Reports: DNI651521-DNI6: Testing Models of Multi-Component Diffusion and Spin Relaxation in Porous Media under Radio Frequency Excitation in NMR

Jamie D. Walls, PhD, University of Miami

Our current research has begun to characterize the diffusion and aggregation processes (AIM IIA) for a "simple" dye, sunset yellow. We have also developed algorithms to efficiently search for relaxation selective pulses (AIM IIB), which can be used to constrain aggregation models used to model the data from AIM IIA.

Design of Relaxation Selective Pulses (AIM IIB):

Research Goal: In porous media, the relaxation times, T2 (transverse) and longitudinal (T1), are inversely proportional to pore size. The goal of AIM IIB was to design radiofrequency (RF) pulses that are able to selectively null the magnetization for a particular choice of relaxation parameters and/or values of the diffusion coefficient. Such pulses could be used for studying diffusion between different pore sizes using a variety of 3D NMR experiments (e.g., diffusion/diffusion, relaxation/relaxation, and diffusion/relaxation correlation experiments).

Progress to date: Currently, our group has developed algorithms that can be used to design RF pulses to null the magnetization for systems with a particular relaxation time, T2 = T2sel and with T1=aT2 (where a ≥1/2). We employed the GRAPE (GRadient Ascent Pulse Engineering) algorithm to perform the optimization of an initial input pulse for zeroing the magnetization at T2=T2sel. This was accomplished by finding the RF that minimized F = Sinhomogenity S T2 ABS[ MMtarget] , where M is the magnetization after application of the RF pulse, Mtarget is the target magnetization profile (where Mtarget = 0 for T2= T2sel), and where we have summed over field inhomogeneities (RF and B0) in order to find pulses robust to such field inhomogeneities. We found that our ability to find the optimal pulse was quite sensitive to the initial input pulse. We did find, however, that optimization of a modified soliton pulses worked well for finding RF pulses that generated an M close to Mtarget and that were robust to field inhomogeneities [Fig. 1]. Other inputs, such as bipolar rectangular pulses and adiabatic pulses also generated similar results but took much longer to converge to Mtarget. This work will soon be submitted for publication.

Fig. 1. (A) Initial 20ms Soliton Pulse (left) along (right) with the magnetization profile after application of the soliton pulse vs. T2 (ms). The theoretical z (purple) and transverse (green) magnetizations, and Mtarget (blue, T2sel=15 ms) profiles are plotted (right) along with the experimental results (*). There was significant deviation of M from Mtarget. (B) The optimized pulse found from our algorithms. The magnetization profile from the optimized pulse was close to Mtarget.

Text Box: Fig. 1. (A) Initial 20ms Soliton Pulse (left) along (right) with the magnetization profile after application of the soliton pulse vs. T2 (ms). The theoretical z (purple) and transverse (green) magnetizations, and Mtarget (blue, T2sel=15 ms) profiles are plotted (right) along with the experimental results (*). There was significant deviation of M from Mtarget. (B) The optimized pulse found from our algorithms. The magnetization profile from the optimized pulse was close to Mtarget.

Future Directions: Our group will employ the current pulse sequence algorithms to find pulses that are T2 selective while being robust to T1 (i.e., independent of a). We will also incorporate diffusion into our algorithms along in order to determine the necessary time-dependent pulsed field gradients that can be employed to selectively zero magnetization based not only upon T1 and T2 but also upon the value of the diffusion coefficients.

Using RF pulses to constrain spectral distribution functions in porous media (AIM IIA):

Research Goal: The relaxation and diffusion distribution and joint distribution functions can provide valuable information about the chemical composition of oil, and also the porosity and pore size distribution. Information about these distribution functions can be inferred from a variety of NMR experiments. The goal of AIM IIA was to determine a set of optimal pulse sequences that could be used to constrain the relaxation and diffusion distribution functions.

Progress to date: As an initial step, our group investigated the aggregation of a well-studied dye, sunset yellow; this system was chosen as a "simple" test case for developing pulse sequences to constrain aggregation models along with applying diffusion selective pulse sequences developed in AIM IIB. Like other organic dye molecules, sunset yellow forms aggregates in all phases, even at concentrations as low as 50 μM. As the aggregate size distribution changes, the observed spectral properties of the dye also change. Figure 2(A) shows the T1 and T2 relaxation times for sunset yellow as a function of concentration. As the concentration increased from 0.0025M (0.1% wt.) to 0.5M (22% wt.), the T1 relaxation increased while T2 relaxation decreased indicating changes in molecular motion due to increasing size of the aggregates.

Besides studying relaxation properties, the observed 1H spectrum was also affected by the aggregate distribution. The protons labeled in Figure 1(B) [solid arrows] were observed to have the greatest change in chemical shift with increasing dye concentration [Figure 2(C)], most likely due to their location next to a naphthalene ring above and below in an aggregate formation; similar results were obtained by Edward et al. for the proton denoted by the hollow arrow in Fig. 2(B). Figure 2(C) illustrates the change in the 1H-NMR spectra with changing dye concentration from 0.0003 M (0.01% wt) (bottom) to 0.75 M (33% wt.). The line widths of the peaks ranged from 3 Hz at the lowest concentration up to 30 Hz. Studies of an aggregation inhibitor, such as the solvent DMSO, and aggregation promoters, such as sodium, magnesium and potassium ions, were also performed.

Figure 2. (A) A plot of T1/T2 vs. concentration for sunset yellow. T1 (measured using inversion recovery experiment) increased with increasing concentration while T2 decreased (measured using the PROJECT pulse sequence). (B)The structure of sunset yellow. (C) Change in 1H NMR spectra of sunset yellow as a function of concentration from 0.01% wt to 33% wt. As the sunset yellow concentration increased, the resonances shifted up-field due increasing size of the average aggregate.

Text Box: Figure 2. (A) A plot of T1/T2 vs. concentration for sunset yellow. T1 (measured using inversion recovery experiment) increased with increasing concentration while T2 decreased (measured using the PROJECT pulse sequence). (B)The structure of sunset yellow. (C) Change in 1H NMR spectra of sunset yellow as a function of concentration from 0.01% wt to 33% wt. As the sunset yellow concentration increased, the resonances shifted up-field due increasing size of the average aggregate.

.

Future Directions: We will compare the data we have collected in sunset yellow to a few simple models of aggregation in dyes (such as the isodesmic model). We will then apply the sequences developed in AIM IIB to further constrain or eliminate these models for aggregation in sunset yellow. Finally, we will apply the techniques we have developed for dyes to begin studying the aggregation in polyaromatic hydrocarbons.