Self-preservation and structural transition of gas hydrates during dissociation below the ice point: an in situ study using Raman spectroscopy

The hydrate structure type and dissociation behavior for pure methane and methane-ethane hydrates at temperatures below the ice point and atmospheric pressure were investigated using in situ Raman spectroscopic analysis. The self-preservation effect of sI methane hydrate is significant at lower temperatures (268.15 to 270.15 K), as determined by the stable C-H region Raman peaks and AL/AS value (Ratio of total peak area corresponding to occupancies of guest molecules in large cavities to small cavities) being around 3.0. However, it was reduced at higher temperatures (271.15 K and 272.15 K), as shown from the dramatic change in Raman spectra and fluctuations in AL/AS values. The self-preservation effect for methane-ethane double hydrate is observed at temperatures lower than 271.15 K. The structure transition from sI to sII occurred during the methane-ethane hydrate decomposition process, which was clearly identified by the shift in peak positions and the change in relative peak intensities at temperatures from 269.15 K to 271.15 K. Further investigation shows that the selectivity for self-preservation of methane over ethane leads to the structure transition; this kind of selectivity increases with decreasing temperature. This work provides new insight into the kinetic behavior of hydrate dissociation below the ice point.

Gas hydrates are nonstoichiometric crystalline solids containing guest molecules within the cages formed by host hydrogen-bonded water molecules at low temperature and high pressure conditions 1 . Three common clathrate hydrate structures have been identified, cubic structure I (sI), cubic structure II (sII) and hexagonal structure H (sH). These structures are comprised of polyhedral water cages, which can trap different guest (gas) molecules. Specifically in each repeating unit crystal, sI contains two pentagonal dodecahedral cavities (12 pentagonal faces, 5 12 ) + six tetrakaidecahedral cavities (12 pentagonal and 2 hexagonal faces, 5 12 6 2 ); sII contains sixteen 5 12 cavities + eight hexakaidecahedral cavities (12 pentagonal and 4 hexagonal faces, 5 12 6 4 ); sH contains three 5 12 cavities + two dodecahedral cavities (three square, six pentagonal, and three hexagonal faces, 4 3 5 6 6 3 ) + one icosahedral cavity (12 pentagonal and 8 hexagonal faces, 5 12 6 8 ) 1,2 . Gas hydrates have drawn attention in the natural gas and oil industries since the 1930s as a consequence of the finding that hydrate formation may lead to plugging of gas pipelines [3][4][5][6][7] . Recently, hydrates have been of increasing interest, mainly because of the substantial amounts of natural gas found in hydrate deposits in oceanic and arctic sediments, and are considered as a potential new energy resource [8][9][10][11][12][13] . Additionally, several researchers have investigated the development of new technologies based on gas hydrates, e.g., separation of gas mixtures, and storage of natural gas, H 2 , or CO 2 [14][15][16][17][18][19][20][21] . The mechanism(s) of hydrate formation and dissociation are of fundamental importance to the different applications of gas hydrates 22 . Fundamental mechanistic insight requires knowledge of the structure and structural changes/

Results
Five groups of methane hydrate samples and five groups of methane-ethane double hydrate samples were formed in a high-pressure optical cell (HPOC) comprising a sapphire window, using pure methane with an aqueous solution of 1000 mg/L of sodium dodecyl sulfate (SDS) at 276.15 K and 5.0 MPa, or a gas mixture containing 68 mol.% methane (C1) and 32 mol.% ethane (C2) with deionized water at 276.15 K and 3.0 MPa. A more detailed description of the experimental apparatus is given in the Methods section. Prior to the dissociation of each hydrate, the system was cooled to a designated temperature and then held for 2 hours allowing the hydrate to equilibrate with the gas phase at this temperature. After this, the hydrate was slowly depressurized to the equilibrium pressure corresponding to the set temperature, and then quickly depressurized to atmospheric pressure in the HPOC cell. From that moment, time-dependent measurements of hydrate dissociation using in situ Raman spectroscopy were started. The hydrate structure and composition were determined from the C-H region Raman band frequencies, corresponding to the vibrational stretching modes of methane and ethane molecules in the large and small cavities of sI and sII hydrate lattices (summarized in Table 1). These frequencies were determined based on the experimental results of Ohno et al. 69 and Subramanian et al. 38,44 , and our Raman measurements on the pure methane hydrates and methane-ethane double hydrates. Additionally, it is known that the gas phase methane molecule has a C-H stretching mode at around 2915 cm −1 , and gas phase ethane has a C-H stretching mode at around 2899 cm −1 and 2953 cm −1 38,44 .
