The structure of natural gas hydrates and the molar ratio of water to gas are one of the more important parameters to estimate the amount of methane gas trapped in gas hydrate in marine sediment. Furthermore, the measurements of structure and composition of natural gas hydrates may also provide the knowledge of their formation conditions as have been documented by the measurements of synthetic methane hydrates (Uchida et al., unpubl. data).
The molar ratio of water to gas is usually examined by the dissociation test. However, the results obtained are largely scattered in each test and are not considered to be a "real" composition of gas hydrates from the viewpoint of crystallography. Only a few technical notes have reported reliable values of the composition of natural samples, even by these kinds of tests. One of the excellent measurements for the composition of natural gas hydrates was conducted by the use of the calorimeter (Handa, 1988). However, because this technology requires that samples be damaged and destroyed, we employed a nondestructive technique to determine the structure and composition of the Blake Ridge gas hydrate.
Some nucleuses with magnetic moments of two directions, such as 1H, 13C, 29Si, cause two different energy levels for those nucleuses in a strong magnetic field. By radiating radio waves whose energy levels are the same as those nucleuses, transitions between the two energy levels may occur when radio waves are absorbed to and released from those nucleuses. This phenomenon is called the nuclear magnetic resonance (NMR). The NMR spectrometry makes use of this phenomenon so as to observe relaxation of energy as a radio wave from the nucleuses. In such a radio wave its frequency becomes slightly changed in the electric environment around the nucleuses, which is known as the chemical shift. Solid NMR spectrometry has recently been well developed with the addition of cross polarization (CP) for improving sensitivity and magic angle spinning (MAS) for gaining high resolution. NMR spectrometry would provide information such as
A Bruker AMX-400 NMR spectrometer with a probe for solid samples was used for the measurements. A plug of hydrate sample was first crushed into silt-size grains, then placed in the zirconia rotor (I.D.= 8 mm, length = 10 mm) of the spectrometer and cooled down to -70 ºC by liquid nitrogen. The rotor was placed in the NMR probe and spun at 3000 rps with dried nitrogen gas atmosphere to maintain low temperature and dry conditions. A thousand measurements were obtained cumulatively in 1 hr.
The 13C CP/MAS NMR spectra of hydrates are shown in Figure 6. The abscissa designates the chemical shifts, indicating relative differences in the characteristic frequency emitted by 13C in slightly different molecular environments. Two peaks are clearly identified at -3 and -7 ppm. These two spectra designate methane molecules occupied in small and large cages of hydrate, respectively. Other hydrocarbon molecules were not detected.
It is concluded that the crystal type of the measured hydrate obtained from the Blake Ridge is Structure I, because 13C NMR spectra indicate only the methane molecule occupations in small and large cages (Fig. 6A). Only one NMR measurement has ever been achieved on the natural gas hydrate sample collected from the Gulf of Mexico (Fig. 6B), which contains methane and propane and was identified to be Structure II (Davidson, Garg et al., 1986). The 13C NMR spectra of synthetic gas hydrates of Structure II and Structure I enriched with 13C methane and propane are shown in Figure 6C and Figure 6D (Ripmeester and Ratcliffe, 1988). These experimental data help us identify gas hydrate from Leg 164 sediments as being Structure I.
Specimens of a dimension
of about 10 mm3 were prepared from core samples in such a way that
only gas hydrate and ice samples were included in each specimen. The specimen
was sealed by Mylar film to avoid sublimation and placed on a polyethylene tube
attached to a goniometer head as shown in Figure
7. Because the diffractometer was operated at room temperature, a
specially designed apparatus with a cold air flow was used for cooling the
specimen without disturbing the measurements. The temperature of the specimen
was kept at about -50ºC during the measurements. The surface of the specimen
changed into ice, and this prevented the specimen from dissociating even at the
atmospheric pressure. The X-ray intensity measurements were made using
Molybdenum K
radiation.
Diffraction peaks were automatically searched to locate 2
angles between approximately 12ºand 15º, a region in which no peaks were
expected from ice crystals but only from gas hydrate. The diffracted X-ray was
accumulated for 120 s in 2
steps of 0.02º, and the specimen was rotated 100º during each step to obtain
the average diffraction of a mass sample.
Figure
8 shows the intensity of the diffraction peaks of the natural gas
hydrate specimen vs. 2
angle. The crystal structure is cubic with a lattice constant
at about -50ºC, averaged over several difficult to analyze specimens. The space group is uniquely determined as Pm3n from the systematic absence of the indices (h, k, and l).
The intensity of the diffraction peaks shown in Figure 8 was compared with that of the ethylene-oxide hydrate (Structure I) whose structure was carefully determined by McMullan and Jeffrey (1965) using a single crystal. The intensity of diffraction was calculated using their structure factors. The measured relative intensity ratio of the natural gas hydrate corresponds qualitatively to the calculated relative intensity ratio. This agreement indicates that the natural sample obtained included only Structure I hydrate.
Because the diffraction intensities of the specimen were measured by rotation, the resulting peaks should indicate the average value of the measured specimen. If the hydrates were fine powder and distributed homogeneously in the specimen, the resulting diffraction peaks were smooth. However, the diffraction peaks had some notched shapes, as shown in Figure 8. This means that there were some coarse particles of the hydrate in the specimen that were located inhomogeneously. This conclusion is supported by Laue photography of the same specimen, which shows the fact that the specimen includes many small, and some large hydrate crystals with random orientations.
