c Instrucciones de Diseño para la Ordenanza 3.

The application of this model to alcohol reorientation in the silica pore will be particularly important when assigning timescales observed in the TDF spectrum to either solvent or solute contributions.

5.5

Time-Dependent Fluorescence Summary

The results presented in this Chapter indicate that in NEMD simulations, both the Stockmayer solute and c153 dye model show TDF spectra that decay on three timescales when confined in hydrophilic silica pores. The Stockmayer solute shows a small net displacement toward the pore interface after excitation, while displace- ments of different sites on the c153 model suggest rotational motion. Additionally, as the Stockmayer solute moves toward the silica interface, its fluorescence energy redshifts. An analysis of fluorescence energy as a function of position relative to the silica interface for selected sites on the c153 model suggest similar redshifting occurs for atomistic dye models. Thus, requirements 1) and 2) in the solute diffu-

sion hypothesis are satisfied in NEMD simulations of simple and atomistic dyes in confinement, a result which is consistent with the equilibrium MD results in Chapter 4. To what extent these appear in the TDF spectrum will require enhanced sampling.

Because chemistry happens at interfaces, which molecular-level effects lead to the satisfaction of each requirement in the solute diffusion hypothesis is of chemical interest beyond explaining emergent timescales in TDF spectra. A series of dye models comprising the Stockmayer solute, modified benzene, p-nitroanisole, and coumarin 153 can be used to determine how solute size, morphology, charge distri- bution, and functional group chemistry affect solute diffusion, reorientation and spectral properties, particularly at the interface. In a similar spirit, a series of con- fining framework is offered to systematically investigate effects from confinement size, charge distribution, surface heterogeneity and interfacial chemistry.

Chapter 6

Conclusions and Future Outlook

The present work is aimed at investigating trends in reorientation dynamics in bulk linear chain alcohols as well as investigating the information content in time- dependent fluorescence experiments in confined systems, which show emergent dynamics beyond those observed in the bulk. While the studies described here only begin to touch upon dynamics in liquid alcohols, many important conclusions have been drawn. Moreover, it is hoped that these results—what worked and what did not—can find utility in the future.

6.1

Conclusions from This Work

Studies of reorientation of linear alcohols have indicated that hydroxyl reorienta- tion slows with increased alkyl chain length[25–29, 129]. The chemical similarity between water and alcohols (i.e., they all contain hydroxyl groups) suggests that alcohol reorientation may be described similarly to water reorientation. Thus, to

investigate the origin of this reorientation trend, molecular dynamics (MD) simu- lations were performed for models of water, methanol, and ethanol. The lack of long-lived non-hydrogen bonded hydroxyl groups eliminates alcohol reorientation viathe Debye mechanism [15]. Additionally, the Ivanov jump model[16] requires that intact hydrogen bonds are orientationally static between hydrogen bond switch- ing events, which is not observed. Instead, the extended jump model proposed by Laage and Hynes[17, 18], and subsequently developed to describe water dynamics near hydrophobic solutes[20], describes well the trend that reorientation slows with increasing alkyl chain length.

The model consists of two contributions to reorientation—a fast, large am- plitude reorientation during hydrogen bond switching (“jump” contribution) and the tumbling motion of intact hydrogen bonds (“frame” contribution). The jump time contribution follows the same trend in reorientation. Notably, the reorienta- tion slowing is not an effect of hydrogen bond strength. Further, hydrogen bond exchange geometries are very similar across the series, as shown by bimodal distributions in the jump angle for each molecule. The trend in the jump time con- tribution then originates with the characteristic hydrogen bond jump time for each, an effect that is well-explained using excluded volume arguments. In essence, the increased alkyl bulk results in fewer incoming hydrogen bond exchange partners, which increases the time for a hydrogen bond switching to occur. The timescale for reorientation of intact hydrogen bonds (“frame” contribution) also increases, but it increases less than the increase in the jump time. Importantly, for alcohols, the relative jump and frame contributions therefore indicate that hydrogen bond

tumbling is the dominant reorientation pathway, whereas hydrogen bond switching is dominant in water reorientation.

For higher alcohols through n-hexanol, reorientation timescales continue to increase with increasing alkyl chain length, and the overall excluded volume fraction continues to increase. However, in higher alcohols, additional timescales are observed with the interesting trend of decreasing amplitude.