Pure methane hydrate. Figure 1 shows the representative Raman spectra for the C-H stretching mode of methane hydrate during the dissociation process at temperatures of 272.15 K, 271.15 K, 270.15 K, 269.15 K and 268.15 K and atmospheric pressure. The Raman spectra collected at t = 0 min denote the re-equilibrium state of methane hydrate at the related temperature, before the rapid depressurzation of the system to atmospheric pressure. Under those conditions, all Raman spectra measured at five temperatures show two peaks centered

Guest
Vibrational modes Peak position (cm −1 ) sI (5 12 ) sI (5 12 6 2 ) sII (5 12 ) sII (5 12 (5 12 ) cavities of sII hydrate, respectively. The coexistence of the new and original peaks indicates the coexistence of the sI and sII hydrate structures. Gradually, the peaks corresponding to the sI hydrate disappeared, and the structure entirely transformed into sII hydrate. The structure transition is displayed by not only the shift in the C-H peak positions, but also by the change in relative intensities of the C-H peaks. The number of large cavities (5 12 6 2 ) for sI hydrate is three times that of small ones (5 12 ), i.e. the PAR values are about 3:1 for methane guest molecules. This characteristic is observed for Raman spectra of hydrates before decomposition or at the initial stage of decomposition. After the structure transition, the PAR value decreases and tends to those of sII hydrate, which is the ratio of the number of large cavities (5 12 6 4 ) to that of small ones (5 12 ). By comparing Figs 3 to 5, one can see that the elapsed decomposition time increases with decreasing temperature, when hydrate structure transitions occur. The structure transition from sI to sII occurred at 145 min, 93 min, and 67 min at decomposition temperatures of 271.15 K, 270.15 K, and 269.15 K, respectively. These observations may be due to hydrate dissociation being retarded with increasing temperature during this temperature region.

Discussion
In situ Raman spectroscopy can be used to evaluate on the molecular-level the self-preservation effect of methane hydrate and structural transition of methane-ethane hydrate during dissociation below the ice point at atmospheric pressure. Two types of the PAR values were determined, corresponding to the molar ratio of guest molecules trapped in large cavities to those in small cavities (A L /A S ), and the molar ratio of methane to ethane molecules in binary methane-ethane hydrate ( ). It should be noted that the peak centered at around 2916 cm −1 corresponds to the C-H stretching frequency of gas phase methane, and overlaps the peak centered at around 2914 cm −1 related to methane trapped in small (5 12 ) cavities of sI hydrate, thus it was unavoidably included in the measurement of A S and A CH 4 .
Pure methane hydrate. Figure 7 shows the variation of the A L /A S values for methane hydrate with time at different decomposition temperatures. Before dissociation, the A L /A S value is about 3 at each temperature, which is consistent with the 3:1 large to small cavity ratio in the sI hydrate unit cell. Since all methane hydrates were formed at higher pressure (5.0 MPa), the occupancy percentage of methane molecules in both large and small cavities was assumed to be close to 100%, which is close to that predicted by CSMGem 1 . Hence, the ratio of guest molecules trapped in large and small cavities should be close to 3. As shown in Fig. 7   be lower than 3. The decrease of the A L /A S value at the dissociation stage is caused by the sudden increase in the relative peak intensity at 2914 cm −1 (Fig. 1a at 93 min, and Fig. 1b at 70 min). This phenomenon could be related to the weak self-preservation effect of methane hydrate at these higher temperatures. In the beginning, the dissociation rate is high and a large amount of methane gas is released. As stated in the Results section, the C-H region band frequency for free methane gas is 2916 cm −1 , which is very close to that for a methane molecule encaged in a small cavity, 2914 cm −1 . That would lead to the overlap of two peaks related to C-H stretching frequencies in gas phase methane and methane encaged in small cavities of sI hydrate. This would therefore result in a sharp increase in peak intensity and peak area at 2914 cm −1 and corresponding decrease in A L /A S value. On the other hand, the rapid release of methane gas from the hydrate would result in a distinct decrease in temperature (since hydrate dissociation is endothermic); thus the release of methane gas stops within a short time, and the A L /A S value tends to increase. With the elapse of time, the temperature increases to the set value again, so the release of methane from hydrate restarts and the A L /A S value decreases again. When the dissociation process tends to the end, the A L /A S value also becomes stable.
