An understanding of fluid dynamics is essential in applications ranging from aerodynamic modeling to heterogeneous catalytic reactions and chemical engineering. However, complex flow dynamics are difficult to simulate due not only to the high computational demands of modeling, but also because few models adequately reproduce experimental observation. In this respect direct experimental methods for studying flow is the best alternative approach, but methods for visualizing multicomponent flow are rare, primarily because of the difficulty in probing past the surface layer and in distinguishing different chemical species. NMR spectroscopy and imaging can provide a solution for both problems because the radio frequency waves used in the method penetrate easily inside conventionally used materials, and the resonance frequency of certain nuclei depends on the molecule in which it belongs and even on the environment of the molecule. Knowledge of flow paths through a material can give valuable information of its structural and chemical properties.

Microfluidic technology is a critical component of so-called “lab on a chip” devices, where traditional chemistry is conducted on length scales that allow for increased control over the dynamics of fluid reactants. Its realm of application is well beyond chemistry, extending, for instance, into the kinetics of fast biological processes such as protein folding and fundamental studies of fluid physics and control. Nuclear Magnetic Resonance is a generic, non-invasive probe of chemistry and fluid motion in microfluidic devices. However, its application is limited by poor sensitivity and limited temporal resolution. We overcome these limitations by employing remotely detected NMR, in which spatial or other information is imprinted by magnetic field gradients over the region of the entire microfluidic device and then carried, by the fluid itself, to an optimized microcoil detector. The resulting amplification of the encoded signal can be used to probe microfluidic flows with unprecedented spatial and temporal resolution.

Using remotely-detected time of flight MRI, we have characterized fluid flow in capillaries and constricted microfluidic channels with gradient moment encoding methods with spatial resolution of ~15 mm and temporal resolution of better than 20 ms. The results reveal correlations between the NMR time-of-flight parameters and local velocity distributions on sub-voxel length scales, yielding information unavailable from other techniques. We are currently applying these methods to the study of flow in porous materials, chromatography, and in chemical reactions including combustion in microchannels. Finally, we have incorporated subsampling and sparse image reconstruction techniques to dramatically reduce the time required to acquire spectra of higher dimensionality.

In the Pines lab, gas flow have been studied by various NMR flow imaging methods, which makes it possible to directly visualize a flow process non-invasively and quantitatively in a three-dimensional fashion. The hyperpolarization of noble gases overcome the low sensitivity of NMR and allows spectroscopy and imaging of the dilute gas phase (pure or within materials) within a reasonably short experimental time. An example of gas flow visualization is the velocity encoded image of the flow around a sphere.

A novel remote detection technique developed in the Pines lab provides an important advance for flow imaging. When studying flow through porous materials, the filling factor of fluid inside the NMR coil is typically low leading to poor sensitivity. Remote detection NMR capitalizes on the mobility of the fluid by separating the encoding and detection steps which enables a measurement of the encoded NMR signal of the fluid as it leaves the sample using a smaller and more sensitive coil with a better filling factor effectively overcoming this limitation. An inherent property of the remote detection method is that it gives information of the time-of-flight of a fluid encoded in a certain region of the sample to the detection coil. By applying a phase modulating or slice selective pulse sequence in the encoding it is possible to obtain time-of-flight snapshot images of the fluid flow through the sample. These images elucidate the flow paths through the sample, from which structural details, such as pore connectivity of the sample, can be deduced. The extreme sensitivity of the chemical shift of ¹²⁹Xe isotope to its local environment makes xenon gas especially suitable as a fluid for remote detection flow imaging. In this way the chemical shift can act as a fourth dimension in the flow images. This information can reveal, for example, multiphase fluid transport in the sample as shown in recent work on the flow of xenon gas through a porous aerogel sample.

Watch movie of xenon gas flow

Applicability of remote detection flow imaging method for studying of real systems was showed by monitoring the flow of xenon gas through porous sandstone rock. The experiment demonstrated the feasibility of this approach for studying flow in large, opaque samples. Time-of-flight images can provide more complete information about the sample than full 3D rendering of pore space, because besides bulk structural information, such as pore volumes and pore sizes, they can reveal the microscopic structure that determines transport properties.

Remote detection flow imaging is not restricted to gas phase flow. It also is applicable to nearly any spin-active nucleus including almost all proton containing liquids. The signal-to-noise ratio obtained from liquids is usually very good without any hyperpolarization because of the large spin density. Because the diffusion of liquid molecules is much slower than gas molecules, relative time resolution of flow images is better, and therefore the experiments may reveal more local details of the sample. Although the chemical shift of most commonly used nucleus in NMR (¹H) is not very sensitive to local environment, the multiphase flow and miscibility of two or several liquids inside materials can also be traced, for example, by selectively encoding certain ¹H resonance of one liquid, which does not overlap with the resonances of other liquids. In some cases the relative short relaxation time of NMR signal of liquid molecules can lead to remarkable diminishing of encoded signal before detection. This problem can be circumvented by observing NMR signal of hyperpolarized xenon gas dissolved in the flowing liquid, because the relaxation time of dissolved xenon is very long (on the order of 100 s). Apart from the advantage of slow diffusion of xenon, the chemical shift information about the sample could be available also in this kind of experiment, because the shift of dissolved xenon depends on the pore size when the size is small enough. Therefore, this experiment can reveal details of multiphase fluid transport, as in the case of gas flow through silica aerogel, with more detailed local information.