect was not observed. Hence, the total GC area counts were reproducible within 0.1 % for a fixed value of the pulse valvernduration. The variation that uses the transfer line is particularly useful for applicationsrnwhere the reactor must be located some distance from the GC system. Other detailsrnon the valve design and test results are also provided.