Preparation of the Raman spectroscopic specimen was almost the same as that used for X-ray diffraction measurements. The surface of the specimen was made flat by a knife edge to reduce signal losses by scattering. The specimen was placed in a cryostat that was specially designed for microscopic Raman spectroscopy. The temperature of the specimen was kept at about -50ºC during measurement by controlling the sample-cell temperature with the flow speed of liquid nitrogen. The sample cell was equipped with a microscope Raman spectroscopy system. The magnification of the system was about 103 by using a 20× long working distance objective lens (NIKON CF M Plan SLWD). The diameter of the incident laser beam on the specimen was about 1 µm.
The Raman spectrometer SPEX RAMALOG-100 equipped with a 1-m, double-dispersed monochromator system was used for this study. The spectra were recorded with a photomultiplier tube detector system. The excitation source was an Ar-ion laser, emitting a 514.5 nm line and providing about 120 mW at the specimen (Fig. 9). A computer system (SPEX DM1B) provided control and data acquisition for the spectrometer system. Routine calibration of the monochromator was done by using the Raman scattering line of silicone plate, 520.0 cm-1. The scattered radiation was collected at the 180° geometry with a slit at 300 µm. Spectra were collected with a 5 cm-1 scanning step, and 3-s integration time steps were done with a measuring range from 100 to 4000 cm-1. This wide-range measurement provides a measure of the content of guest molecules within the sensitivity of Raman scattering. Narrow-range measurements were performed from 2800 to 3000 cm-1 to investigate precisely the methane-molecule vibrations in the hydrate. Spectra were collected with a 0.5 cm-1 scanning step and a 5-s integration time/step.
The wide-range measurements of the Raman spectra for natural gas hydrate and for coexisting ice in the same specimen showed that both spectra have several peaks that result from the vibration of H2O molecules: the translational mode is observed at about 300 cm-1 and stretching vibrational modes are observed from 3000 to 3800 cm-1. Comparison of the peak shapes indicates that vibrational conditions of the host molecules of the natural gas hydrate are almost the same as those of ice crystals. This is consistent with the fact that both crystals are constructed by the tetrahedral hydrogen-bonding network of water molecules.
On the other hand, no
significant peaks were observed other than the clearly resolved double peaks
observed at about 2900 cm-1. These double peaks are the 
1
symmetric band of methane. This result indicates that the specimen
contains almost pure methane gas. The dissociated gas measurements of the
natural gas hydrate by gas chromatography showed that more than 99% of the
content is methane (Table 2). Figure
10 shows an example of the results obtained by the narrow-range
measurements. The spectrum of the natural gas hydrate is shown by a solid line
and, for a comparison, that of the artificial methane hydrate is shown by a
dotted line. The peak positions of the natural gas hydrate are 2904.1 cm-1
and 2915.1 cm-1 for the larger peak and the smaller one,
respectively. Although the 1
stretching mode of methane vapor is the single peak at about 2917 cm-1,
this mode is known to be split into two peaks in the large and small cages of
Structure I hydrate. This result quantitatively corresponds to that obtained on
the artificial methane hydrates. Therefore, it is concluded that the natural gas
hydrate samples are almost pure methane hydrates. This is consistent with the
X-ray diffraction measurements.
To estimate the hydration number of the natural gas hydrate from Raman spectra, we applied a calculation method with a thermodynamic model, which was used for the evaluation of the hydration number of methane and propane hydrates from 13C NMR measurements (Ripmeester and Ratcliffe, 1988). In the absence of guest-guest interaction and host-lattice distortions, the chemical potential of the water molecules in a Structure I hydrate, µw(h), is given by
L
) - ln (1 - S)]
/ 23, (2)
where µw(h*)
is the chemical potential of water molecules in the lattice with empty cages; L
and S are the
large- and small-cage occupancy by a guest molecule, respectively; and R is the
gas constant. When the hydrate is in equilibrium with ice, the left side of
Equation 2 becomes
µw*,
(3)
where µw*
is the chemical potential of the empty lattice relative to ice. Davidson, Handa
et al. (1986) reported that µw*
was 1297 ± 110 J/mol for a xenon-hydrate sample prepared under equilibrium
conditions at 0°C. The integrated line intensities of Raman spectra allow the
relative populations of two sites to be determined. After allowing for the fact
that there are three times as many large cages as small ones in type-I hydrate,
the cage occupancy ratio L/S
= 1.25. This, together with the above µw* value, gives
absolute occupancies L
= 0.97 and S
= 0.78. The hydration number n can then be obtained by the form n = 23 / (3L
+ S), giving
a value of 6.2. This value includes some errors resulting from the uncertainty
in µw*
and the evaluation of peak intensities.
Hydration numbers of natural gas hydrates have been reported by Handa (1988), which were measured by using a calorimeter. He determined the hydration number of the sample obtained from the Middle America Trench slope sediment off Guatemala was 5.91, and the number obtained from the Green Canyon area of the northern Gulf of Mexico to be 8.2. The result obtained in the present study compares well with the former. The latter was discussed to be larger than expected, which may have resulted from the effect of the free water produced by the dissociation of gas hydrates (Handa, 1988). Other recent values for n of artificial methane hydrate, determined by using direct as well as indirect methods, range from 5.8 to 6.3 (Ripmeester and Ratcliffe, 1988). The present value is in good agreement with these results. Therefore, it is concluded that the composition of natural gas hydrate determined in the present study is almost similar to that of pure Structure I methane hydrate.