Several attempts have been made at explaining these, a number of which proved unsuccessful. As observed for supercooled water, the extended jump model can result in the longtime reorientation becoming split between two timescales–kshort

and klong—as a consequence of the distribution of jump rate constants, which in

alcohols can be related to the distribution of excluded volume fractions. Correlation functions in the excluded volume fraction have been calculated in a preliminary fashion, but their calculation is expensive. Accordingly, alternative hypotheses were investigated. Trends in nonspecific, local coordination shells and exchange rates between them failed to describe the dynamics. Additionally, a more global picture that higher order structures show different dynamics also fails to describe reorientation in higher alcohols.

A reasonable explanation for these effects was found, however, through ad- ditional calculations. Local hydrogen bond relationships successfully explain the bimodal distribution of hydrogen bond jump angles. Specifically, hydrogen bond exchange occurs at low jump angle when the current and future acceptor are hydrogen bonded to one another, while the exchange generally occurs at higher jump angle when these molecules have no hydrogen bonding relationship.

Two-dimensional free energy plots (in distance and angle coordinates) show the growing-in of new free energy minima (and associated barriers) with increased alkyl chain length, which successfully explains the emergent timescales. The quantitative agreement between free energy barriers and reorientation timescales, however, may require additional considerations such as solvent viscosity.

Additional timescales also emerge in the time-dependent fluorescence (TDF) spectra of nanoconfined systems, but their origin remains similarly unclear. One hypothesis—the solute diffusion hypothesis—is that because the confining frame- work introduces an anisotropy in the system, solute properties, including fluores- cence energy, change across the confining framework, and that emergent timescales actually report on motion of the reporter dye itself. For this to be observed, the solute position within the confining framework must depend on the solute charge distribution, the fluorescence energy must depend on the solute position, and the net effect—the motion across the confining framework with a change in spectrum— must occur on experimentally accessible timescales and with measurable amplitude. The first two qualifications were addressed using equilibrium MD simulations of a model Stockmayer solute dissolved in ethanol and confined within a silica pore. The results indicate that for 5, 10, and 15 D dipole moments on the solute, the solute position within the hydrophilic pore is strongly biased, while it is more modestly biased in the hydrophobic pore. Additionally, spectral properties, includ- ing fluorescence energy, absorption energy, and spectral line width, all depend upon the solute position within the hydrophilic pore, but not the hydrophobic pore. Free energy decompositions indicate that interactions with the silica interface are

primarily responsible, while contributions from solvent organization play a smaller role. Investigation of several locations within the hydrophilic pore suggest that specific chemistry, i.e., hydrogen bonding with silanol groups at the pore interface, is primarily responsible these effects.

To determine the extent to which solute motion manifests in the TDF signal, two different dye molecules—the simple Stockmayer solute and the more chemically relevant and well-studied coumarin 153 (c153) dye—were dissolved in ethanol and confined in a hydrophilic silica pore. The relevant post-excitation solute properties— solute displacement within the pore and solute fluorescence energy—were followed for 200 ps in non-equilibrium molecular dynamics (NEMD) trajectories. The re- sults for the solute displacement indicate a net movement toward the pore interface for the Stockmayer solute, while a net rotation is suggested for the c153 solute. Additionally, the Stockmayer solute shows a redshift in fluorescence spectrum near the pore interface, while the results for the c153 solute suggest that a similar change can occur for certain sites on the molecule. The TDF spectrum for c153 decays much faster than that for the Stockmayer solute, an effect which may be attributable to solute morphology. Additional timescales are not resolvable, and confirmation that changes to the TDF signal can arise from solute motions within the nanoconfining framework will require further investigation, preferably with the use of enhanced sampling techniques.

From the difference in TDF spectra and static spectra as a function of their position on the atomistic dye molecule, it is clear that both morphological and chemical effects are important in determining the solute contribution to the TDF

spectrum. Obviously, the two solutes provided here do not constitute a solute series large enough to determine how shape and chemistry alter spectral properties. To this end, two additional solute models—modified benzene (mBz) and para- nitroanisole (pNA)—have been preliminarily constructed. The adjustability of the dipole moment of mBz means that comparisons of results between mBz and the Stockmayer solute can be made, with differences attributable to the shape and size of the molecule, i.e., the distribution of surrounding solvent. Similarly the dipole moment of mBz can be tuned to that of pNA, so that differences in the results for each solute can be (at least in part) attributed to specific functional group chemistry. This growing-in of effects may lead to better interpretation of c153 spectral properties and assist in identification of which dye properties are important if a dye molecule is to contribute to the TDF spectrum. This is of particular importance, as no general principles have yet been identified to predict which confined systems show emergent timescales and which do not.