Conversely, the A L /A S value remains above 3 during the whole dissociation process at lower temperatures (270.15 K, 269.15 K and 268.15 K). This indicates the self-preservation effect is significant and the hydrate dissociation rate is very low within the examined time under these conditions, and results in the very slow release of methane from the hydrate, with little influence on the Raman spectra.  Fig. 8. A L refers to the total area of peaks corresponding to the occupancies of methane and ethane molecules in large cavities (5 12 6 2 for sI or 5 12 6 4 for sII); A S refers to the total area of peaks corresponding to the occupancies of methane and ethane molecules in small cavities (5 12 ), where the latter is around zero. Just like the case for methane hydrate, the A L /A S values for methane-ethane hydrate are around 3 before the dissociation at all five temperatures, which is in agreement with the relative intensities data of the various CH 4 and C 2 H 6 sites measured using 13 C magic-angle spinning (MAS) NMR 45 . Similar to that for methane hydrate, the A L /A S values of methane-ethane double hydrate also fluctuate at the early stage during the dissociation. Though the extent of fluctuation is more severe than methane hydrate. The reason may be that the dissociation rate of binary methane-ethane double hydrate is higher than that of pure methane hydrate at the same temperature and pressure, which is in agreement with Stern et al. 64 .

Methane
The most significant charateristic of the methane-ethane double hydrate dissociation behavior, which is different from pure methane hydrate, is the structural transition during hydrate dissociation at 271.15 K, 270.15 K, and 269.15 K. As observed at these three temperatures, the A L /A S value is firstly maintained at around 3, and then decreases continuously and is finally stabilized at less than 1 after a certain period of decomposition time. Considering the 1:2 large to small cavity ratio and the representative peak positions of sII methane-ethane hydrate, this is consistent with the structural transition from sI to sII, which is also shown by Figs 3 to 5. Coversely, it can be seen from Fig. 8 that the A L /A S value at 272.15 K fluctuates drastically with a range from approximately 3 to 5, but shows no obvious downtrend. The reason may be that the dissociation at higher temperature close to the ice point is significant. Therefore, obviates the occurrence of the structure transition for methane-ethane hydrate during the dissociation. In contrast, at lower temperatures such as 268.15 K as shown in Fig. 8, the A L /A S value remains above 3 during the whole dissociation process. This indicates the self-preservation effect is more significant and the hydrate dissociation rate is very low within the examined time under this conditions, which results in the unchanged structure of hydrate and slow release of gas from the hydrate.
