UCLA

More than 85% of all chemical industry products are made using catalysts, the overwhelming majority of which are heterogeneous catalysts that function at the gas-solid interface. Consequently, much effort is invested in optimizing the design of catalytic reactors, usually by modeling the coupling between heat transfer, fluid dynamics, and surface reaction kinetics. The complexity involved requires a calibration of model approximations against experimental observations, with temperature maps being particularly valuable because temperature control is often essential for optimal operation and because temperature gradients contain information about the energetics of a reaction. However, it is challenging to probe the behavior of a gas inside a reactor without disturbing its flow, particularly when trying also to map the physical parameters and gradients that dictate heat flow, mass flow, and catalytic efficiency. Although optical techniques and sensors have been used for that purpose, the former perform poorly in opaque media and the latter perturb the flow. NMR thermometry can measure temperature non- invasively, but traditional approaches applied to gases produce signals that depend only weakly on temperature, are rapidly attenuated by diffusion, or require contrast agents that may interfere with reactions. In this dissertation, we present a new NMR thermometry technique that circumvents these problems by exploiting the inverse relationship between NMR linewidths and temperature caused by motional averaging in a weak magnetic field gradient.

The motional averaging behavior of gases is fundamentally different from that of liquids, which can be explained using a more detailed theoretical description of the dephasing function that accounts for position autocorrelation effects. The traditional view of nuclear-spin decoherence in a field gradient predicts that in a fluid, NMR linewidth should increase with temperature; however, in gases we observed the opposite behavior. Furthermore, in an inhomogeneous field, the nuclear free induction decay signal exhibits fundamentally different time dependence between gases and liquids. The Carr-Purcell-Meiboom-Gill (CPMG) experiment has been used for decades to measure nuclear-spin transverse (T2) relaxation times. In the presence of magnetic-field inhomogeneities, the limit of short interpulse spacings yields the intrinsic T2 time. In gases, CPMG unexpectedly fails to eliminate the inhomogeneous broadening due to the non-Fickian nature of the motional averaging.

We exploit the motional averaging behavior of gases by non-invasively mapping gas temperatures during the hydrogenation of propylene in reactors packed with metal nanoparticles and metal-organic framework catalysts, with measurement errors of less than 4% of the absolute temperature. These results establish our technique as a non-invasive tool for locating hot and cold spots in catalyst-packed gas-solid reactors, with unprecedented capabilities for testing the approximations used in reactor modeling.