Assessing Intrusion by the Capillary During Spatially Resolved Mass Spectrometry

Michael  Harold, University of Houston

The impact of a  probe of capillary inlet spatially-resolved mass spectrometry (SpaciMS) on the reactant conversion during a catalytic reaction in a  monolith reactor is systematically examined. The accuracy of the concentration measurements were checked during propylene oxidation  by excess O2 over Pt/Al2O3 washcoated monoliths having a range of channel diameters (100 to 600 cells per square inch, cpsi). By comparing the spatial concentration profiles of propylene sampled with capillary probes of two different diameters (170 and 363 µm) with optical fiber temperature measurement using optical frequency domain reflectometry (OFDR) in an adjacent channel, we conclude that the Spaci-MS provides accurate measurement when the limiting propylene is depleted in the downstream section of the reactor (at a position equal or larger than about 0.6 of the  channel  length). For certain conditions the suction by the probe compensates for the blockage effect so that the concentration profile measured by the larger probe diameter (363 µm) is similar to that measured by the smaller capillary. The experiments reveal that the axial position of the probe in the channel does not influence the flow profile in the 100 cpsi monolith channel, nor does it affect the amount of flow deflected to  surrounding channels for the 600 cpsi monolith.  When the propylene is depleted further upstream the results are impacted by transverse concentration gradients, confirmed in experiments in which the reactant channel concentration near the monolith exit is compared to the effluent mixing cup concentration.  Under such conditions the accuracy of the concentration measurements cannot be resolved.  The difference between the location of propylene depletion and local temperature maximum, DZ, provides a useful metric for capturing the collective impacts of flow deflection and transverse gradients.  The complexity of the flow and reaction suggests that at least three dimensionless groups are needed to classify the different behaviors.