To further investigate the mechanism of the structural transition, another set of PAR values, , was calculated and compared, where A CH 4 and A C H 2 6 refer to the total area of peaks corresponding to the occupancies of methane molecules and that corresponding to the occupancies of ethane molecules in hydrate lattice respectively. Figure 9 shows the variation of the A A / CH C H 4 2 6 value with the elapsed decomposition time at each temperature for methane-ethane double hydrate. At higher temperature, e.g., 272.15 K, the A A / CH C H 4 2 6 value remains at around 0.5 as shown in Fig. 9. Hydrate dissociates rapidly and ethane and methane gas molecules are rapidly released due to the weaker self-preservation effect at higher temperature. In contrast, when at a lower temperature, e.g., 268.15 K, the A A / CH C H 4 2 6 value tends to decrease during the dissociation process. The structure of methane-ethane double hydrate is maintained, but the reduced release rate of methane is higher than that of ethane from the hydrate cavities. (Here, the reduced release rate of methane or ethane is defined as the ratio of the release rate (mol/h) to original content of methane or ethane (mol) in the double hydrate). When at 271.15 K, 270.15 K, and 269.15 K, the dissociation process could be divided into three stages. The first stage is the single dissociation of sI hydrate, the second one is the structural transition from sI to sII, and the third one is the single dissociation of sII hydrate. The fluctuation of the A A / CH C H 4 2 6 value in the first stage should be caused by the release of ethane and methane gas molecules. In the second stage, the A A / CH C H 4 2 6 values increase gradually with the elapsed time at these three temperatures, demonstrating that ethane and methane molecules are not released in the same proportion to their original composition ratio in the sI hydrate lattice, i.e., the reduced release rate of ethane is higher than that of methane from the hydrate cavities. Ethane molecules can only occupy the large cavities (5 12 6 2 ) of the sI hydrate lattice, while methane can occupy both small (5 12 ) and large cavities (5 12 6 2 ).
The large cavities become unstable and collapse when the gas molecules escape from them, while some of the small cavities may still remain stable for a certain period of time when methane molecules escape from them 70 . All these factors result in decomposition of double sI hydrate involving the number ratio of large cavities (5 12 6 2 ) to small cavities (5 12 ) no longer retaining the required value of 3:1 for the stabilization of the sI hydrate lattice. In this case, some of the 5 12 6 2 cavities can transform into larger ones (5 12 6 4 ) to construct a sII lattice with excess small cavities (5 12 ). This could be a key feature of the mechanism of structure transition which occurs during the decomposition of double methane-ethane hydrate. Dec et al. 45 used 13 C MAS NMR to study the decomposition of methane + ethane structure I hydrate. Although the sI/sII transition was not detected due to the different temperature conditions, the NMR resonances also clearly demonstrate that C 2 H 6 sI large cages decompose more readily than CH 4 sI small cages.
In the third stage, methane-ethane hydrate converts completely into sII hydrate. As seen from Fig. 9  tion of the structure transition from sI to sII. It seems that self-preservation is selective, i.e., methane shows a stronger self-preservation effect than ethane. To illustrate the mechanism of hydrate dissociation, the schematic diagram for the dissociation process of methane-ethane double hydrate with the elapsed time is shown in Fig. 10.
In summary, the dissociation behavior of pure methane hydrate and methane-ethane double hydrate at temperatures ranging from 272.15 to 268.15 K and atmospheric pressure has been investigated systematically via in situ Raman spectroscopy. The sI methane hydrate shows a significant self-preservation effect during dissociation at temperatures from 268.15 to 270.15 K, which has been verified by the C-H peak position and the variation of the A L /A S value, and leads to the stability of methane hydrate at these temperatures. However, the self-preservation effect is weak at higher temperatures (≥ 271.15 K), as shown by the dramatic change in the spectra and fluctuations of A L /A S values for methane hydrate.
The self-preservation effect also occurs at lower temperature (lower than 271.15 K) for methane-ethane hydrate. The structure transition from sI to sII occurred during the decomposition process of methane-ethane hydrate when at 271. 15  This work presents the important influence of the self-preservation effect on the stability of methane hydrate and structural transition of methane-ethane double hydrate during dissociation at temperatures below the ice point. The new findings obtained in this work should be important to improve our understanding of the kinetic behavior of hydrate dissociation below the ice point. This information can be helpful for future investigations of the mechanism of hydrate formation or dissociation by Raman spectroscopy, and could help to provide important insight for the further development of gas storage by hydrates.

