59th Annual Report on Research 2014

The main objective of this research project is to apply advanced molecular simulation methods to the exploration of the methane hydrate formation mechanism. In the first year of the project, the main research outcome was the successful development of an effective order parameter , which is required in the application of advanced sampling methods. The order parameter is capable of distinguishing effectively the hydrate phase from ice and liquid water, while does not favor the reorganization of a particular hydrate phase. The integration of the order parameter into the forward flux sampling (FFS) method allows computing hydrate nucleation rates.

In the second year, we further our research following this achievement. Specifically, employing our developed computational method, we carried out extensive computational studies, which yield several important findings regarding hydrate nucleation. First of all, we have demonstrated that hydrate nucleation rates can be obtained explicitly from molecular simulations under a condition where the spontaneous hydrate nucleation becomes too slow to occur in the brute-force simulation. To best of our knowledge, this is the very first time where hydrate nucleation rate is computed without subject to the assumption of nucleation theory. Therefore the obtained nucleation rates can be used to directly verify the validity and applicability of nucleation theories in methane hydrate formation. Second, we obtained the ensemble of hydrate nucleation trajectories, which are statistically significant to verify the proposed nucleation mechanisms. Particularly, Analyzing the obtained large ensemble of hydrate nucleation trajectories, we show hydrate formation at 220 K and 500 bar is initiated by the nucleation events occurring in the vicinity of water−methane interface, and facilitated by a gradual transition from amorphous to crystalline structure. The latter provides the very first direct support to the proposed two-step nucleation mechanism of methane hydrate.

In the process of reporting our findings, we have spent significant efforts (over six months) checking the convergence of our computed hydrate nucleation rate. This not only ensures our reported nucleation rate is reliable, but more importantly, allows identifying an important issue that might have been largely overlooked when the forward flux sampling method is applied. The issue concerns the importance of the initial sampling of the FFS method. In FFS, the rate is expressed , where is the initial flux rate reaching the first interface from basin A and is the probability for a trajectory that starts from and eventually reaches B. While the computation of the initial flux rate was considered straightforward, the sufficient initial sampling obtained in this step was found crucial for the convergence of the final rate, because the computation of the growth probability is carried out using the configurations collected at the interface when the initial flux rate is computed. It was demonstrated in our work that orders of magnitude difference in the final rate could be resulted when the insufficient simulation time (<300 ns) is used to perform initial sampling of hydrate nucleation at 220 K and 500 bar.

With these successes, we have also started investigating the variation of hydrate nucleation behaviors with the change in the thermodynamic conditions and the presence of foreign substances. We have found that the change in the temperature and pressure can affect the preferred hydrate nucleation pathway, hence the most relevant nucleation mechanisms. This is because hydrate nucleation is a complex process that may involve several steps. The free energy landscape may vary with the change in the thermodynamic conditions. Therefore care needs to be taken when generalizing a nucleation mechanism obtained in one condition to a different one. To better understand the nucleation pathway and mechanism, it is also important to quantitatively describe the free energy landscape. To achieve this goal, we have implemented the algorithm to compute free energy profile, based on both forward flux sampling and backward flux sampling.

As the most relevant hydrate formation proceeds via the assistance of the foreign substance, through the heterogeneous nucleation process, we are also currently working to tackle this challenging problem. To start with, we investigated heterogeneous ice nucleation on a graphene surface. Interestingly, the calculated heterogeneous ice nucleation rates on graphene surface are found to fit well according to Turnbull's theory, indicating that the theory of heterogeneous nucleation may also apply quantitatively to the case of ice nucleation. The next question we would like to answer is what characteristics of a foreign surface would be the key to the control of hydrate nucleation.

 

The ACS PRF DNI award provided the PI and the group the precious support for pursuing the research goal of understanding hydrate nucleation. The project has been supporting one postdoctoral scientist and involving one graduate student and one undergraduate student. Over the two-year project period, the award has also allowed developing new research directions. Understanding the general nucleation behaviors in complex system has now become one of the main research interests of the PI.