Methods
Procedures. Methane and ethane, both with a purity of 99.9 mol.% (supplied by Beifen Gas Industry Corporation, China), were used in the preparation of the gas mixture and formation of hydrates. Sodium dodecyl sulfate (SDS, analytical reagent) was commercially available (supplied by Beijing Reagents Corporation, China) and used without further purification. Deionized (DI) water was used to prepare hydrates.  5) close V-3 and pressurize the system to 5 MPa using the pressure generator; (6) maintain the system pressure at 5 MPa for 24 h to allow the gas to fully dissolve in the solution and until the hydrate is formed; (7) slowly cool the HPOC to a temperature below the ice point and maintain at this temperature for 2 h; (8) open V-2 to slowly decrease the pressure to equilibrium pressure corresponding to the current temperature, then quickly lower the system pressure to atmospheric pressure; (9) when the system pressure was reduced to atmospheric pressure, the hydrate dissociation process started and was monitored by Raman spectroscopy. (10) Change of the initial decomposition temperature of hydrate in step (7) and repeat (7)- (9). It should be noted that a large number of such experiments were repeated and every experiment was conducted on different interface spots.
In these experiments, the laser passed through the sapphire window and focused on hydrate as shown in Fig. 12. Moreover, the cell was affixed to an X-Y-Z translational stage so that the laser can focus on a specific hydrate position. It should be noted that the experiments were repeated many times and measurement was conducted on different spots of the hydrate.
High pressure optical cell (HPOC). A stainless steel cell similar to Subramanian et al. 38,74 was designed for safe operation at pressures of up to 80 MPa, as illustrated in Fig. 12. The whole HPOC is about 115 mm in diameter and 35 mm in height, and constructed from three major parts, which are fixed together by screws: the front plate, the middle plate, and the back plate. The built-in channel in the HPOC is designed to circulate coolant and control temperature. A temperature sensor inserted through a RTD well is placed only a few millimeters from the sample chamber bottom. One gas inlet in the middle of the sample cell wall allows the sample gas to enter the sample chamber and connects with a pressure sensor. The sample cell is 1.4 mL in volume and has two sapphire windows (6 mm in thickness). Raman. The Raman spectra were obtained on a HORIBA XploRA Raman system equipped with a 1800 grooves/mm grating and 20× microscope objective. A 532 nm wavelength laser, with excitation power of 20 mW, was used to focus on a fixed middle spot of the hydrate surface in all experiments, and the focused diameter was approximately 5 μ m. The spectral resolution was 1 cm −1 . Routine calibration of the monochromator was performed by using single crystal silicon. The data acquisition time for one measurement was about 180 s, and the exposure time was 60 s for each accumulation. The spectra were averaged over 3 accumulations. The collected Raman spectra were analyzed using Labspec 5 spectral analysis software, in which deconvolution of peaks, peak-fitting, and calculation of the peak areas were performed (see Supporting Information). The peak-fitting  parameters were the same for all series of experiments. Initially, the deionized water is loaded inside the HPOC for a week under the experimental pressure to ensure full gas saturation. The formation of hydrate and Raman measurement of hydrate dissociation requires two days. Approximately 140 independent Raman spectra are acquired in one experiment. These spectra were collected on different locations of the hydrate sample.
Quantitative analysis of Raman spectroscopy. The quantitative analysis of Raman spectral data is based on Placzek's polarizability theory [75][76][77] as given by where A i is the integrated peak area (intensity) of a Raman active peak for the guest species i over a finite range, I o is the irradiance on the sample, σ i is the Raman scattering cross section of i at a certain wavelength of the exciting source, N i is the number of guest molecules i in the irradiated volume, η i is the instrumental efficiency of the optical and electronic response 78 . When assuming I o , σ i and η i of different guest species in hydrate lattices being identical 76 , the following equation can be used to evaluate the relative proportion of two guest species a and b trapped in hydrate lattices.
This equation can be used to evaluate the relative proportion of different guest molecules encaged in the hydrate lattice. Similarly, we can also evaluate the relative proportion of occupancies of guest molecules in two different types of cavities by where N L and N S are the numbers of guest molecules trapped in large and small cavities, respectively; A L and A S denote the peak area corresponding to the occupancies of guest molecules in large and small cavities of hydrate lattices, respectively.